This is the first video I've seen of Gregory Maxwell. This adds some confirmation for me of my upthread speculation about Greg seeing himself as critic and the smartest person in the room. He specifically states in this video that his role is more as a reviewer than a doer (even his stated goal is maximum impact with the least coding...which is a desirable goal but only if it is not the only one), right after admitting that he was wrong in 2004 about decentralized consensus being impossible. The audacity. Socrates taught us that recognizing that we are not omniscient is a primary attribute of cognition.
(Edit: in the "Selection Cryptography" portion of the video, he elaborates on why his role is appropriate — "Pragmatic has its place, but beware against biasing against competence")
No doubt this is a very smart guy with powerful crypto+math domain knowledge who can add considerable analysis and even new ideas. You'd definitely want him on your team (I would) if he can contain himself to a non-leadership role. But hand him the keys and you are likely to go too far down dead-end paths—e.g. CoinJoin—because my impression of him so far (limited interaction) is he is more of a narrow space thinker who doesn't pay as much attention to what is going on in the kitchen when he is in the basement (unless if he a lead on a very narrow space, orthogonally contained project domain such as an audio codec). And this is precisely what I told him the very first time he spanked me in public in these forums; I warned him that I am more of a pragmatic generalist and that we tend to paradigm shift around people like him (which is precisely what I am hoping to do accomplish this year). The first exposure I had to Greg was when I was very impressed by his forum post containing analysis of a proposed proof-of-work hash for something bytemaster was proposing (I forget the details).
I am taking a deeper look at Blockstream and side chains today for the first time. I will report back my findings shortly.
P.S. I am only 10 minutes into the linked video and it is particularly poignant so far. I highly recommend it. So far it appears to be making the case for Monero. It admits Tor is weak against the State, which is a concept I was promulgating since 2013 and was initially resisted (afair by Greg and many others). Good to see that my work in the forum in 2013 finally was accepted. I will say as AnonyMint, I was pushing hard for greater in anonymity starting in 2013 with some posts I made in the anoncoin thread.
The anti-"Eclipse" (Sybil attack mitigation) slide at the 11:45 minute mark is a very important point. An inherent weakness in PoW is the simultaneity requirement which makes network topology so critical (think orphan rates, selfish mining attacks, etc) and which is also difficult to anneal because of the self-referential relativity of the paradigm. This is one of the main fundamental flaws I want to address in crypto currency. I had referred to this recently as "anti-aliasing". Ironically where Greg says Satoshi's solution to decentralized consensus achieved something "not quite as strong" as that which he thought he had proved was impossible, the same applies to my solution to centralized consensus which implicitly resists centralization and scales to micropayment volume.
Another interesting point at 21 minutes, that multisig negatively impacts decentralization and scaling. This is another fundamental aspect my solution fixes. Schnorr as a potential solution to this problem for Bitcoin lacks optional accountability (or loses some of the scaling and decentralization advantages with Greg's TREE scheme) and has a simultaneity issue vaguely analogously related to the problem with CoinJoin.
Regarding Greg's redefinition of Cryptography at the 40+ minute mark, he is on the right track but I fundamentally disagree with his definition because it presumes we could both exist and information could be entirely free, i.e. it is vacuous because it presumes mutually incongruent assumptions. The generative essence is that the entropy of the universe is trending to maximum (i.e. not infinity) according to the Second Law of Thermodynamics (refer to the upthread philosophical discussion about existence and theoretic physics). I would instead define, "
Cryptography is the art of structuring information such that hidden entropy doesn't collapse to 0 over known domains in time and computability".