What a bunch of BS. It's obviously easy to say that in hindsight. There are a million reasons for why he did not take the investment, but in none of the published conversation was anything about him thinking SBF is lying. In fact, knowing Elon, if he had really said that at some point before FTX collapsed, he sure would have published it by now to prove how smart he is.
He was skeptical even though SBF wanted to give him money, not the other way around. And people Elon knew were pushing for SBF. Elon ended up ghosting him anyway. Look at the chat logs that were revealed in court.
Michael Grimes [IBanker at Morgan Stanley]: Do you have 5 minutes to connect on possible meeting tomorrow I believe you will want to take?
Elon: I'm backlogged with a mountain of critical work
matters. ls this urgent?
Michael: Wants 1-Sb. Serious about partner w/you. Same security you own
Michael: Not urgent unless you want him to fly tomorrow. He has a window tomorrow then he's wed-Friday booked
Michael: Could do $5bn if everything vision lock. Would do the engineering for social media blockchain integration. Founded FTX crypto exchange. Believes in your mission. Major Democratic donor. So thought it was potentially worth an hour tomorrow a la the Orlando meeting and he said he could shake hands on 5 if you like him and I think you will. Can talk when you have more time not urgent but if tomorrow works it could get us $5bn equity in an hour
Elon: Blockchain twitter isn't possible, as the bandwidth and latency requirements cannot be supported by a peer to peer network, unless those "peers" are absolutely gigantic, thus defeating the purpose of a decentralized network.
Elon: ["disliked" "Could do $5bn ..."]
Elon: So long as I don't have to have a laborious blockchain debate
Elon: Strange that Orlando declined
Elon: Please let him know that I would like to talk and understand why he declined
Elon: Does Sam actually have $3B liquid?
Michael: I think Sam has it yes. He actually said up to 10 at one point but in writing he said up to 5. He's into you. And he specifically said the blockchain piece is only if you liked it and not gonna push it. Orlando referred Sams interest to us and will be texting you to speak to say why he (Orlando) declined. We agree orlando needs to call you and explain given everything he said to us and you. Will make that happen We can push Sam to next week but I do believe you will like him. Ultra Genius and doer builder like your formula. Built FTX from scratch after MIT physics. Second to Bloomberg in donations to Biden campaign.
Will MacAskill [co-creator of effective altruism movement, Oxford professor, Chair of board for Global Priorities Institute at Oxford]: Hey - I saw your poll on twitter about Twitter and free speech. I'm not sure if this is what's on your mind, but my collaborator Sam Bankman-Fried (https://www.forbes.com/profile/sam-bankman-fried/?sh=4de9866...) has for a while been potentially interested in purchasing it and then making it better for the world. If you want to talk with him about a possible joint effort in that direction, his number is [redacted] and he's on Signal.
Elon: Does he have huge amounts of money?
Will: Depends on how you define "huge"! He's worth $24B, and his early employees (with shared values) bump that to $30B. I asked about how much he could in principle contribute and he said: "~$1-3b would be easy-$3-8b I could do ~$8-15b is maybe possible but would require financing"
Will: If you were interested to discuss the idea I asked and he said he'd be down to meet you in Austin
Will: He's based in the Bahamas normally. And I might visit Austin next week, if you'd be around?
Will: That's a start
Will: Would you like me to intro you two via text?
> Elon: Blockchain twitter isn't possible, as the bandwidth and latency requirements cannot be supported by a peer to peer network, unless those "peers" are absolutely gigantic, thus defeating the purpose of a decentralized network.
Anyone able to evaluate this comment? Is this another "poorly batched 1000 RPCs slowing down home timeline" remark?
It's true, and the RPC comment is also very plausible
It's definitely not as ridiculous as what lots of experienced engineers are oddly saying
e.g. simply open up Twitter in Chrome's devtools and look at the networking tab -- I get 148 requests that finish in 6.3 seconds
That is very plausibly "1000 poorly batched RPCs" that makes Twitter slow in other countries -- if it's 6 seconds for me, it's easily 20, 30, or 60 seconds for others
It will be shocking to me if some executive attention on latency can't halve it, in short order. The problem with latency is that no one team is in charge of it -- every team is incentivized to use as much latency budget as possible to ship their feature.
Larry Page harped on at this at Google for a decade, until he kind of gave up / moved on, and Google became nearly as slow as every other website.
So Elon is absolutely doing the right thing with respect to latency -- he's not wrong, and he's not micro-managing.
----
Yes Elon is not infallible -- he made flatly wrong claims about self-driving for years, despite experts in the field telling him otherwise, to his financial benefit. And Tesla is being rightly investigated for those claims (years too late, probably)
It doesn't mean that Twitter isn't slow as hell for extremely basic reasons
I made a similar comment on Reddit and got voted down because Elon fanboyism has apparently turned into Elon hate. People really love to "pick a side" and ignore the facts on the ground.
Simple changes like enabling HTTP/3 and IPv6 would dramatically improve performance in locations outside of the United States. Google developed those specifically to work around performance issues in places like India, or mobile networks pretty much anywhere. Since Twitter is mobile-first, it surprises me they haven't jumped on board with these kinds of advancements.
What were all those thousands of people doing over there!?
Theoretical TPS limit for mainnet and L2s isn't enough to support the scale of twitter. The writes will get expensive very fast unless you offload that and only send a batched proof. Putting actual content on blockchain is a non-starter as well.
Most of the blockchain based social graphs end up using centralized indexer.
Everyone is also dependent on centralized RPC providers for querying and sending transactions since it requires a lot of resources to maintain your own.
So peer to peer isn't possible at their scale depending on how you interpret it.
