6th HLF – Hot Topic: Blockchain and distributed ledgers: Presentations "Applications"
2 views
Formal Metadata
Title 
6th HLF – Hot Topic: Blockchain and distributed ledgers: Presentations "Applications"

Title of Series  
Author 

License 
No Open Access License:
German copyright law applies. This film may be used for your own use but it may not be distributed via the internet or passed on to external parties. 
Identifiers 

Publisher 
Heidelberg Laureate Forum Foundation

Release Date 
2018

Language 
English

Content Metadata
Subject Area  
Abstract 
Blockchain and distributed ledgers: Will the reality live up to the hype? Will distributed ledger provide a ‘reset’ button for the internet and other networks? Many of the primary privacy risks prevalent today are due to an increasing centralization of information. A decentralized network is potentially more secure but not without its vulnerabilities. During this session, a panel of experts will illuminate how distributed ledgers work, discuss their potential and explore how the world of finance and other application areas could be reshaped. Cryptocurrencies and their escalating, volatile values have successfully captivated the public. However, the rise to fame has not brought a thorough understanding of the underlying technology along with it and distributed ledgers remain largely misunderstood. A better comprehension of the technology is increasingly vital due to its potential ramifications in finance and regarding privacy. Distributed ledgers could conceivably reshape finance through cryptocurrencies and smart contracts, cure data protection issues with social media and redecentralize the internet. In short, a chance to hit the ‘reset’ button. Simultaneously, the very aspects that make distributed ledgers so promising are the same that make it vulnerable. Though replicability, immutability and being appendonly are enormous strengths, they are equally large burdens when used maliciously. The Hot Topic was coordinated and will be moderated by Eva Wolfangel, European Science Writer of the Year 2018, a science journalist with over 15 years of experience covering a range of scientific issues and technological developments and highlighting their significance for the public. In order to unravel the technology behind distributed ledgers and its potential implications, Wolfangel has enlisted the help of experts with backgrounds ranging from academia to industry. Through discussions and an open debate, the speakers aim to distinguish the implausible from the practical and distill how the distributed ledgers will further influence our lives. Experts: Mihai Alisie is the cofounder of Ethereum blockchain applications and founder of Akasha, a social network based on the EthereumBlockchain and the InterPlanetary File System. Demelza Hays is researching the role of cryptocurrency in asset management in the Business Economics program at the University of Liechtenstein. Dexter Hadley’s expertise is in translating big data into precision medicine and digital health at the University of California. His background is in genomics and computational biology and he has training in clinical pathology. The opinions expressed in this video do not necessarily reflect the views of the Heidelberg Laureate Forum Foundation or any other person or associated institution involved in the making and distribution of the video.

Related Material
00:00
Internet forum
Musical ensemble
Word
Email
Internetworking
Universe (mathematics)
Bit rate
Cartesian coordinate system
01:45
Laptop
Context awareness
Presentation of a group
Algorithm
Real number
Combinational logic
1 (number)
Drop (liquid)
Mereology
Computer font
Total S.A.
Theory
Field (computer science)
2 (number)
Prime ideal
Data management
Derivation (linguistics)
Social class
Mathematics
Video game
Strategy game
Spherical cap
Energy level
Presentation of a group
Area
Algorithm
Spacetime
Regulator gene
Uniqueness quantification
Moment (mathematics)
Bit
Total S.A.
Weight
Binary file
Cryptography
Repeating decimal
Derivation (linguistics)
Internet forum
Data management
Arithmetic mean
Content (media)
Lattice (order)
Data storage device
Lie group
Serviceoriented architecture
Units of measurement
Reverse engineering
08:45
Database transaction
Decision theory
Equaliser (mathematics)
Computer font
Weight
Information privacy
Total S.A.
Perspective (visual)
Variance
Video game
Strategy game
Spherical cap
Computer configuration
Singleprecision floatingpoint format
Universe (mathematics)
Information security
Descriptive statistics
Area
Hidden surface determination
Simulation
Constraint (mathematics)
Channel capacity
Regulator gene
Point (geometry)
Bit
Measurement
Maxima and minima
Internet forum
Arithmetic mean
Oval
Data storage device
Crosscorrelation
Order (biology)
Convex hull
Figurate number
Information security
Mathematical optimization
Point (geometry)
Slide rule
Existence
Token ring
Constraint (mathematics)
Real number
Calculation
Similarity (geometry)
Drop (liquid)
Numeral (linguistics)
Theory
Number
Twitter
Social class
Crosscorrelation
Liquid
OSI model
Subtraction
Weight function
Variance
Volume (thermodynamics)
Weight
Cryptography
Local Group
Number
Lie group
Infinite conjugacy class property
Universe (mathematics)
Strategy game
Mathematical optimization
19:32
Database transaction
Bit rate
Information privacy
Roundness (object)
Insertion loss
Hypermedia
Descriptive statistics
Physical system
Constraint (mathematics)
Regulator gene
Bit
Motion capture
Category of being
Radical (chemistry)
Arithmetic mean
Data management
Process (computing)
Frequency
Order (biology)
Phase transition
Pattern language
Mathematical optimization
Slide rule
Random number
Beat (acoustics)
Constraint (mathematics)
Skewness
Distribution (mathematics)
Branch (computer science)
Measurement
Arithmetic mean
Term (mathematics)
Lipschitz continuity
Normal (geometry)
Weight function
Series (mathematics)
Pseudonymization
Prisoner's dilemma
Theory
Volume (thermodynamics)
Skewness
Line (geometry)
Covariance matrix
Limit (category theory)
Shape (magazine)
Local Group
Error message
Resource allocation
Estimation
Universe (mathematics)
Strategy game
Invariant (mathematics)
Mathematical optimization
Distribution (mathematics)
Outlier
Decision theory
1 (number)
Basis (linear algebra)
Parameter (computer programming)
Mereology
Weight
Likelihood function
Variance
Duality (mathematics)
Spherical cap
Strategy game
Universe (mathematics)
Arrow of time
Information
Curve
Process (computing)
Optimization problem
Normal distribution
Moment (mathematics)
Fitness function
Term (mathematics)
Measurement
Benchmark
Internet forum
Right angle
Information security
Kurtosis
Resultant
Fundamental theorem of algebra
Reverse engineering
Row (database)
Moment (mathematics)
Limit (category theory)
Vector potential
Focus (optics)
2 (number)
Liquid
Software testing
Kurtosis
Subtraction
Distribution (mathematics)
Twin prime
Distribution (mathematics)
Projective plane
Mathematical analysis
Variance
Cryptography
Similarity (geometry)
Contrast (vision)
Skewness
Negative number
Units of measurement
Empennage
Series (mathematics)
32:05
Point (geometry)
Momentum
Linear regression
Bell and Howell
Horizon
Mathematical analysis
Thermodynamic equilibrium
Variance
Planning
Data management
Statistics
Kurtosis
Normal (geometry)
Traffic reporting
Operations research
Pay television
Axiom of choice
Information
Mathematical analysis
Theory
Limit (category theory)
Measurement
Internet forum
Order (biology)
Software framework
Normal (geometry)
Condition number
Bayesian network
Mathematical optimization
33:18
State observer
Artificial neural network
State of matter
Artificial intelligence
Faculty (division)
Wave packet
Internet forum
Medical imaging
Computational intelligence
Medical imaging
Personal digital assistant
Universe (mathematics)
Resultant
Physical system
34:53
Focus (optics)
Group action
Standard deviation
Graph (mathematics)
View (database)
Distribution (mathematics)
Information and communications technology
Iterated function system
Statute
Limit (category theory)
Surgery
Medical imaging
Incidence algebra
Bit rate
Average
Right angle
Office suite
Physical system
37:08
Point (geometry)
Sensitivity analysis
Code
View (database)
Machine vision
Direction (geometry)
Digital signal
Bit rate
Twodimensional space
Surgery
Computational intelligence
Field (computer science)
Number
Image resolution
Medical imaging
Latent heat
Field (mathematics)
Freeware
Sensitivity analysis
Resultant
Reading (process)
Electric current
38:52
Point (geometry)
Pattern recognition
Statistics
Model theory
Equaliser (mathematics)
Scientific modelling
Computergenerated imagery
Artificial neural network
Hierarchy
Parameter (computer programming)
Shape (magazine)
Computational intelligence
Rule of inference
Power (physics)
Facebook
Atomic number
Computer hardware
Graphics processing unit
Series (mathematics)
Texture mapping
Spacetime
Artificial neural network
Machine vision
Cellular automaton
Mathematical analysis
Physicalism
Line (geometry)
Instance (computer science)
Set (mathematics)
Equivalence relation
Word
Process (computing)
Graph coloring
Thermal radiation
Universe (mathematics)
Equation
Film editing
41:42
Algorithm
Scientific modelling
Computergenerated imagery
Zoom lens
Visual system
Shape (magazine)
Regular graph
Computational intelligence
Weight
Wave packet
Medical imaging
CNN
Hypermedia
Average
Computerassisted translation
Social class
Area
Hidden surface determination
Curve
Pattern recognition
Standard deviation
Algorithm
Texture mapping
Machine vision
Objekterkennung
Weight
Set (mathematics)
Prediction
Demoscene
Local Group
Category of being
Word
Graph coloring
Right angle
Object (grammar)
Figurate number
43:46
Slide rule
Algorithm
Information
Electronic program guide
1 (number)
Prediction
Frequency
Medical imaging
Prime ideal
Mathematics
Machine learning
Personal digital assistant
Computer network
Right angle
Subtraction
God
45:01
Point (geometry)
NPhard
Hidden surface determination
Dot product
Spacetime
Scientific modelling
Bit
Bit rate
Information privacy
Rule of inference
Medical imaging
Machine learning
Network topology
Case modding
CNN
Principal component analysis
Internetworking
Right angle
Computerassisted translation
47:08
Server (computing)
Scaling (geometry)
Correspondence (mathematics)
Set (mathematics)
Raw image format
Subset
Number
Medical imaging
Arithmetic mean
Process (computing)
Medical imaging
Bit rate
CNN
Average
Right angle
Right angle
Traffic reporting
Subtraction
48:30
Algorithm
Artificial neural network
Linear regression
Gradient
Scanning tunneling microscope
Sampling (statistics)
Prediction
Wave packet
Computer network
Vertex (graph theory)
Glattheit <Mathematik>
Software testing
Right angle
Negative number
Website
Traffic reporting
Subtraction
Resultant
49:46
Scale (map)
Algorithm
Link (knot theory)
Freeware
Algorithm
Core dump
Bit
Set (mathematics)
Maxima and minima
Medical imaging
Medical imaging
Screensaver
Right angle
Automation
Subtraction
Traffic reporting
51:26
Point (geometry)
Recurrence relation
Statistics
Observational study
Computergenerated imagery
Artificial neural network
Bit rate
Mass
Computational intelligence
Total S.A.
Semantics (computer science)
Number
Local Group
Medical imaging
Unified Communications
Term (mathematics)
Negative number
Pairwise comparison
Position operator
Process (computing)
Line (geometry)
Set (mathematics)
Windows Registry
Sign (mathematics)
Website
Right angle
Negative number
Physical system
Electric current
54:05
Operations research
Characteristic polynomial
Scientific modelling
Computergenerated imagery
Data center
1 (number)
Proper map
Number
Revision control
Medical imaging
Finite element method
Image resolution
Ontology
Software testing
Right angle
Empennage
55:29
Point (geometry)
Threedimensional space
Group action
Digital signal
Twodimensional space
Wave packet
Image resolution
Medical imaging
Wave
Medical imaging
Singleprecision floatingpoint format
Program slicing
Logic
Subtraction
Task (computing)
56:56
Electronic data processing
Sensitivity analysis
Touchscreen
Statistics
Observational study
Touchscreen
Lemma (mathematics)
Iterated function system
Analogy
Digital signal
Set (mathematics)
Number
Frequency
Medical imaging
Negative number
Subtraction
Mutual information
58:11
Hidden surface determination
Touchscreen
State of matter
Physical law
Artificial neural network
Set (mathematics)
Information privacy
Windows Registry
Portable communications device
Computational intelligence
Medical imaging
Permanent
Computer hardware
Physics
Software framework
Right angle
Computerassisted translation
Physical system
Physical system
59:49
Standard deviation
Context awareness
Block (periodic table)
Physical law
Heat transfer
Picture archiving and communication system
Picture archiving and communication system
Communications protocol
Local Group
Medical imaging
Data storage device
Right angle
Traffic reporting
Physical system
1:01:10
Surgery
Scaling (geometry)
Spacetime
Closed set
Channel capacity
Core dump
Interface (computing)
RAID
Cryptography
Local Group
Database normalization
Data storage device
Computer network
Uniform resource name
Computer hardware
Hard disk drive
Computer science
Right angle
Data storage device
Quicksort
Information security
Block (periodic table)
Local ring
Physical system
1:02:35
Point (geometry)
Surgery
Algorithm
Forcing (mathematics)
Sampling (statistics)
Auto mechanic
Physicalism
Core dump
Set (mathematics)
Group theory
Surgery
Malware
Internetworking
Computer network
Right angle
Natural language
Quicksort
Information security
Subtraction
Physical system
1:03:46
Rule of inference
Slide rule
Building
Freeware
Algorithm
Closed set
Group theory
Staff (military)
Number
Data model
Napster
Whiteboard
Hypermedia
Office suite
Information
Quicksort
Information security
Task (computing)
Electric current
Physical system
1:04:57
Algorithm
Cellular automaton
Virtual machine
Open set
2 (number)
Data model
Architecture
Internet forum
Internetworking
Design by contract
Hill differential equation
Encryption
Data type
Task (computing)
Electric current
1:06:33
Elliptic curve
Internet forum
Mathematics
Cryptography
Process (computing)
Presentation of a group
Angle
Computer science
Cartesian coordinate system
Cryptography
Field (computer science)
Sinc function
1:07:49
Area
Readonly memory
Scaling (geometry)
View (database)
Expression
Expert system
Interactive television
Bit
Information privacy
Information privacy
10 (number)
Facebook
Readonly memory
Energy level
Regular expression
Species
1:09:12
Point (geometry)
Context awareness
Service (economics)
Algorithm
Information privacy
Collective intelligence
Facebook
Touch typing
Normal (geometry)
Game theory
Informationsgesellschaft
Physical system
Default (computer science)
Scale (map)
Default (computer science)
Link (knot theory)
Scaling (geometry)
Theory of relativity
Graph (mathematics)
Electric generator
Spacetime
Information
Bit
Multilateration
Cartesian coordinate system
Message passing
System programming
Speech synthesis
Pattern language
Smartphone
Right angle
Identical particles
1:11:18
Empennage
Spacetime
Information
Source code
Moment (mathematics)
Projective plane
Maxima and minima
Expected value
Inclusion map
Frequency
Touch typing
Condition number
Theorem
1:12:43
Laptop
Point (geometry)
Area
Block (periodic table)
Software developer
Projective plane
Content (media)
Laurent series
Propositional formula
Shape (magazine)
Local Group
Chaining
Internetworking
Core dump
Right angle
Digitizing
1:14:10
Point (geometry)
Slide rule
Software developer
Skewness
Planning
Electronic mailing list
Term (mathematics)
Group theory
Mereology
Number
Internet forum
Frequency
Wave
Simulation
Units of measurement
Digitizing
Row (database)
1:15:21
Point (geometry)
Simultaneous localization and mapping
Projective plane
Cartesian coordinate system
Computational intelligence
Connected space
Number
Revision control
Internet forum
Degree (graph theory)
System programming
File system
Vertex (graph theory)
Right angle
Subtraction
Mutual information
Computing platform
Asynchronous Transfer Mode
Alpha (investment)
1:17:19
Server (computing)
Beta function
Information
Curve
Internet forum
Message passing
Hausdorff dimension
Videoconferencing
Iteration
Theorem
Quicksort
Identical particles
Spectrum (functional analysis)
Alpha (investment)
1:18:32
Area
Mobile app
Service (economics)
Graph (mathematics)
Motif (narrative)
View (database)
Projective plane
Bit rate
Prediction
Group action
Cartesian coordinate system
Internet forum
Integral transform
Mixed reality
Utility software
Computer network
Right angle
Software framework
Information
Quicksort
Block (periodic table)
Computing platform
1:20:10
Point (geometry)
Dataflow
Algorithm
Game controller
Context awareness
Programming paradigm
Shift operator
Information
Feedback
Bit rate
Supercomputer
Internet forum
Facebook
Centralizer and normalizer
Internetworking
Order (biology)
Data center
Right angle
Internet der Dinge
Quicksort
Extension (kinesiology)
Information security
Identical particles
1:22:16
Point (geometry)
Building
Presentation of a group
Password
Design by contract
Disk readandwrite head
Mereology
Physicalism
Perspective (visual)
Rule of inference
Field (computer science)
Cryptography
Fluid
Bit rate
Number theory
Smart card
Energy level
Circle
Data structure
Form (programming)
Condition number
Physical system
Area
Rule of inference
Block (periodic table)
Cellular automaton
Projective plane
Physical law
Physicalism
Coordinate system
Set (mathematics)
Publickey cryptography
Local Group
Connected space
Category of being
Data mining
Universe (mathematics)
Order (biology)
Design by contract
Selforganization
Right angle
Quicksort
Identical particles
Communications protocol
Fingerprint
Transzendente Zahl
1:26:27
Satellite
Database transaction
Building
Context awareness
Scientific modelling
Virtual machine
Limit (category theory)
Computational intelligence
Event horizon
Theory
Mathematics
Internetworking
Software framework
Information
Data conversion
Cybernetics
Block (periodic table)
Theory
Interface (computing)
Mass
Group action
Cartesian coordinate system
Sign (mathematics)
Message passing
Mathematics
Lie group
Telecommunication
Units of measurement
1:28:36
Database transaction
Group action
Context awareness
Game controller
Programmable readonly memory
Augmented reality
Information privacy
Web 2.0
Mathematics
Internetworking
Touch typing
Encryption
Data conversion
Computing platform
Form (programming)
Scaling (geometry)
Theory of relativity
Database transaction
Selforganizing map
Basis (linear algebra)
Interactive television
Group action
Limit (category theory)
Cartesian coordinate system
Internet forum
Phase transition
Selforganization
Pattern language
Right angle
Iteration
Local ring
1:31:38
Collaborationism
Context awareness
Musical ensemble
Block (periodic table)
Projective plane
Inclined plane
Cryptography
Cartesian coordinate system
Local Group
Elliptic curve
Internet forum
Mathematics
Mathematics
Touch typing
Equation
Computer science
Subtraction
00:01
[Music]
00:23
so I think we start welcome everybody just a few words to me I'm Eva I'm your host I'm a science journalist and I'm writing for many years about future technologies and how they are changing our lives I'm quite sure that blockchain technology somehow will change our lives as well but there are many open questions until now some compared with the early days of the internet where he had to dial up with this strange noise into the internet and were happy when you received at least one email so this is early times as well at blockchain technology and we are going to discuss some of these these open questions and trying to look into the future what is going to come so here this is the application sessions so we have three talks about very interesting and promising applications public blockchain technology we do it like this that everybody talks about 25 minutes and we have five minutes for short Q&A it's more for understanding not for discussing the whole thing because discussing we are going to do later on when we have the panel discussed discussion with all the whole audience and all the speakers together so I hand over to Adam Elsa highest know he is Russia is researching at the University
01:47
of Liechtenstein about the role of cryptocurrency in asset management isn't going to tell us about that welcome to Elsa wonderful thank you very much for having me so basically I have a background in economics so it's this is gonna be a little bit more of the theory that you heard earlier applied to economics and finance so basically at the University of Liechtenstein I've been teaching a course on Bitcoin and blockchain for the past two years and we discussed many topics in that field specifically related to economics but one problem with theory and academia and teaching economics specifically is that you basically study money your whole career but you never seem to have very much of it so a couple years ago a wealth management firm in Liechtenstein said well we'd like to create a cryptocurrency fund and would you like to join the team and I said sure of course that sounds great I can finally apply all of this theory into practice so I have basically one foot in academia and one foot in the industry and my papers and my research is a combination of those two roles and also I just wanted to thank the Heidelberg laureate meeting for having me here I think it was really great to see what field Diffie here I teach about him and I think without his work I probably wouldn't be giving this presentation now so I definitely want to thank Heidelberg for having me here okay so this
03:21
presentation today is going to basically start with an overview of financial aspects of cryptocurrencies I'm not going to talk about the ethics or the technology I'm just going to jump in with some assumptions on those topics and then I'm going to discuss basically what strategies people have been using and how those strategies have fared for people as far as their returns go and their risk then I'm going to follow up with further research areas for people that are in mathematics or in economics and are interested in researching cryptocurrencies and I always have to say this at every single presentation but this is not investment advice okay so basically that's the disclaimer here so don't leave this presentation and spend all of your life savings on cryptocurrencies I mean if you want to but not it's not my fault okay so basically people have been investing in cryptocurrencies since 2009 invest investors and hedge funds hedge fund managers have also been using algorithms such as mean reversion and other trading strategies to trade cryptocurrencies since 2011 and a lot of people don't realize this but it's not only hedge funds that are investing in cryptocurrencies so today institutional investors and pensions can actually invest in cryptocurrencies as well in Europe because there are regulated funds under the u s it's an AI f md regulation so basically there's fund in germany and there's a few funds in Liechtenstein and a few funds in Switzerland that allow large investors to invest in cryptocurrencies directly okay and these funds legally they require a custodian bank so basically this is a bank take for example any large European bank and this Bank is licensed to actually store cryptocurrencies for investors so this does not exist in the US currently the u.s. does not have any licensed custodian bank or prime broker dealer but Europe actually has several licensed banks that can actually hold cryptocurrencies for investors and this is very important because the custodian bank is financially and legally obligated to store those cryptocurrencies so if the custodian bank loses your cryptocurrencies they're actually financially liable okay and this is really unique about Europe and this is why Europe is attracting so much capital in so many blockchain entrepreneurs currently now these custodian banks of course will only take on this level of risk when they can find an insurance company that will insure their holdings so the second part of custodian banks is insurance companies and currently Switzerland has an insurance company that is insuring custodian bank holdings of cryptocurrencies but this is still a relatively new field and there's a lot of space for competition the third part is auditors so for example Ernst and Young is now beginning to audit cryptocurrency holdings held by regulated banks okay so but auditors need to basically be there to verify that the bank's claim is honest so if they claim they have 1,000 Bitcoin they actually Ernst and Young will come in and check that those thousand Bitcoin exists then they report it to the insurance company and it basically works like traditional finance strange my fonts my font package does not work on this laptop so I have a very cute curly handwriting the font here which you can't really quite read very well but in general this chart basically this is just an overview okay so basically the idea here is that there's over 2000 cryptocurrencies so just to get in a broad understanding of the audience how many people in the audience actually hold cryptocurrencies currently okay okay so that's that's that's quite a few of you okay wonderful so this might be a little bit repetitive for some of you and a little bit new for some of you so basically there's over 2000 cryptocurrencies that already exists the total market value of all of the cryptocurrencies is already over 220 billion it almost eclipsed one trillion in early 2018 and this is just a few of the coins that are above a billion dollars in market cap and I've defined market cap here for anybody that's not really familiar with financial argan okay so basically there's already many coins that are above a billion dollars so this is already very relevant for investors but just to put this into context if you think about cryptocurrencies as an asset class which this weather asset is and represents unique asset class or not is very subjective and it's not exactly a science but in general if you did think about it like this you could see how small it is compared to all the other assets that investors normally invest in so here we have derivatives and these are all proportionate proportionate little bubbles here so this is derivatives making up the largest financial market in the world then we have real estate global real estate 220 trillion we have global bonds global money this is all of the M ones of all the fiat currency this is stocks gold of course is just a little drop in the bucket and then crypto is very small at the moment so it still has quite quite a way to grow if it's going to become relevant for institutional investors
08:46
okay so again my fonts look very funny here but this is the 90day correlation moving correlation between the S&P 500 and the price of Bitcoin so basically there's no trend this is basically just showing that when the stock market goes up it doesn't mean that Bitcoin goes up and when the stock market goes down it doesn't mean that Bitcoin goes down okay so basically there's no clear relationship between stocks and bonds and this is interesting from a financial perspective I mean sorry no clear relationship between stocks and cryptocurrencies and this is interesting for investors because when you see this kind of relationship this means that you could actually potentially diversify away some of the risk in your stock market holdings because now you're going to add an asset to your portfolio that performs in a different way okay and here you can see just some some descriptive statistics it's not correlated with any other asset class either so Nasdaq bonds gold real estate okay so it's basically uncorrelated with all other major asset classes and this is a major reason investors are becoming interested in cryptocurrencies okay this is another broad overview of financial topics with cryptocurrencies this is basically just a simulation if you had added Bitcoin to your portfolio what would have happened to your portfolio so a major measurement of portfolio return is the Sharpe ratio so the Sharpe ratio is kind of a ratio of your risk to your return because just gaining return isn't relevant for all investors a lot of investors actually want to minimize their exposure to volatility on the market okay so basically these numbers that will be available for anyone afterwards that wants to further research this but in general your Sharpe ratio clean proves when you add just a little bit of cryptocurrency and that's simply because the returns have been historically so significant and the the low correlation between cryptocurrencies and other assets okay so this is a nice chart that I made recently this is all of the the bear markets that Bitcoin has had since its existence so basically there's been about eight bear markets the bear market is defined when the price drops and continues to drop from its high so basically it turns around it's no longer in a bull market it's actually on the decline and yeartodate so January of this year we're at minus 51 percent for Bitcoin so maybe everybody's familiar with this who invested who raised their hands they've probably they're probably fully aware that the crypto currencies are have lost half of their value since the beginning of the year since 2013 they're still plus four thousand percent so it's it's it's still it's still relatively performed very well compared to other asset classes so here's some some different durations of bull markets I unfortunately invested in the longest poem of bear market excuse me that existed this one it lasted for two years it lost 84 percent of its value during that bear market and least to say I had about two years of sleepless nights and probably added a decade of age to my life during those two years because I didn't really realize what I was investing in and and how risky it was as an asset class and that's also what motivated me to continue to do this research okay so that's an overview of the financial topics the financial market on cryptocurrencies now I'm going to talk about a paper that I'm writing for the University of Liechtenstein and from my research okay so in this paper I'm basically going to talk about what investment universes are I'm going to define risk with cryptocurrency and then I show how I collect data and then I basically calculate different performance measures on that data and basically the point of this paper is basically just to ask the question if I do invest what is the optimal investment strategy okay so if I if I'm going to put some money in this as a diversifier from my portfolio how should I go about putting all of my money in should I put it all in Bitcoin should I put it in a handful of tokens should I put it in Silvio's Algar and you know how should I go about doing this okay and there's a few things of practical consideration that are important to consider first of all the quiddity constraints okay so many cryptocurrencies have daily trading volumes that are so low which means the amount of people actually investing in that coin is so low that it actually doesn't have the capacity to hold large investments so if you put a large investment into that cryptocurrency you're actually going to move the price with your own transaction and you're gonna actually make the market fluctuate so basically you want to have no coins in your portfolio that have shallow order books or have single point of failure order books so here's what a single point of failure order book is
14:09
this is where your coin is only treated on one exchange that's domiciled in a specific country and then that country decides to ban all cryptocurrencies and then your money's stuck on an exchange and you can't get it out okay so this is one constraint another constraint is insurance constraints so coming from the fun side we in our fund we actually cannot invest in any coins that do not have cold storage options at this this is simply because an insurance company will not insure the banks holdings if they cannot be relatively sure that they actually can securely store them okay so this is another constraint legal constraints okay so basically privacy coins we can take we cannot invest in privacy coins because basically this is a gray area legally so if you invest in Mineiro or Z cash or  you may or may not be supporting illicit activity so this is another constraint and finally the final constraint is rebalancing constraints so transaction cost so a transaction cost is basically every time you convert from Bitcoin to Fiat or Fiat to Bitcoin you have to pay a fee okay and if this fee is relatively high then you don't want to convert back and forth too often right so basically one thing is that you have to consider how often you're going to be rebalancing your portfolio is it going to be every quarter you change what you hold every month you change what you hold every week every day et cetera okay so in this paper I basically like in two things together cryptocurrencies in the stock market and I do this because everybody knows the stock market well and we can easily just say does traditional financial theory applied to cryptocurrencies so we can basically just consider these as two very similar markets and in fact the regulator's consider these things as two very similar markets so the SEC in the u.s. recently announced that all initial coin offerings are security offerings in their opinion okay which basically is if anyone's not familiar with an initial coin offering what that is is that's basically when a group of entrepreneurs comes together and says we have an idea we want to raise capital and what we're going to do is we're going to create a aetherium ERC 20 token or a similar cryptocurrency we're going to issue this to our investors and in exchange our investors are going to give us fiat currency or another cryptocurrency and we're going to give them a token and that's going to represent some kind of security as an investment in the entrepreneurs ideas okay and the sec basically said this this these are securities okay and fin mah in switzerland re also said recently the same thing they said that about 95 percent of all i ciose our security offerings and so the regulators think of it that way but so do entrepreneurs because you can see here that entrepreneurs are actually increasingly using initial coin offerings to raise capital so in quarter 2 of 2018 just previous quarter the ico market accounted for 45% of the traditional IPO market and 31% of venture capital market okay so basically entrepreneurs and regulators are seeing this as just a new technology a new way to raise capital okay and also on the stock market there's various strategies that you can use to invest in stocks stock picking for example you can say I think visa is undervalued or I think it's overvalued therefore I'm going to invest or sell buy or sell visa stock this is called an active trading strategy because you're actively making an investment decision there's also passive investment strategies for example market cap weighted liquidity weighted mean very mean variance optimization and one over in okay so basically I think the one that's the most interesting here is the one over in strategy because this strategy does not rely on any data and any theory and it's basically as old as mankind so basically the idea here is that diversification is good and this is an ageold wisdom so basically one over N is just let's say you have in assets in your investment universe so you have you know you go to the you go to the dog races or the horse races and there's ten horses there and basically you can't figure out which horse is going to win and then what you do is you basically just put an equal amount of bet on all of the horses in the race so you know you're gonna win you might not win very much but you know you're gonna win because you've invested in all the horses at the race okay so this doesn't rely on any data or any theory and just for definitions market cap weighted is basically where you take the market capitalization of the coin so we saw with Bitcoin it was around I think 220 billion no sorry sorry excuse me what was what sorry no what was um Bitcoin out no yeah around okay on this slide it seen 200 above 200 okay so like let's say that you take Bitcoin and what you do is you basically invest
19:34
according to the market capitalization so if Bitcoin has if Bitcoin makes up 80% of the market then 80% of your portfolio is invested in Bitcoin okay and then if etherium makes up 20% of the market you invest 20 percent of your portfolio and aetherium so there's basically this is market cap weighted liquidity weighted is basically daily trading volume so the coins that have the highest daily trading volume are the coins that make up the largest weights in your portfolio etc meanvariance optimization mint variance optimization this is basically an optimization problem and I will go into this too much but there's another paper recently written by Brown and mesto that do this for cryptocurrencies and they also have some really interesting results on meanvariance portfolios for cryptocurrencies okay second part is defining the investment universe so basically it's again it's just like opening up a textbook on the stock market all you have to do is just read the textbook on the stock market and you know exactly where the cryptocurrency market is going so we can break it up into different groups we have blue chip stocks like IBM and Apple and then we have blue chip crypto currencies like Bitcoin and aetherium we have emerging market stocks from different countries in Asia and South America and then we have emerging market cryptocurrencies that are from South America South America and Asia finally we have penny stocks and we have penny coins penny coins is not the term that's normally used for this group the term that's normally used for this group maybe some of you know it starts with s it ends with T it's four letters okay so this this is normally you know that we also it's the same it's the same you have you have broad categories and you have coins that fit into these and right now I'm in the current currently in the process of just looking at the descriptive statistics of these different groups to see for patterns but it's definitely clear that in the literature on financials on traditional finance you can see that penny stocks have a negative return so in general when you're investing the likelihood that you get a positive outcome is is statistically proven to not exist so just it doesn't exist statistically but it's it could potentially be the same with penny coins because there's so many different things going on here for example the car taxi ico is anybody familiar with the car taxi I see oh I hope nobody here involved is involved with that project so recently the Securities Exchange Commission in the u.s. created this blockchain Explorer which allows the regulators to see where each transaction is moving throughout the blockchain because they're not exactly fully anonymous right they're pseudonymous unless you have a privacy coin like when arrow works or something like that but with car taxi that what the regulators could do is they could trace back all of the transactions to the original wallets and what they found is that car taxi claimed something like they had raised seven million dollars but what they actually found out is that they had only actually raised two million dollars and then the creators of the ICO had cleverly taken the two million dollars and then put it through a whole system of new wallets and then reinvested it back in their own wallet to make it look like they had raised much more so that they could get more media hype and so that they could get more investors because then other people would think oh wow that coin must be very popular maybe I should invest in it right so there's definitely a lot of weird market manipulation going on in these penny coins okay then defining risk the next step okay so a lot of the papers on cryptocurrency so far that have come out peerreviewed journals basically use variance as a measure of risk and they also use sharp rage the Sharpe ratio is a performance measure however there's already literature from traditional finance that says that variance may not be the best measure of stock market performance because it does not consider the third and fourth moments of the distribution okay so basically you can see here that these two distributions you could say of returns or of any outcome these two distributions have the same first and second moments of the distribution right the first mean and the mean and the variance are the same but actually stock market returns do not follow a normal distribution and neither do cryptocurrency market returns so most likely there's actually a better measure for risk and a better measure for performance than either the variance or the Sharpe ratio and that's basically what I argue for in my paper okay so just for time purposes and also to not complicate this to to too much I'm happy to ask questions about it afterwards but basically stock in order to use sharpen invariance you would basically need to assume some kind of normal normally distributed returns but definitely there are several papers that already confirmed that Bitcoin  litecoin Bitcoin cash and aetherium and many other cryptocurrencies have a skewed and leptokurtic return distribution so basically this is where you have I think I showed in the next slide this is where you have this high peak and fat tails right kind of like the Nassim Taleb Black Swan distribution so you have basically positive skew with cryptocurrencies and you also have outliers that really change the distribution from from a normal one so this is a normal distribution set around the same mean for Bitcoin and bitcoins this blue line and the red line is a normal distribution okay and yeah I there's also a paper that does this it was in the Journal of risk finance last year and they
25:50
basically cover this for all the cryptocurrencies and I've just done some very brief analysis of different penny coins to see if there's a negative skew similar to the stock market okay so the solution that I proposed is the Omega ratio so it already exists it's already being used in traditional finance and basically what this does is it incorporates all the information available in the distribution so it incorporates the third and fourth moments kurtosis and skewness and what I do is I collect data from coin market cap basically I take all of my cryptocurrencies I take my strategies I start with an initial an investment amount and then I see what happens to that initial investment amount okay so out of all the strategies that I discussed previously the one over in the market cap weighted active stock picking volume weighted strategies meanvariance optimization what strategies do you think performed best on the stock market so if you had to go out on the stock market invest today what strategy would you use which one trendfollowing okay trendfollowing mean reversion okay what over N great 1 over N so historically 1 over N outperforms all these other strategies and there's a few reasons for that basically trendfollowing and fundamental analysis fit into the active trading strategy branch and the passive trading passive trading strategy branch basically includes any strategy where you don't have to take daily decisions basically you just invest on day one you walk away and you come back later on ok and one over N is basically one of those passive trading strategies where you just sprinkle a little bit of around on all the different stocks and then you see which ones grow and which ones don't make it and definitely with the stock market one over and performed the best it's very hard to beat there are some active trading strategies that use mean reversion and they actually do outperform benchmarks by one over N but it's it's again going back to the bell curve of returns it's most likely the case that these investors actually got lucky and had a return that was above the mean just because statistically some investors will have returns that are above the mean right but it's very statistically unlikely that two years in a row an investor can beat their benchmark and by the third year it's very statistically unlikely so you don't have you have they've done many papers on fund managers right so fund managers cannot consistently beat their benchmarks unfortunately which is it's very hard because I know this fact and then at the same time in my job I have to go and justify my fees to someone so it's it's really it's a really difficult dilemma here I struggle with cognitive dissonance every day but most of the time I just tell them the truth and then hope that they'll invest with me but that doesn't usually work I usually go for the you know hotshot investor that'll tell them they can beat the beat the market every year and then my question is is it the same for cryptocurrencies so basically I ran this round the test and you find some so you find some interesting results with cryptocurrencies you know what's weird is that first of all one over five so if you just invested in the top five cryptocurrencies beginning in 2013 and then you rebalance your portfolio every quarter you actually performed better than if you had only invested in Bitcoin okay so in the first phase we see some benefits of diversification because actually this portfolio has a higher terminal value if you had invested a thousand dollars in the top five coins rebalanced every quarter you actually would have earned more including rebalancing transaction fees than if you had just invested in Bitcoin but then we see that as you add ten to your portfolio actually your terminal value starts to drop and by the time you invest it in the top thousand cryptocurrencies your portfolio is really taking a hit okay so this charts a little bit hard to see here but if you have any if you'd like to see the paper afterwards I'm happy to distribute it but basically what's weird here is that it looks like there's some limits to diversification within the cryptocurrency asset class okay so basically that's what I try to show on my paper now my next paper next series of papers I'm going to try to be going to try to understand why this limit to diversification exists and here's some possible reasons ambiguity aversion which is in the literature on financial markets this already exists basically this is the concept that you will only invest in a stock that has a very long history of data and so that there's no parameter uncertainty about your distribution of returns so you're pretty much sure what your covariance matrix is going to look like legal constraints so for example at our fund we can only invest in about ten different coins legally okay so it could be that oh okay so I'm running out of time here okay skew and kurtosis also penny coins possibly they have a negative return statistically okay so I'm just gonna go ahead and finish up here conclusion stocks and cryptocurrencies are two different ways to raise capital for entrepreneurs some cryptocurrencies are extremely volatile and this might suggest that there could be some benefits diversification however it seems like there's limits to cryptocurrency diversification and the Sharpe ratio may not may not be an appropriate performance measure and variance in either because the higher moments of the distribution matter for cryptocurrency returns okay and then finally just for a plug for myself I
32:06
make a report it's sponsored by Fanta Bell Bank in Switzerland and this report is available for free in German and in English every quarter and we basically this is just financial analysis of the cryptocurrency market for investors so I hope to yeah take some questions
32:29
thank you so much Demelza so time is short one question who is the first that's you there's someone with the micro come yeah my question is really simple so I think it's really nice to introduce our moments but if we have a few points and really like I tell I guess it's really noisy to measure the third order or the fourth order momentum and I'm just wondering if we couldn't measure something else like for example just l1 norm or other stuff like this to have I or information you said suggests measuring what instead one norm or just yeah potentially definitely a big major limitation would be the lack of data
33:20
available I mean that's definitely one problem with this whole market as a whole it's the results are definitely certainly noisy and I do weekly observations here so it would be very hard potentially that could be a way to overcome that that problem though certainly so sorry for all the other questions but I promise we have more time later on at the panel discussion thank you thanks a lot thank you see you later at the panel now I would like to welcome Dexter Hadley he's from University of California he is a medical background and he is talking about distributing cancer imaging for artificial intelligence and that's what I think an interesting use case for blockchain
34:06
technology welcome Dexter thank you I just like to thank Eve personally for inviting me the conference and all of you for showing up so yeah I'm a physician I guess by training but I like to say I spent 10 years in medical school figuring it out how to not practice medicine and this is what I've been doing in the meantime so I'm a faculty member at the University of California which is one of the only state health systems left in America like staterun there's six medical schools at UCSF which is the San Francisco which is the flagship there's more like 4 million patients at our University probably more if you consider the whole state and obviously I'm talking about
34:55
cancer imaging this is all of the cancer imaging that has been done at UCSF in the last two decades about 5,000 exams have been run on patients and you can see chest xray CT mgs mammogram that's what I'm gonna focus on the yellow but you can sort of see it's a long distribution and this is the cumulative distribution so if you look at
35:21
mammograms there's been about this is current till 2016 that if you add 2017 and there's about a million mammograms that we have in our system that doctors can look up but can anybody guess mammograms have been since the 80s anybody have any idea what happened before 20052006 why the distribution is flat because we threw him away so film all right so we transition from an array of fill mammography it's a digital mammography if the statute of limitations is seven years for criminal liability so after seven years millions of dollars to store the film they didn't think it was
36:04
I beg to differ but this is you know you can't really see on this graph but um around 1985 there should be right here there should be a shaded region of the graph which shows when the fda regulated mammography in america my office a ten billion dollar industry at least in america and what this graph shows is it really doesn't work right so this is the average size of tumor excised at surgery right so if and the fda implemented mammography standardization in the 80s you'd expect big tumors to go down because you're catching small tumors earlier what you see is small tumors going up meaning you're taking out clinically irrelevant you know you're identifying and surgically extracting from patients clinically relevant cancer because the rate of large tumors is about the same right so this is and there's many different papers that show this this is in the New England very prestigious one and by and large you know manual breast exam in the bathroom works as well as no better than mammograms and this is why
37:10
so firstyear radiology resident has to look at this which is you know to to burst two views so you have four images and based on these four relatively low resolution images you're supposed to find free cancer or early cancer and then this is an example where no radiologist is gonna find cancer on this but this is confirmed pathology confirmed cancer this is an example of radiologically opaque cancer and just remember that buzzwords jargon radiologically opaque and yeah so this
37:44
full field diagnostic mammography is just code for twodimensional nog Rafi like I said about a ten billion dollar business because 30 million 30 to 40 milligrams every year FDA regulates every single one of them you know the sensitivity and specificity is okay it's on the 90s but every percentage point you know one in eight women in their lifetime is going to get breast cancer so every percentage point either direction is hundreds of thousands of women right in America at least probably in in Europe as well but this is just incredible ninetyfive percent of women are going to have a false result over their lifetime one in two women are gonna get called back with added anxiety added surgery added mastectomy for no good reason and this is over their lifetime so one in two women will have a false reading over their lifetime and ten percent will have some kind of surgery because of it so these are these are bad numbers well computers have been around for 20 years and they really haven't worked because the way computer vision has worked before this era of deep learning was
38:53
basically mimicking a human's knowledge so humans can't do a good job of this neither can computers if you just mimic so I'm in a big data lab and AI lab and deep learning many of you are in statistics I hope you've heard about this but this is supposedly revolutionising just about every industry today arguably except medicine so deep learning is different to traditional computer vision in the sense that the computers figure this out given label data there's no rules there's no textures there's no colors there's no shapes or shades or anything or that go into this kind of analysis but you know a lot of computation theoretically this has been around since the 50s or the idea of artificial neural networks functionally they're coming out today because we have GPU compete now very large Harley power parallelized hardware and guys like Jeff Dean I met him yesterday I have figured out tricks to make these very large computational problems tractable all right so it's a day and think about it like this so most of statistics maybe I shouldn't say that in this audience but a lot of statistics is based on the equation of a line right y equal MX plus C or B depending on I was gonna British background which means you know the two parameters in that equation so with two points you can plot a line and then predict every Y given an X so the deep learning model is that predict whether for instance there are more parameters in those models than there are atoms in our universe but I know this audience likes to talk about universe but there's no way you could deterministically predict these parameters very large parameter spaces so two points are not gonna cut it you need very big data which is the age we're living in today especially in medicine and don't take my word for it
40:39
but you know I learned in medical school you know anybody has any idea this is Marie Curie the mother of Radiology Nobel Prize winner in chemistry and physics I believe died of radiation poisoning but the way our brain recognizes this is Marie Curie is you know light hits the retina the cells in the retina goes through a series of hierarchical distinct networks that interpret colors and shapes and this and that and the other before the higher cortex recognizes this is a picture of an old lady or Marie Curie well Google and Microsoft and and and Facebook they've built similar computational equivalents whereas these convolutional neural networks are deep article networks every single layer is multiple neurons that are interconnected and each of those are parameters that need to be set by very large data sets for them to work accurately but given mark davis has given big compete these things work better than humans and don't take my
41:42
word for it so this is the imagenet competition there's a very large standardized data set that came out of Princeton and after Stanford of a on average a thousand objects into a thousand categories so on average a thousand categories per objects and these are just regular pictures of dogs and cats I would imagine they scraped of social media and just took the tags to build this data set but and the competition is you know every dot and this is a performance of an entrance entrant into this competition you know given up all that set of say 1200 images predict the class of those thousand classes the training set is a thousand images of every class and you can see before deep learning came on the scene where you're telling the computer look for this color or this shape or this texture to figure out a dog from a cat best accuracy we could achieve was 70 75 % Toronto University of Toronto Jeff Hinton's group Alex net is the first deep learning approach to this competition and significantly outperformed everybody else they stopped the competition this year because today humans have been beat for two or three years now so they're at 98 99 % which means when you submit a lowresolution you know dark image of a cat to Google Google has a better chance of guessing that's a cat than a human right so today in feels like regular computer vision regular object recognition and entity recognition computers outperform humans
43:10
Google figure this out and paid 10 ophthalmologists to look at pictures like this which is a funder scoffing exam of your eye and make a prediction about some diabetic retinopathy they did at one hundred and twenty five thousand times right and trained the model and said you know for the statistician this is an area under the receiver operating curve every dot is a human compared to the computer only this guy here's a zoom in at the same curve only this guy after from the algorithm right so give me enough time enough knowledge we could learn his knowledge is biases and not before him the next big thing in deep
43:48
learning in medicine was the Stanford paper so it's very clever how they did this so they claim that this network can predict skin cancer better than a dermatologist something like one in 30 biopsies for skin cancer or not cancer this algorithm takes it down to one in ten the funny story behind this is nobody has hundreds of thousands of Bobbsey confirmed melanoma or skin cancer period right taken with an iPhone or whatever so what they did is it took you know maybe 2030 Bobbsey confirmed images of moles and then literally photoshopped them on different peoples ethnicities and skin and artificially created this dataset so they artificially showed in this the data said deep learning works this slide changes depending on you know
44:34
what's the biggest thing going on right now but I don't know if you noticed Apple's keynote last week or two weeks ago whenever it was but people wear EKGs on their hands now with the new Apple watch and deep learning is the one that tells you interprets the tracing right so if you're in a favored atrial fibrillation or god forbid assist aliy deep learning algorithms are the ones that sort of give you information on this raw data so you've probably figured
45:02
out by now but but my goal is to apply the same kind of machine learning that can outperform humans on dogs and cats to you know ductal carcinoma in situ lobular carcinoma in situ so on and so forth where am I going to get a million images of that label is the question well various people have tried one this
45:25
is data from the tree mammography challenge so they basically uh again this is medical data it's not dogs and cats you could put on the Internet right so this is private data and they're all kind of rules that sort of surround how this data can be used and this competition was to use six hundred thousand dollar images of cancer that have been labeled by pathology because of the privacy restrictions on this very valuable data they released 50 of these points which is a twodimensional I think is tis New Year no no no some kind of principal component space blue is non cancer red is cancer so we ran this 50 points to the original model this model right the first Alex next model the one
46:08
that won this competition and already you can see the red segregates from the
46:12
blue we didn't do not anything this is mod this is data trained on dogs and cats right this is these are models trained on dogs and cats and we could already see discrimination and when you look a little bit closer as to how these things are clustering you see what you see it's kind of hard from this angle but around every mole when you go for a mammogram they put a marker it's hard to see but there's mark mole markers here so all of these dots have mole markers that's why they're clustered together there's plate artifacts so the lead plate they put on you is another reason that these things are misclassified and implants so that's what implants look like on a mammogram and all of these points here are related because they're implants so with a little bit of with more data we think we can reduce this kind of artifact right again this model is trained on dogs and cats of that's more data with more mole markers with more plates with more of everything we can reduce that to really emphasize the cancer signal so like I
47:08
said this competition there was six
47:10
hundred and something thousand images I already have more of this than done that
47:14
competition and here's the data we have in UCSF over the last twenty years so we have nine hundred thousand images sitting on a server somewhere a mammogram is four images per exam all right so that's the scale so this is the raw images they're called DICOM that's just what they're called and for every DICOM if you divide this by four there so there's about this many radiology reports all right so there's 166 or thousand reports that corresponds to these nine hundred thousand images because it's four to one plus or minus and a subset of those so the rate you know every image gets read by radiologists and if it's suspicious ago some pathology so a subset of these three thirty three thousand or so of those reports went to pathology meaning it's suspicious for cancer and a pathologists read it so these are the actual exams and these are the number of patients so on average you know 10,000 patients have 32,000 exams two or three for patients right so our job is how do we turn this stack of data that's completely unrelated they're in different departments radiology is one Department pathologies in other department and they don't talk to each other right so how do we turn that into a label data set that we can use well we use deep learning and I'll show you what we did so here's though here's the
48:31
problem where is the cancer diagnosis it's in a report that looks like this patient you know there's a whole preamble here but the cliffnotes it is because the impression here are the samples that were sent ABCD and E two of them are breast this one and that one the rest of them are lymph nodes and the cancer is in this one all right you can read it if you like who's gonna sit there and read 30,000 of these things so what we did is you read 3,000 we read 3,000 and then we trained a different kind of neural network Oh STM network to predict over the remaining 27 odd thousand reports whether it's the left breast that was positive the right breast that was positive or neither breast was positive right so those are the possible outcomes and we did pretty well right so we
49:21
trained we read manually read 3000 news reports we tried a bunch of different NLP predictors and we found we were able to publish our results in a peerreviewed journal I mean here are
49:33
the au C's across the 1 2 3 4 5 6 7 different algorithms you see like regular old logistic regression does very well to predict left breast right breast cancer not cancer we published
49:47
this not not too important but if you want to learn more you can go here last
49:54
month that came out but now we have a tool where all this routine data I just showed you we can set up an algorithm to label these imaging right so woman shows up January 1st by larold screener which is imaging it suspicious she comes back in February for another set of diagnostic imaging that's two views versus for a little bit different magnification middle of the February she goes for the first biopsy because this has been suspicious that biopsy is a needle biopsy a fine needle pathologist looks listen says what this woman has cancer comes back two weeks later she has a lumpectomy which is indeed curative you know so the fine needle the pathologist thought it was DCIS which is carcinoma in situ so not full invasive cancer but by the time she came back for the unpackin which is more tissue more precision and a diagnosis so to speak they upgraded her to invasive ductal carcinoma so she has fullblown Frank cancer the lumpectomy is curative because you could see later on in August she's now negative for any pathology right so you know we can read these things off of the reports now and given that we can read them we can assign the imaging the worst possible clinical label right so I DC invasive ductal carcinoma is worse than DCIS so both of these imaging should be labeled IDC because it precedes the IDC that was confirmed and this one is negative does that make sense now we have a very straightforward algorithm so now we can
51:27
take our you know 900,000 images and put them in terms of who had cancer and who didn't pathology again cancer is a pathology diagnosis so we have 36 odd thousand negative cases of images 21 are positive thousand positive cases of pathology confirm cancer and 700 odd cases never went to pathology they're all negative right so this is pathology and what's interesting now we can have a war of the specialties because radiologists also score these things based on what they think this is this by rat score zero one two three four five and six four five and six is confirmed cancer zero is come back for another scan basically maybe cancer maybe not one and two is not cancer so you can see just based on the numbers the pathology more or less lines up with the radiology at least the scoring so now we have a massive data set how do we know what's right well how do we even know that
52:23
breast cancer screening doesn't work well places like this which is a national multimillion dollar funded multiyear study over the last 20 years the breast cancer surveillance consortium has been literally calling up patients getting their notes reading the notes and recording which mammogram was true positive which mammogram was false positive true negative false negative and calculating the statistics that tell us who the best cancer screening really has not worked as we thought it has all right so we one of the biggest sites for this BCSC is in San Francisco so we took 200 of our patients again these people are called up you know this is a very timeconsuming process 20 years to get this amount of data that the bcs he has so we asked them for these 200 patients from UCS what did you find are they true positive radiology findings or negative or false positive or false negative 15 cases we disagree with them because of semantics so they're an epidemiological surveillance consortium so they only call the first screening of cancer the first positive is the only one they care about they don't call subsequent ones positive so we could fix that semantically we missed two cases very honest mr. cases because these patients went somewhere else for their pathology so we never had the pathology to begin with so we estimate somewhere about our precision and accuracy is around 95 point something percent 99.5% so we're very confident that you know we can recapitulate what took 20 years for this consortium to do in a few months with computers and how do we know it works
54:05
because given the labels the imogen we can learn breast cancer screening right so we took this is a very simple data set we're working on much bigger ones but this is training a model to predict whether a mammogram should go for biopsy and we beat humans currently because all of these have gone to biopsy and our model would predict would save I don't know the exact numbers because this keeps changing but the fact is we can learn which means that the labels we ascribe to the images are not random which means that we're actually pulling out left breast right breast bilateral cancer or not so this is encouraging and if you can do one and zero breast cancer or not cancer there's a whole pathology
54:49
there's a whole ontology of breast pathology that we could predict an extract from pathology notes alright so for now we've you know so there's lymphoma metastasis breast cancer proper and then all the benign stuff all the stuff that is not cancer a long tail of atypical findings on pathology for epithelial findings and as you can see a long tail of non epithelial of non cancer breast pathology so we're working on the next version of this which is basically taking an image and predicting things like DCA s vs. LCS which no radiologist can do we're working on that because we have such big datasets but the future is in
55:31
threedimensional breasts radiology all right so as I started off this talk you know we transition from an era of film into twodimensional mammography so we're going through the transition from 2d into 3d mammography and here's what instead of for you know a 2d mammogram is something like um 10 megabytes you know for images a 3d mammogram is something like 80 slices 10 gigabytes so you have a much higher resolution arguably higher noise arguably a more difficult task for a human to do there's no difference between 2d and 3dimensional performance for radiologists and how do I know that
56:14
because the FDA regulates every single mammogram that's run in America so we know film took 15 years to extinguish this is the 2 dimensional wave let's say and we're in the 3 dimensional wave this is probably up to here by now because we're a year out and these are the nine thousand facilities that the FDA regulates and what they report as the technology that they use for breast cancer screening right so so which means we have about 10 to 15 years it took BCSC 20 years to collect data to show us that cancer screening doesn't work and that's based on film all right where that data can hardly be use for for prospective training because we don't do film anymore we hardly do 2d anymore we're moving to 3d so my point
56:56
is we have a very small period of time to collect this data process it and learn from it if we expect to change these statistics like I said 2d is really no difference in sensitivity it's probably harder for a firstyear resident going and looking at you know 10 gigabytes of data versus 10 megabytes but here's the really damning statistic
57:16
if you ask me one in four cancer screen or missed one in four right so why Cannot cancer be missed it could be missed because it's just fast growing it wasn't there when you screened at the beginning of the year you know every year every two years depending on your stratification you get screened the cancer might just be fast growing okay all the radio I'll disagree it wasn't there it's there now it thought forbid if you're this radiologist but you could have made a mistake as a radiologist so you miss this interval but this is the this is the cancer I showed you in the beginning these are called radiographically occult breast cancers you know seven to thirty percent or so of cancers or in this category and if you look at the screening population it's a very tiny number you need something like three thousand screeners to pick up one of these radiographically occult breast cancer so with the data set we have now in fact with the Hobie CSE for twenty
58:11
years across six different states they've screen sorry they've screened
58:15
over two million women ninetyfive thousand cases of cancer there's less than 300 interval cancers in that data set and that's over 20 years multisite collection so if we expect a thousand examples of radiographically who caught breast cancer  you know like we have a thousand dogs and cats that we can discriminate we need a lot more data than UCSF ever has at least ever had even if they kept the film
58:38
so where did block 10 come into all of this no one institution really has enough data to solve some of the problems that are very clear in breast and I'm sure other cancer radiology right so I came up with this idea CCSF where some digital sponsors the hardware to get five million mammograms donated across institutions because no one institution has that amount of data and the way it's gonna work is you know the
59:06
way it works now patients donate data to physicians they you know they go to their physician physicians put the data in the EHR and scientists are supposed to get this data from the EHR but it doesn't work like that not in America at least it's a capitalist health system and big health systems they go on to their data in spite of the laws that exist that actually mandate portability of the data you know this hightech Act says everybody has to have to electronic medical records this HIPAA Act is supposed to make the data portable but there's also a privacy law that everybody hides behind and I can understand if I was a big healthcare institution I would just not give out data and avoid being fined so what we built is on existent systems
59:51
right so all of these have systems so the imaging that the doctor sits in in Germany and looks at imaging for breast radiology is the same PACs system the same communication protocol they set in San Francisco and look on so any doc any PACs can be managed to another pacs seamlessly for the last twenty years we've had that so the images that's the infrastructure is ready there hl7 is the new kid on the block not that you but this adoption of hl7 for the other kinds of data the pathology reports the radiology reports the labs the drugs so on and so forth also exists today so we put together this infrastructure to prairie PACs directly you know the law says if a patient wants me to see their report nobody can stop that that that health care institution has to share with me either as a doctor or as a scientist or anybody the patient has the right to their data it's their data so we built this infrastructure and
1:00:44
now we've very simply asked the next month is Breast Cancer Awareness Month and we're in pink right and from next month women can sign up and get this data delivered to their phones or to their website and pull it up with their phone and from you know for the first time ever I arguably they can see they're imaging but like I just said if I want five million mammograms each one is ten gigabytes of to to of 3d data that's a ton of storage that's 250,000
1:01:10
dollars a year to store that data in Amazon that petabyte scale storage I don't have two hundred fifty thousand dollars here sort of data in Amazon which is where this cryptic urn would not cryptocurrencies I'm getting confused in the first talk which is where blockchain and sort of the idea of cryptography comes in because today this IP is you know so here's the cost something like two hundred quarter million dollars a year to store petabytes kill data in Amazon right so ipfs allows us to distribute that costs right so women can actually donate hard drive space to distribute redundancy of this data this data can never be lost it cannot be destroyed it should not be destroyed it's invaluable data so essentially I can just have one Western Digital donated hardware and redundantly back it up over IP FS so anybody who wants to contribute women can actually own their data see their data touch their data and eventually sell it which is where this is going so
1:02:04
just to close I'm out of time centralized data stores are insecure I don't need to tell this group of computer scientists that but when you see the show up on the Grey's Anatomy finale like it's a big deal right so
1:02:17
this is Europe right so I don't know if you know but a couple years ago the National Health Services or system in England was hacked asking for Bitcoin this is a nefarious he's a Bitcoin we don't any don't definitely don't endorse it but look why did this happen and happen because NHS has a centralized
1:02:37
network right so there's one point of failure it's attacked and it's shut down the hospital you know patients miss surgeries and miss you know care was affected people were affected but who could predict sort of the geopolitical forces that led to this if you look these three guys up in internet this is the one picture you're gonna find because NSA developed the malware that WikiLeaks publicized and then North Korea weaponized to go shut down the NHS health system like who could predict that like that's crazy so the security is one aspect that blockchain is supposed to solve but
1:03:12
besides that the way that the data works today there is no mechanism to buy patient data there's no mechanism to go get patient data these deals are made behind closed doors which means you're only going to get the data that the hospital is willing to sell you now this is medicine right this is not physics and it's not and it's not chemistry which means why would I think that data from the west coast where there's a whole different set of exposures and ethnicities and biology is going to be comparable to data under on the East Coast we've seen this before in genetics
1:03:42
96% of genetic samples are Caucasian and we don't want to see this in the age of
1:03:47
AI because Caucasians are not getting the most of the diseases so just a close
1:03:53
and I set this up very nicely for the next talk on the details we want to move to a more distributed system of storing this data that empowers not the healthcare systems to own it but the patients that you know they pay their pain and their suffering they should really own a benefit from I think I don't have this all figured
1:04:10
out yet but I think blockchain is the way to sort of democratize and sort of build trust into the system and I'm down
1:04:18
to one slide here now this is how it works now Google goes around at bicep all the data it goes by its John's data Jack's data and Jos data and there's no
1:04:28
other way for Google to get it like I'm not chastising Google how else they're gonna get the data they got caught in England they paid a bunch of fines because apparently it's illegal and it's not just Google a number of startups are
1:04:38
having the same problem this was last week sloankettering people are upset they're starting to realize what's going on we've seen this before like we've
1:04:46
seen this with Napster and lamb wire and BitTorrent and movies and media but what I cannot be happy with is that this is people we're talking anybody knows that
1:04:58
this is this is Henrietta Lacks no biologist in the crowd you might have heard of HeLa cells seven eight pharmaceutical companies have been founded on this woman's genes she had cervical cancer in the 50s I think she died in the 50s and some lab rat took her cells immortalized them and every biologist knows about HeLa cells they're her cells never compensated poor black woman from John Hopkins it's the saddest story ever Oprah made a movie about this you should check it out and yeah so blockchain is supposed to
1:05:28
move us to federated types of machine learning where the data can stay private the algorithm moves around we don't just buy up the data and lose value for the patients but this is a feature this is going on at Oxford I believe Anju terraces at Oxford and I will end there
1:05:44
thank you thank you very much Dexter I'm so sorry but I have to skip questions now because we are running very late but again I promise me have time for questions later on but I would
1:06:00
like to welcome me hi ELISA know his founder of Akasha and cofounder of ethereal and founder of Bitcoin magazine so he's very very innovative just Akasha is a new type of social networks that's his idea for the future of the Internet and I think he will tell us much more about that in a few seconds welcome so hello and guten tag Heidelberg I'm very
1:06:35
excited to be here and deeply honored this is in my opinion truly one of those standing on the shoulders of giants scenarios and I hope to showcase some of the applications the work of some of the Heidelberg laureates in the field of mathematics and computer science have enabled and I just like to make clear
1:06:58
from the beginning that this is not going to be a presentation about cryptocurrency I think Demelza is from the University of Liechtenstein made a great job at covering that angle this is going to be more about the deeper implications and applications of this technology and if you look at the bigger picture we are now approaching the first decade since blockchain technology so to say has entered our reality and I think the laureates in in this audience or those that will watch online have contributed greatly and today it's almost as these applications and the stuff that we are able to build on top of their work is a ripple through time of some of the work that they have done decades ago like for example elliptic curve cryptography so my goal today here
1:07:50
is to share some of like our story as a kashaf at the foundation our research areas which I hope will be interesting for some of you and also I invite you to collaborate this is not by no means a charted territory on the contrary so this is gonna be another selfdiscovery and for everyone like the on this note I don't think there is such a thing as a watching expert yet because it's too early and that being said one
1:08:23
of the team that made us want to start on this road is that we have we find ourselves today in this society and Dexter touched a bit about the implications there were the deeper implications in our daily lives where we have outsourced our privacy our freedom of expression and collective memory as a species to a handful of corporations and I'm talking about planetary scale here and this would be an overly simplified
1:08:51
view of the interactions we have on a daily base you know you post something on Facebook you read an email but at all times you are going through at least one entity in most cases we are talking about tens or even hundreds of third parties that access and monetize in all kinds of way your data and to kind of you know showcase a dark side of this
1:09:13
like what happens when you change that data when you post that message when you add that reaction a Facebook and this is not very well known and it surprised me that actually has a pattern that can predict when you'll die so this also kind of makes you wonder like Google and all these big corporations where where this is going as an information society and what it means for for our future and the children that will follow on that you know today are born with smartphones in their hands so to say and how will their future look like especially when you add AI to the picture and that being said Akasha is a nonprofit born at the
1:09:53
intersection of blockchain and collective intelligence and I'll touch about on the collective intelligence a bit later and we seek to advance the design and application of technologies for sustainable systems at city country and planetary scale so it's all about like localizing and trying to find the solution in the right context but also heavy having it interoperable at the global scale and we're doing this because we want to
1:10:19
see a world where human rights are default rather than features and I'm talking about things like freedom of speech and privacy a world where people own their online lives and identities so not like a service that grants you access to the service and which can be revoked at any time and the world where you're you know human relations so to say your social graph is not owned by a corporation which again like your friends that you're talking with and you're exchanging information at any point if Facebook or whatever social network you're using that is centralized can determine if you are still able to contact your friends which is not okay right kinda makes you wonder what kind of work we leave to the next generation and I hope with this work and the work that will continue to go on in this space we believe a better world you know what to be proud of like yes I've memorized oh maybe 10 years from now and to give you a bit of a context my story
1:11:19
in this space began in 2011 when I discovered Bitcoin and it was one of those moments like is this real and when I was trying to understand the technology and if this is a real innovation or you know scam I I spotted the need for a coherent source of information but for people that are trying to learn about it and also for journalists that were trying to find someone to talk with like what is this thing how does it how does it work and a few weeks later after I you know made an impression I okay this is real I got in touch with vitaly put her in and then that was kind of the early days of Bitcoin magazine and my story continues in 2012 where I served as
1:12:06
editorinchief for the magazine and in this time I think this was a you know an invaluable opportunity to immerse ourselves in this space and to see it from inside out not from outside in and that gave us the opportunity to meet interesting people share interesting ideas and see many projects rise and fall in this period of time it also kind of gave us an idea of what this technology can do what it cannot do where reality meets expectations you know it's not gonna solve all our problems but maybe it can help us all some at least and this is how in 2013 the idea behind the theorem
1:12:45
started to take shape in Vitalik mind and I joined him in in this project as a cofounder and it's worth mentioning that aetherium as as a concept came in an era where most of the block chains were singleuse case so if you take Bitcoin as the first example of this the use case of that party needs to be decentralized digital currency right and in in this case we proposed the idea of a generalpurpose blockchain that would I'd say liberate and empower developers to cocreate on top of a shared infrastructure this kind of points back to Internet and how you have all these various devices from phones to laptops able to communicate with each other in the same way if you'd have a blockchain ecosystem where everything is fragmented it will be harder to inter communicate between them right so that's like the synergy kind of core value proposition of the project in 2014 I landed in
1:13:43
Switzerland together with a group of people where we set the base for a team Switzerland it was a search in the early days we tried to find the right jurisdiction for obvious reasons still gray area in most but luckily the content of truck was openminded enough and it was very welcoming and that gave us jurisdiction that we could call home so to say and fast forwarding a couple of months in and I suppose you know the
1:14:10
slide that the massage showed with the mm I see rows and coins like we I'm sorry we kind of started this it wasn't called a nice your back in the day but we were trying to find a way in which we can raise resources to pay developers and people that would help us to actually make the dream a reality so to say and this is how in 42 days et Liam
1:14:35
and if you look at it this is still an ongoing so what's on number one but
1:14:39
aetherium actually set the world record eighteen point approximately eighty eighteen point five million dollars raised through a peertopeer currency
1:14:49
Bitcoin in this case to build the next wave of a digital infrastructure so it's almost like a cyberpunk inception movie where it's like the digital economy is building itself out and also to frame this this was happening in a period of time where we had this vibrant community of people selforganizing around the world so this was not topdown plan and just sharing knowledge among each other the social learning was a very important part of this in 10,000 member in 2014 and now going to 2015 in July 30 we had
1:15:24
ATM blockchain Genesis which you know moved ATM as a project from the dreamland to a platform where other people can build their dreams and also like to make a connection with the first presentation like if we look now like
1:15:38
fast forward in three years right this is the top ten of the highest crowd
1:15:43
funded projects and you see like from eighteen point five millions like the numbers jumped like four billions or hundreds of millions but like what's
1:15:53
interesting and what I like to attract your attention is that seven out of ten are done on ETV so if this is one of the
1:16:01
lowest hanging fruits of this technology but it enabled people to raise funds for four projects in a new way by passing the need to go to essentially you know a VC or other set up that involves a third party and in this case like in the three years that have passed we have now surpassed 1 million people
1:16:23
selforganizing around the world so 1 million 150 thousand people around the
1:16:28
world like epic right 100x and going back to 2015 this was when you Kieran was launched I also told Vitaly can the rest that I was always excited about the things that can be built with this thing so I never saw like the launch of the blockchain as a finish time more like a starting point and this is when Akasha is an application as a decentralized social network started to take shape and in 2016 we launched the alpha version which this is a preview of what it was able to do basically what makes this application different from other applications is that it was running completely decentralized so basically if someone would install this application on their computer in the background they would install an aetherium node and an IP FS node and ipfs being the interplanetary file system which acts as a
1:17:19
complementary piece of the puzzle to store the information and the blockchain being as more of an index and as you see
1:17:26
since we are using Bachchan for identities and people usernames every identity is also a wallet so this opens a very intriguing ways for people to not exchange only status messages and posts but also wealth in their social network so this adds another dimension to social crowdfunding of sorts and then going to 2017 we unveiled our beta which is an iteration we learned like we looked at the Alpha ok we can do better but the purpose of the Alpha was to prove that this idea can actually work to have something that does not rely on servers and the iteration was going more towards parallel experience and in this particular video we also explored the idea of having a multilayered curation so it's not like you have just a black and white approach like you have the thumbsup thumbsdown but you have actually a spectrum of opinions that build when you overlaid some sort of emergent picture of our reality and in the meantime you know in this time while we were focused on our own thing the ecosystem around the theorem and people building various things and not only theum the blockchain ecosystem as a whole has underwent like
1:18:34
like a sort of Cambrian explosion of things and if we zoom in hopefully let's
1:18:38
see if this is visible we have a bunch of projects in all sorts of areas from exchanges and trading you know financial
1:18:46
services graph tech wallets and if we
1:18:50
move down we have prediction markets
1:18:53
utilities insurance and healthcare legal
1:18:57
so as you can see many many many areas and somewhere here on the right we also have let's see we have also akasha as a
1:19:06
social network on the right but while we while we saw this
1:19:12
unfolding so it was like a coevolution of sorts on the platform on top of which we were building was also evolving as we were moving forward and we understood the need of shifting our understanding of what we are doing taking again a couple of steps back and looking at what
1:19:27
we actually want to achieve and this made us understand that we need to move
1:19:32
beyond like thinking of the application as a social networking app and rather putting it into an ecosystem view because this was already kind of a network of sorts but what was lacking was this inter connective tissue that will leverage the synergy of this ecosystem so now even if these applications and all these projects could talk to each other they're not very good at doing so in this concept the Akasha kind of starts to begin morphing into like this emergent framework rather than something overlaid on top and moving on something that you
1:20:11
know to summarize this to some extent the shift from the current paradigm to the next one is this Reedy centralization so the early ideas of the internet and so on all revolve around having no central point of failure but the hyper centralization we saw manifesting you know Facebook billions of people Google billions of people in half a planet have led to this like super how to call it like an information dead star of sorts right that is not necessarily contributing back to society and if you have that pyramid of data information knowledge wisdom like our society as a whole even if it's an information based society has a very detailed knowledge and arguably less wisdom right because all that data that we are producing all that information that we are producing is not siloed captured and monetize analyzed by these corporations and that the knowledge or wisdom that is derived from that is not necessarily to the betterment of humanity rather the betterment of the algorithm serving ads and in this paradigm you see the user in the decentralized paradigm the user as being in control of their flows and this involves the social networks which can be you know a social network at work a social network at home a social network with your friends and then you have your own things in the house in this context the Internet of Things and the user through this identity and security layer which could be framed as a blockchain infrastructure but you know it's not like the user has to understand blockchain in order to use it just like you don't need to understand electricity in order to turn on the light you just press something about and turn on turn off and then this feedback again can happen locally on the user's device so is there's no need for that huge data center somewhere to collect that we've moved past that stage we all basically have super computers in our pockets and to kind of give you an idea of where
1:22:06
we're going and if there might be some overlap between our interest our research centers around three main pillars like what who and how so when I say who it refers to who are you who are
1:22:19
you are you delimited by your physicality as you know I am this trillion of cells just standing here and giving this talk am i at rate or is it my identity is limited to the passport my government gives me the same with a username that you have on a social network same with a cryptographic identity which can be represented by a key pair but you know you have hundreds of projects that operate in this field there are hundreds of initiatives that kind of search to break down and this challenge from this perspective but something to note here that this is a binary universe where you either have it or you don't you're either in or you're out right but that's not really how the world works in reality we are more like this fluid multidimensional beings it's not like you can be summed up to just one trait or what one passport or to one driving license right or to one username it's more like we have this separate and sometimes sometimes only overlapping circles of friends reputation and knowledge that we can build on top of and this is the area where we focus and this is an area that's not to say too popular hopefully this will change in the future it's also a more complex to solve but you know what's the point of researching something if it's not challenging and when we touch on the what you know and
1:23:47
as you can see the presentation it does not focus very much on the technology but rather the implications and where humans and technology kind of come together and in this case how many of you are familiar with da OS can I see a show of hands okay mom not bad I've seen a few hands so basically a decentralized autonomous tributed autonomous organization is a new form of organizing for people right so if we look back a couple of hundred years ago we had the corporation as a way or a legal structure around which people individuals could come together towards some shared goal right so in that case it could be a mining company which has the goal of you know extracting coal from the mine a petrol and so on but in this case you know you also have the system which has a whole if you look at the Constitution the laws and all this stuff can be seen as a sort of protocol at societal level in this jurisdiction so this is the accepted social stuff this is what's not accepted and we operate within that so it's like the green light we can go the red light no you stop so in the same way you can define this kind of rules that help us organize and coordinate it with each other and just that instead of using you know papers and complicated laws and and so on you have this ability of creating an organization where you define these laws in a transparent manner using smart contracts by the way how many of you are familiar with smart contracts can I see a show of hands okay that's better so the smart contracts are like the building block on top of et reham which can be anything mathematically defined as a set of condition so it's like an if this then that kind of scenario but the main purpose here is that this organizational form that we come up with as a society a couple of hundred years ago we're all relying on one central point a president a CEO you know a general manager but in this scenario where everyone has connectivity everyone has an access to you know to being connected with the whole group it's not so much a problem of coordination and getting people together which was a problem a couple of hundred years ago but not so much today so the stakeholders the participants in this group can all signal like where do you want to go similar to how you have the Flamingo Birds you know like singling signaling with their heads and depending on where the most point that's where they go so in a similar fashion which might seem like chaotic you can have some sort of order emerge and that's the decentralized part and now coming back to you know the setup of
1:26:29
this event this might seem familiar to some this is a schematic from the mathematical theory of communication also known as the Shannon Weaver model which treats the unseen the communication right a theory of communication and this it's fantastic this is like one of the building blocks that enabled us to build many things from like betting satellites understanding black holes and you know most of the internet and applications we used today are still built on top of this and this works great because a computer's message so when you have messages signals that travel to this framework this applies very well however when you add people like humans into the picture it's almost like we're missing a piece of the puzzle and this is a where conversation theory comes into play and again I'd like to point out that this is like decadesold research that was done by Gordon Paz and nowadays is continued by Paul pangoro through his work and in this scenario you have a setup where the participants that use the signaling the messaging framework can coordinate and come together to achieve a common goal and this is formalize this is like one of the applications of cybernetics in the context of how humans and machines can you know leverage their overlap to increase their and in in this context like putting back the blockchain you know this is decades old this is all fun and so on but if you put back in the blockchain context now you have this idea of a
1:28:14
conversational Dao so on one side you have the transactions and people being able to signal but you don't necessarily have the interface that would enable these participants to easily and efficiently come together and towards a common goal and this is quite exciting and this is a quote from pal pangoro that says well creation has shifted from prior
1:28:39
knowledge like patterns IP and you know the classic to the ability to gain new knowledge in action because also the
1:28:46
world around us is like constantly shifting and accelerating in this change right it's a lot more fast than it was a couple of hundred years ago when we had this form of organization that might have been you know ok but these days I think we are reaching the limits of what this form of organization can do at the planetary and you know country or local scale and also something to note that
1:29:10
this is taken from economy of insights conversations as transactions you know
1:29:16
so again putting into the context of of blockchain and to kind of wrap this up
1:29:21
and touch also on what Dexter was talking about when it comes to data as in something that can be applied also on the medical side there is a question here is it something you own some people might say that but if you look at it as something you own you get to like some weird ethical debates as in you know you either own this jacket or you don't I have it or you do but not the both of us and I think in the in the next iteration of what this web kind of infrastructure opens we have to go beyond that and we got we have to go towards an era where our understanding of what the data is is not something we own but something we actually are because from those patterns going back to what phase you know like predicting when you can die that's actually us it's almost like a doppelganger of ourselves in a digital form and this is why again privacy and encryption and this platform should you know excite people because it can empower them and again we take control over this whole infrastructure that was developed around us while we were sleeping or scrolling to the feed and to wrap it up how we can do that self sovereign technologies are defined as something but you know there are many many definitions but one of them is that they serve you and no one else and in the context where we explored how these entities are in the middle and exploiting basically this human desire of connecting with another because we're social humans social animals and then you have like the social network as the killer application of the Internet no surprise there right we also can create this infrastructure where that desire of each other connecting with another is done on a peertopeer basis like it was done hundreds of years ago like it was done thousands of years ago before we had this infrastructure developed that just captured and monetized our relations and our interactions with each other so to put
1:31:39
it in you know in a nutshell the collaboration with Paul pangoro for us was a was an example how transdisciplinarity can help us and how sometimes old research you know in the decades old research can be applied in the new context and in a world in a you know in a world where things can be everything is a remix and sometimes old ideas in a new context can come up with new solutions we think it's it's the same now and I think we live in an age where the gap between when math is being invented and when it is being used is collapsed so you see we have these decades old lag and in research like for example elliptic curve cryptography which has at the heart of many block in projects and it's almost like we see this ripple through time right there was these smart guys coming putting down equations and now decades later there are these cypher punks you know just stumbling upon something yes that that can work and then you just combine it in a way to make a meaningful application and I think new and all mathematics and computer science can truly make a difference in years rather than decades if we come together and we explore the the interest the shared interest and shared goals we have as you know a group of a community just appreciating the intellectual challenge and the curiosity in finding new solutions to sometimes all problems and the idea here is that we invite everyone that is interested in exploring this or has some inclination towards blockchain as a you know a tangible application in our real world to get in touch and basically to turn that beautiful mathematics into a meaningful application that can truly make a difference for people around the world thank you
1:33:41
[Applause] [Music] you