Her spat against Yann LeCun felt very wrong to me. That is not how fellow academics/researchers should treat each other.
Maybe Yann LeCun was wrong, the way she conveyed the message disregarding any knowledge he might have on this was definitely not civil.
She could have chosen a different way to address the complaints here, not by getting overtly confrontational. Maybe the review group got it wrong, maybe they were right. But that is not how you handle disputes.
I agree that people should try to be kind and charitable to each other, and this conversation didn't show that.
That said, I think we also have to be cognizant of the way "civility" is not objective/neutral, often relies on codifying the norms of a dominant majority, and also preserves the status quo.
It might not have been very "civil" for instance when Google employees walked out of the office, but I also don't think it was "civil" when Google paid $90 million to someone fired for sexual harassment.
Sometimes big problems can't be met just with civil discourse, it's why many of us still voted for Biden (for instance) even after he told Trump to "shut up."
What about paying $90 million to the key person in creation of android? Honestly, he is a scumbag, but that doesn’t nullify his contribution to google (?) it’s not that he was paid for sexual harassment, it’s that he got paid for being creator of one of the most important products in the world.
He sold Android to Google in 2005 for a hefty sum, and he drew a large salary and stock package every year since. Why in the world would he need a $90M exit package on top of all that when he's being fired for severe workplace misconduct? He already got paid for those contributions.
Because Android is arguably worth a hell of a lot more than the $50 million + salary that he received and if they didn't kick him some money, they would never be able to make a similar acquisition deal in the future. Nobody would sell a technology like that for that little and Google would have to forever compete with future-Androids.
If they had fired him under more favorable circumstances his payout would have been significantly higher.
He sold the company to them in 2005. If he didn't think that was a fair price he wouldn't have agreed to the sale. That was when he was compensated for his creation. After that, he was compensated through salary.
I don't understand your point of view here. If I sell something to someone else, I'm not expecting to get more payout for it fifteen years down the line unless some kind of residual was explicitly negotiated in the original sale agreement. Acquisitions would never happen at all if the acquiring company was on the hook for unbounded amounts of money decades later just because they made it successful. This point of view seems very un-businesslike. Andy Rubin was not owed that $90M, it shouldn't have been paid to him, and Google has admitted to its mistake and will not be paying out in similar cases going forward.
1. It was an aquihire. Usually there's some performance secondary payout strings attached.
2. It doesn't matter what their deal was. It matters what public perception is. I'm talking entirely about public perception.
Regardless of what their deal was, Google has to send a signal that _impacts future deals_. If the general developer public thinks that Android was acquired for too little, then we will all be reluctant to sell our projects to Google. That's the whole point.
It has nothing to do with him thinking Google treated him fairly. It has to do with whether we think we could keep a company going long enough to make more money than his payout on _our_ work.
Google needs to make more Android-like deals in the future.
And all of that needs to be weighed against the huge worldwide negative PR resulting from the payout, the global walkout of tens of thousands of employees expressing their anger with the decision, and the depressive effect on hiring that has for specific targeted groups. You can't just analyze the payout in isolation; you have to analyze its sum total effects. Turns out, the payment wasn't worth it. Google has said as much and won't be doing it again in the future.
> If the general developer public thinks that Android was acquired for too little, then we will all be reluctant to sell our projects to Google. That's the whole point.
I'm still not understanding the point here. If you agree to sell your company for price $X, it doesn't matter who you're selling to; you've sold for that amount. If you think you should get more than $X, then don't sell to Google, or anyone else for that matter, for only $X. If you want less up front but more money over time as part of an incentive structure, then negotiate for that as part of the acquisition (which Andy Rubin did do and was paid out for). All of this is independent from a severance payout fully 15 years after the acquisition.
> Google needs to make more Android-like deals in the future.
I think you'd be hardpressed to find a founder that wouldn't sell to Google for what they consider the right price if he didn't get $90M. Do you think any football players don't want to play for the Patriots, because Aaron Hernandez wasn't supported by them? People get that all bets are off with criminally poor behavior.
Also, take a step back and consider how there was a worldwide walkout of 10k+ employees. That wouldn't have happened if Google's hands were tied here and they had been contractually forced to pay that money out to Andy Rubin. The outrage was because they knew about his misconduct and yet still chose to let him "retire" and give him a big amount of extra money, rather than firing him and giving him the $0 he was legally entitled to.
In settling on terms favorable to two of the men, Google protected its own interests. The company avoided messy and costly legal fights, and kept them from working for rivals as part of the separation agreements.
Real life isn't a melodrama. I suspect there's plenty of nuance here.
You shouldn't get severance packages if you're fired for egregious misbehavior. Google has admitted it was a mistake and will not be paying out severance packages in these situations going forward.
> "civility" is not objective/neutral, often relies on codifying the norms of a dominant majority, and also preserves the status quo.
In this case, the norm in question was respectful argumentation grounded in empirical evidence. Why does it matter that that norm belongs to a dominant majority?
> Why does it matter that that norm belongs to a dominant majority?
It matters
a. if the norm is actually universally applied - I find that this norm of respectful argumentation is not universally applied.
Remember: this is a twitter argument entirely unrelated to what Timnit is under discussion for now. Elon Musk and Linus have had unrelated arguments where they engaged in very offensive argumentation (calling their opponents pedophiles, telling them to kill themselves) without any evidence, yet I rarely see these issues brought up on unrelated issues about them. But for Timnit, it's a scarlet letter she has to bear.
b. I think we substantially underestimate how frustrating it might be to be a scientist of color where people are constantly discounting your expertise (because, yes, Timnit is an expert.) I'm sure seeing that same sentiment for the ith+1 time can evoke strong emotions and we are not necessarily seeing the context.
Regardless, maybe not an incredible look for Timnit but not super huge in the context of a large body of work.
I've tried to piece together the debate with LeCun, but I can't find anything she said that seemed all that bad. It seemed like, if anything, she just simply wouldn't let him off the hook. And it's very possible I missed several tweets that were really toxic...
Gebru did not engage LeCun in a reasoned debate. She said he should stop taking, listen to her and her allies (and only them). She said "Let us lead here and you follow" and "Just listen", essentially telling the inventor of convolutional neural nets that it was now his role to sit and shut up.
That's how woke Twitter works: you define the group of people who are allowed to speak as only you and your allies. You pursue the meta-argument whenever someone not in your group attempts to speak.
> Let us lead [here] (I'm assuming "her" was a typo) and you follow.
Aggressive, sure, but also seems:
a. reasonable for AI ethics experts to "lead" in issues surrounding AI ethics
b. not the cardinal sin of discussion that it appears to be viewed as here.
You'll see people here defending Linus or Elon Musk for saying objectively way worse stuff than this.
Somehow Timnit is defined by this discussion, but Elon isn't defined by calling a rescue diver a pedophile with no evidence or Linus telling someone to kill themselves.
OK, but imagine that you are the expert in your field and you have work for this person, who controls your career and compensation, who has a tendency to say that domain experts need to shut up and defer to her.
If the domain is AI ethics, then Yann is not a domain expert, she is. Indeed, it seems like this exact expansion of Yann's "domain" is what she is frustrated by, and you are playing exactly into it.
> the expert in your field and you have work for this person, who controls your career and compensation
Must be harder when your CEO boss has a tendency to call domain experts pedophiles, yet I see Musk praised frequently here.
I'm not the one saying people who criticize "domain experts" deserve to be fired. I think both are within the realm of discourse and shouldn't be retaliated against.
I think your sounds ask her actual reports what they think of her management (hint: essentially all of them have said "she's the best manager I've ever had" or the equivalent in the past week).
If you work in an area that is written off, you need a strong advocate, and from everything I've seen, she was that strong advocate both for herself and for those who reported to her.
> Even amidst of world wide protests people don’t hear our voices and try to learn from us, they assume they’re experts in everything. Let us lead her and you follow. Just listen.
She says talking down to an actual expert in ML who gave an opinion on ML
"Ethics" does not really encompass the topic. The topic was that a net trained mostly on white people's faces upsamples a downsampled photo of a half-black man (Barack Obama) into a generic white face. LeCun's point is this is no surprise. If you trained it on leopards, it would upsample pixelated human faces into leopard faces instead. LeCun is an expert on how these nets work.
And nowhere did Timnit disagree that the cause he identified was a cause of bias in the system. Her contention is that there are others, and reducing the problem to just "get better datatsets" is a bad idea (and a way for researchers to abdicate responsibility).
To use an example that comes up: do we solve the ethical qualms of a system that predicts incarceration rate based on facial structure by getting a better dataset? Is it even ethical to attain such a "better" dataset, what does "better" even mean in the context?
Not sure why you are being downvoted. I see how reasoned your response is to what I said. It is a clear contrast to how the AI ethics researcher engaged in the Twitter thread. Two totally different hypothetical conversations:
A: Training data is a cause of this bias
B: But not the only cause
A: True
vs.
A: Training data is a cause of this bias
B: I am so sick of this. You are not permitted to speak.
A: I am deleting my Twitter account.
Any support for Dr. Gebru seems to be against the HN ethos at the moment, I have some theories as to why ;)
> You are not permitted to speak.
I really don't think this is a fair reading of her tweet. I don't see any way to read her actual tweet claiming Yann can't speak at all on the subject. I do see a plea to engage more with domain experts (which to reiterate: my understanding is that Yann has a history of ignoring on the subject)
Reading these threads is really disappointing. I can in may ways already empathize with Timnit -- its exhausting seeing people who are typically bright just completely miss the point.
But if you change to a different topic like privacy then these same people really get all of this nuance that they completely seem to miss here...
And plenty of people attacking him. Every Linus thread has a massive debate about him being an asshole. Just like in this thread there are plenty on each side. Where’s the double standard?
I was thinking threads from before his big "shit, i'm an asshole, sorry. gonna take time off" phase. Maybe I have a selective memory, but I'm remembering threads from those couple years before and after Susan Fowler and Uber's culture implosion. It felt like there was a shift in how the whole industry felt about brilliant assholes around then (thank god).
I think the threads in articles specifically about Linus are better comparisons to this, rather than articles about computing that involve him. But, hey, folk like Torvalds and Gebru both really rub me the wrong way, so it's reasonable that I might be selectively remembering threads where HN said I'm right and that he's an asshole (and I did a lazy googling of "linus torvalds asshole site:news.ycombinator.com" which is biased too), so who knows.
Agreed, there's even an article floating around that purports to summarize the most offensive tweets. In my opinion, this whole thing is being vastly blown out of proportion.
By the judgement of her colleagues inside google this was not a normal internal review process, and according to her own writing, she tried to engage with the review process but was blocked.
Also, to quote LeCunn:
“I’d like to ask everyone to please stop attacking each other via Twitter or other means. In particular, I’d like everyone to please stop attacking @timnitGebru and everyone who has been critical of my posts… Farewell everyone.”
Maybe Yann LeCun was wrong, the way she conveyed the message disregarding any knowledge he might have on this was definitely not civil.
She could have chosen a different way to address the complaints here, not by getting overtly confrontational. Maybe the review group got it wrong, maybe they were right. But that is not how you handle disputes.