OpenAI offers $10M pay packages to poach Google researchers | Hacker News Reader

OpenAI offers $10M pay packages to poach Google researchers

theinformation.com

59 pointschoppaface49 comments

Hardwalled articles (that is, paywalls without workarounds) aren't allowed on HN - see https://news.ycombinator.com/newsfaq.html and https://news.ycombinator.com/item?id=10178989.

theinformation.com used to open certain articles for HN readers but they stopped responding to our emails years ago

Are paywalls ok?

It's ok to post stories from sites with paywalls that have workarounds.

Here are other non-paywalled articles reporting on the news, but some say The Information was the first to report.

https://nypost.com/2023/11/13/business/openai-reportedly-try...

https://www.businessinsider.com/openai-recruiters-luring-goo...

That said, the headline says the most of it: OpenAI-Google poaching wars similar to Facebook-Google now about 10 years ago, though the pay packages are about double today.

Note that the creator of Kubernetes almost took a poach to Facebook, but Google countered and he stayed to create Kubernetes. So the astute HN reader will look at the article headline and then look beyond, as often the article doesn’t tell the whole story.

Also context: YC has vested interests in OpenAI and HN moderation is part of YC. It is beyond me why HN moderation here decides to target my submission with needless references (one of those links is over 7 years old now) and have that appear at the top of the comments. Instead of hey maybe linking to non-paywalled articles. Which would be in the spirit of making HN a place about conversation and learning. And which would be much more aligned with how YC compensates HN moderation. Given there are many hard walled / paywalled articles posted to HN and do not receive such attention from HN mods as has done here, I hope the HN readers are able to weigh how HN mods target certain people in these forums.

> Given there are many hard walled / paywalled articles posted to HN

I've never seen a hardwalled article on the frontpage, without any workaround whatsoever, in the last month at least. If you can link to a few recent examples this would be more credible.

Search for yourself articles with any sort of wall are posted here all the time and HN moderation does not consistently comment and moreover allows people to post gift links which everybody here likes but in fact violate TOS.

https://hn.algolia.com/?dateRange=all&page=0&prefix=false&qu...

https://hn.algolia.com/?dateRange=all&page=0&prefix=true&que...

Also this 7yo flamewar: https://news.ycombinator.com/item?id=12683017

Nobody's targeting you. You posted a hardwalled article, which is against HN's rules; that's all.

sort by date, several articles from this site posted in the past few weeks with zero HN moderation commenting: https://hn.algolia.com/?dateRange=all&page=0&prefix=false&qu...

While HN moderation might argue X Y or Z, the evidence confirms targeting, the comments here by HN moderation stigma both submission and discussion, and the substance of the article is primarily the headline anyways.

HN moderation and YC leadership have a well-established desire to antagonize, see e.g. https://techbrosf.weebly.com/

> target my submission with needless references (one of those links is over 7 years old now) and have that appear at the top of the comments.

This is... dramatic.

Moderation comments appearing at the top of a thread is pretty common and makes perfect sense to me. One of the 'references' you're talking about is the site FAQ, and the other is a thread explaining the policy change re: paywalled articles; I'm not sure why its age is relevant.

@dang, why are Medium paywall articles still allowed on HN? They recently went hard paywall.

If it's hard paywalled, it's not allowed.

Detecting these things is another matter. We don't know how to write code to do that, and it's impossible to keep track of manually. Even for a given site it's different for different regions, different users, etc.

> Join us now to lock in a stock package at the current valuation of $27 billion and benefit from the impending increase.

Oh, so not $10M.

Weird coincidence that it’s the exact same amount that Research In Motion paid to poach top engineers from Google / Microsoft back in the 2000’s to help grow their Blackberry network capacity (assuming the Blackberry movie is accurate)

Doesn’t seem unreasonable to me. Top athletes can make hundreds of millions. For a company looking to lock down talent in a fast moving market, it would be strange to not be offering large compensation packages.

I find it hard to believe it’s the case for an average researcher, we know their pay package for most L6 engineers is a bit under 1m. But maybe a “fellow” or distinguished engineer?

Yes, it probably applies to the most qualified hires, so probably folks who are already around $1M total comp where they are.

But the most important point is, you don't get that money now, and maybe never. The most likely scenario where you get that kind of a payday is if they go public and if the pre-IPO pie-in-the-sky valuation holds for a good while.

It might. But it's far more of a gamble than your FAANG stock.

I assume an engineering fellow at Amazon or Google is making well over 1M, so that actually makes sense to need to give them 3x or more for drastically less liquid comp.

What's the fastest way to jump this ladder. Swe at Google, cross train yourself on LLM in 20% time and at home, internal transfer, then get competing offers?

Most of these people presumably have PhDs from top tier universities. So... don't think you'd be able to jump aboard this train without going back to school for a bit.

Most of these people are the best in the world at what they do. Like literally, the best out of tens of thousands of other smart people doing the same thing. If you think that you will become the worlds foremost expert in something if you "go back to school for a bit", then I don't know, maybe you should just go for it.

Publish a 10000 citation paper at NeurIPS.

Be born in 1980. Have reasonably well-off parents. Spend your childhood learning math and computers. Travel a little and learn a bit about the world. Study hard and do your homework. Get in to a top university and work on exactly the right problems. Get a job at google some time between 2003 and 2009. Spend at least a decade solving hard problems there, making good visible progress. Now wait to rake in the sweet sweet offers.

Spend at least a decade solving hard problems

That's the hard part. If you can do that you absolutely deserve all the sweet offers coming your way.

Also, be born with an extremely IQ.

Publish high impact papers or develop a social circle of influential folks and impress them

Money hack: leave open AI, join Google, land openAI, repeat.

Isn't this the common advice for software engineers?

That's not a hack, that's practically an industry standard.

archive.ph?

you can put the url immediately after archive.ph/ and it works

I am fascinated how this outlet (theinformation.com) spreads clickbaity rumors and sells them behind paywall.

They are very legit and well researched articles, probably the most accurate in terms of tech news(semaphore is pretty good as well). However, for niche audience news, subscription is probably the only sustainable way to build a business

this is not my observations. I researched few news stories like that, and it was like "someone told someone" without a chance of verification.

I bet this one is also like that.

Not naming sources is one thing, but which news stories from The Information were false? IMO they have an amazing track record, to the point where half the tech news you read is some journalist's distorted summary of something they read in The Information. There's a reason they can charge $400 per year and people pay.

I didn't say "false", I said rumors which are not possible to verify.

The service they provide is verifying rumors, essentially. Due to the nature of the subject matter it's not possible for them to show you all the data you need to verify things yourself. If you don't want to trust their verification that's your prerogative but IMO rather silly considering their excellent past track record. And that's coming from someone who is extremely critical of most journalism these days.

> IMO rather silly considering their excellent past track record

this is absolute handvawing

Their past reporting on non-public information that I was personally familiar with has been 100% correct. This particular news about $10m packages was just corroborated independently by some well known researchers I follow on Twitter. Based on that and their prior track record of things they reported first that later became public and were true, I trust their reporting on other things I can't immediately verify.

> This particular news about $10m packages was just corroborated independently by some well known researchers I follow on Twitter.

can you post the link?

https://x.com/ericjang11/status/1724231320317120676

> https://x.com/ericjang11/status/1724231320317120676

he was speculating in that tweet obviously.

It's a screenshot of his DMs. Someone he knows is telling him they know it's real. The name is cut off presumably because he doesn't want to out his friend who is disclosing non-public information.

Yes, so we went from "corroborated independently by well known researcher", to researcher responded "psyops" on DM from unknown author, lol.

Seems like you won't accept anything short of Sam Altman himself announcing it on stage. I guess you'll just have to be satisfied not knowing.

yes, that's how professional journalists and outlets work: they announce news if they are verifiable. Tabloids spread rumors instead. There are many rumors to spread.

I closed three banners and a payment popup before being told I couldn’t read the entire article.

It might be easier to use chatGPT to build something that is worth $10M. For example, there is the guy that made a crypto-currency.

I'm curious why would they need to do this. They are ahead of Google (and are arguably less evil), so the best people should naturally want to go to them. Besides, they probably already have all the people they need, now they just need more GPUs to test ideas.

I dunno, I think it's pretty evil to misuse the word "open" like that

Would you rather work for Google developing cutting edge AI models, or for OpenAI?

Reply on news.ycombinator.com