The WSJ dashes our hopes for a quiet Christmas by dropping on Christmas Eve a further extension of all this reporting: "Sam Altman’s Knack for Dodging Bullets—With a Little Help From Bigshot Friends: The OpenAI CEO lost the confidence of top leaders in the three organizations he has directed, yet each time he’s rebounded to greater heights", Seetharam et al 2024-12-24 (Archive.is, HN; annotated excerpts).
This article confirms - among other things - what I suspected about there being an attempt to oust Altman from Loopt for the same reasons as YC/OA, adds some more examples of Altman amnesia & behavior (including what is, since people apparently care, being caught in a clearcut unambiguous public lie), names the law firm in charge of the report (which is happening), and best of all, explains why Sutskever was so upset about the Jakub Pachocki promotion.
Loopt coup: Vox had hinted at this in 2014 but it was unclear; however, WSJ specifically says that Loopt was in chaos and Altman kept working on side-projects while mismanaging Loopt (so, nearly identical to the much later, unconnected, YC & OA accusations), leading to the 'senior employees' to (twice!) appeal to the board to fire Altman. You know who won. To quote one of his defenders:
“If he imagines something to be true, it sort of becomes true in his head,” said Mark Jacobstein, co-founder of Jimini Health who served as Loopt’s chief operating officer. “That is an extraordinary trait for entrepreneurs who want to do super ambitious things. It may or may not lead one to stretch, and that can make people uncomfortable.”
Sequoia Capital: the journalists also shed light on the Loopt acquisition. There have long been rumors about the Loopt acquisition by Green Dot being shady (also covered in that Vox article), especially as Loopt didn't seem to go anywhere under Green Dot so it hardly looked like a great or natural acquisition - but it was unclear how and the discussions seemed to guess that Altman had sold Loopt in a way which made him a lot of money but shafted investors. But it seems that what actually happened was that, again on the side of his Loopt day-job, Altman was doing freelance VC work for Sequoia Capital, and was responsible for getting them into one of the most lucrative startup rounds ever, Stripe. Sequoia then 'helped engineer an acquisition by another Sequoia-backed company', Green Dot.
The journalists don't say this, but the implication here is that Loopt's acquisition was a highly-deniable kickback to Altman from Sequoia for Stripe & others.
Greg Brockman: also Stripe-related, Brockman's apparently intense personal loyalty to Altman may stem from this period, where Altman apparently did Brockman a big favor by helping broker the sale of his Stripe shares.
YC firing: some additional details like Jessica Livingston instigating it, one grievance being his hypocrisy over banning outside funds for YC partners (other than him), and also a clearcut lie by Altman: he posted the YC announcement blog post saying he had been moved to YC Chairman... but YC had not, and never did, agreed to that. So that's why the YC announcements kept getting edited - he'd tried to hustle them into appointing him Chairman to save his face.
To smooth his exit, Altman proposed he move from president to chairman. He pre-emptively published a blog post on the firm’s website announcing the change. But the firm’s partnership had never agreed, and the announcement was later scrubbed from the post.
Nice try, but no cigar. (This is something to keep in mind given my earlier comments about Altman talking about his pride in creating a mature executive team etc - if, after the report is done, he stops being CEO and becomes OA board chairman, that means he's been kicked out of OA.)
Ilya Sutskever: as mentioned above, I felt that we did not have the full picture why Sutskever was so angered by Jakub Pachocki's promotion. This answers it! Sutskever was angry because he has watched Altman long enough to understand what the promotion meant:
In early fall this year, Ilya Sutskever, also a board member, was upset because Altman had elevated another AI researcher, Jakub Pachocki, to director of research, according to people familiar with the matter. Sutskever told his board colleagues that the episode reflected a long-running pattern of Altman’s tendency to pit employees against one another or promise resources and responsibilities to two different executives at the same time, yielding conflicts, according to people familiar with the matter…Altman has said he runs OpenAI in a “dynamic” fashion, at times giving people temporary leadership roles and later hiring others for the job. He also reallocates computing resources between teams with little warning, according to people familiar with the matter. [cf. Atlantic, WaPo, the anonymous letter]
Ilya recognized the pattern perhaps in part because he has receipts:
In early October, OpenAI’s chief scientist approached some fellow board members to recommend Altman be fired, citing roughly 20 examples of when he believed Altman misled OpenAI executives over the years. That set off weeks of closed-door talks, ending with Altman’s surprise ouster days before Thanksgiving.
Speaking of receipts, the law firm for the independent report has been chosen: WilmerHale. Unclear if they are investigating yet, but I continue to doubt that it will be done before the tender closes early next month.
the level of sourcing indicates Altman's halo is severely damaged ("This article is based on interviews with dozens of executives, engineers, current and former employees and friend’s of Altman’s, as well as investors."). Before, all of this was hidden; as the article notes of the YC firing:
For years, even some of Altman’s closest associates—including Peter Thiel, Altman’s first backer for Hydrazine—didn’t know the circumstances behind Altman’s departure.
If even Altman's mentor didn't know, no wonder no one else seems to have known - aside from those directly involved in the firing, like, for example, YC board member Emmett Shear. But now it's all on the record, with even Graham & Livingston acknowledging the firing (albeit quibbling a little: come on, Graham, if you 'agree to leave immediately', that's still 'being fired').
Tash McCauley's role finally emerges a little more: she had been trying to talk to OA executives without Altman's presence, and Altman demanded to be informed of any Board communication with employees. It's unclear if he got his way.
So, a mix of confirmation and minor details continuing to flesh out the overall saga of Sam Altman as someone who excels at finance, corporate knife-fighting, & covering up manipulation but who is not actually that good at managing or running a company (reminiscent of Xi Jinping), and a few surprises for me.
On a minor level, if McCauley had been trying to talk to employees, then it's more likely that she was the one that the whistleblowers like Nathan Labenz had been talking to rather than Helen Toner; Toner might have been just the weakest link in her public writings providing a handy excuse. (...Something something 5 lines by the most honest of men...) On a more important level, if Sutskever has a list of 20 documented instances (!) of Altman lying to OA executives (and the Board?), then the Slack discussion may not have been so important after all, and Altman may have good reason to worry - he keeps saying he doesn't recall any of these unfortunate episodes, and it is hard to defend yourself if you can no longer remember what might turn up...
Approximately four GPTs and seven years ago, OpenAI’s founders brought forth on this corporate landscape a new entity, conceived in liberty, and dedicated to the proposition that all men might live equally when AGI is created.
Now we are engaged in a great corporate war, testing whether that entity, or any entity so conceived and so dedicated, can long endure.
What matters is not theory but practice. What happens when the chips are down?
So what happened? What prompted it? What will happen now?
To a large extent, even more than usual, we do not know. We should not pretend that we know more than we do.
Rather than attempt to interpret here or barrage with an endless string of reactions and quotes, I will instead do my best to stick to a compilation of the key facts.
(Note: All times stated here are eastern by default.)
Just the Facts, Ma’am
What do we know for sure, or at least close to sure?
Here is OpenAI’s corporate structure, giving the board of the 501c3 the power to hire and fire the CEO. It is explicitly dedicated to its nonprofit mission, over and above any duties to shareholders of secondary entities. Investors were warned that there was zero obligation to ever turn a profit:
Here are the most noteworthy things we know happened, as best I can make out.
Later, when we know more, I will have many other things to say, many reactions to quote and react to. For now, everyone please do the best you can to stay sane and help the world get through this as best you can.