Much of our time online continues to be spent in centralized silos. A wide range of projects continue to chip away at building decentralized alternatives, but it's fair to say that decentralized platforms continue to play catch up. This talk looks at a few of the key problems underlying many schemes for decentralization, and reflect on how different projects are tackling them and what remains challenging. In particular decentralized platforms grapple not only with how to efficiently scale, how to balance resiliency and capacity, how to name and discover content and users, but also have to consider privacy and security. In 2020 we've seen increased adoption of both federated (matrix, activity pub) and decentralized (ssb, ipfs, cryptocurrencies-writ-large) platforms. Lots of decentralized engineering has been happening this year. The Interplanetary Filesystem restructured its DHT to get content discovery times below 5 seconds. Matrix grew to more than 2,500 federated servers, and got a contract with the German education system. Planetary.social extended the Secure Scuttlebutt gossip protocol to mobile. The crypto-currency community has settled on some strategies for parallel computation to address scaling pains, and are continuing to work on building systems for decentralized governance and organizations. On the flip side, there remain plenty of big questions these systems are hitting. You can't just keep scaling a DHT without the overhead of connection setups and metadata overwhelming performance. Consistency remains computationally expensive at scale, and decentralization of power and wealth are not solvable with engineering alone.