I don't know much about Simon DeDeo, but is there any reason that I should believe that he can make a fair assessment of Google Brain/Facebook/DeepMind?
Not sure why this was downvoted, but maybe it came off a bit rude or critical. I'm not trying to criticize DeDeo, but I did a quick skim of his work and couldn't find that much overlap between what Google Brain does and his research.
So he's saying don't go to industry and waste your talents in intellectually lame projects, instead come do a PhD, it's more stimulating and rewarding. But what happens when the students get their PhD? There's very few academic jobs, they're gonna go back to industry no? And at some point in the write up he also says academia sucks, so I'm confused as to what his suggestion is, it seems to me that either way you are a servant doing "grad student descent", only that in academia you get to claim to have a bigger intellectual dick, but in industry you have more money. It feels like there's no good alternative.
After my PhD I received a very nice industry position where I can do more research than my friends who are professors. I don't have to sit in committees or teach: I'm employed to conduct research. I'm also fairly compensated, versus the slave wages I was paid in graduate school.
I am flying back from ICML in Stockholm, so it's hard to argue my company doesn't support my academic interests.
I like Simon but he's off-base here. And I don't even work for FAANG/Deepmind, where I suspect I'd have considerably more substantial resources. Industry is by no means perfect but it's a great landing spot post-PhD.
Perhaps a better question ... how long is your situation going to last? Academia offers stability to an extent .. say post tenure. Industry does not imho.
I don't believe my situation is rare: this position was considerably easier to attain than a faculty position.
Stability is better and worse. It is better in terms of resources (I don't have to apply for grants, I can get summer students, and not for grad student descent! ) but it certainly has less job security than tenure.
Statistically, most PhDs will work in industry because academia is a pyramid. Do PhDs get higher paid jobs in industry than academia? Unsure.
Entry postdoc salaries in the UK are on par with industry outside London (£30-40k). Over time industry will probably pay more, but professors earn £60k minimum so they're hardly underpaid.
Being a postdoc is something of a sweet spot. You get paid enough (finally) to enjoy life a bit, you have immense working flexiblity and you aren't lumbered (yet) with the bureaucracy of the university system. The only way to advance is usually to take on teaching loads and more admin.
But.. postdocs are almost always contractual, lasting 1-3 years. This has benefits: most people do a couple of postdocs before getting tenure and it allows you to move around the world if you like. Academia is only permanent if you get tenure, and even then it's still dependent on your research output and teaching performance. If you got a job in a grad scheme, you'd have a more stable job after finishing your PhD.
Getting tenure is hard. This is absolutely not the normal route for PhDs, as much as they think it is. There are far fewer permanent positions than there are postdocs, so it's not uncommon for people to move abroad just to get a stable job.
Speaking from experience as a postdoc, we earn less for sure, but from what I can see a postdoc in the US earns about the same. In the rest of Europe things are a bit different. At the extreme end, in Zurich you might earn 100k CHF as a postdoc if you're lucky (but you pay 30EUR for a pizza). Even at really top-tier places like TUM, you're going to be on 35-40k EUR.
A typical professorial scale in the UK (most universities publish this information publicly) ranges from £60-120k.
You have to realise two things. First, that FAANG are crazy outliers. Second, that UK universities are publicly funded and with few exceptions, justifying six figure salaries from the taxpayer is difficult. A lot of professors make good money on the side by doing consulting.
I believe this is different in the US, as a lot of universities (like Stanford) have huge private endowments. I'm not sure how it works at places like Oxford where the university owns half the centre of town.
A senior engineer in London or a tech hub like Cambridge might make six figures, but that's after several years probably. For engineers, there are much higher salaries to be found elsewhere in Europe in places like Munich.
I think it's weird and unfair to say that research is lesser because it has a tangible goal. It's like saying engineers' discoveries matter less than theorists'.
The argument that "basic" or "pure" science is better or more enlightened than "applied" science exists in every field and is as old as science itself.
If you want to build machines that monitor people and sell them more ads faster, go for it. If you want to find problem where you can take a working-class job, model the man or woman who does it, and build a net to put them out of a job without compensation, be my guest.
A major divisive narrative I see in ML/AI is whether philosophical questions as to the ends of the work are worthy of time.
This is starkest in the debate about providing ML based solutions to governments. Some group says "we're not on board philosophically" the others shrug and move on.
I don't think this is much different than any other engineering/science discipline honestly.
So I think DeDeo, as part of academia is trying to say: Academia has knowledge as it's ultimate goal, corporations just have profit and product. So therefore:
"But if you want, at some point in your flourishing career, with your mind and your soul, to join the two-thousand year old parade of intellectual progress, you are not going to do it at Google."
The strongly negative response, I think is in part because it's derisive of corporate work, but also because there seems to be a bias for the idea that "I'm not interested in the philosophy around why I am doing this work"
It may be because I don't use Twitter, but I found that extremely difficult to follow. It's unfortunate this debate didn't land somewhere more conducive to long-form discussion.
Nobel prizes have been won by both commercial and military labs. The idea that they don’t do science is ridiculous. Moreover, just because your conducting empirical research and haven’t explained your results yet with a theory doesn’t mean you’re not doing science.
Scientific results can come about through many avenues. Sometimes it’s just pure curiousity, and sometimes it’s because you’re trying to solve a practical problem and end up doing research that yields breakthroughs.
And each environment has their negatives, it’s not like academia doesn’t have toxic politics, slaving of grad students, publish-or-perish, the race to get grants, etc
I mean he may bring up some legitimate issues but his whole ‘I met a few people and let me extrapolate this to the industry as a whole’ is itself anti scientific.
Data does not seem to support. If we look at NIPS for example this year. Google had more papers accepted than CMU or Stanford or any other educational organization.
21 comments
[ 4.9 ms ] story [ 61.5 ms ] threadAfter my PhD I received a very nice industry position where I can do more research than my friends who are professors. I don't have to sit in committees or teach: I'm employed to conduct research. I'm also fairly compensated, versus the slave wages I was paid in graduate school.
I am flying back from ICML in Stockholm, so it's hard to argue my company doesn't support my academic interests.
I like Simon but he's off-base here. And I don't even work for FAANG/Deepmind, where I suspect I'd have considerably more substantial resources. Industry is by no means perfect but it's a great landing spot post-PhD.
I don't believe my situation is rare: this position was considerably easier to attain than a faculty position.
Stability is better and worse. It is better in terms of resources (I don't have to apply for grants, I can get summer students, and not for grad student descent! ) but it certainly has less job security than tenure.
Entry postdoc salaries in the UK are on par with industry outside London (£30-40k). Over time industry will probably pay more, but professors earn £60k minimum so they're hardly underpaid.
Being a postdoc is something of a sweet spot. You get paid enough (finally) to enjoy life a bit, you have immense working flexiblity and you aren't lumbered (yet) with the bureaucracy of the university system. The only way to advance is usually to take on teaching loads and more admin.
But.. postdocs are almost always contractual, lasting 1-3 years. This has benefits: most people do a couple of postdocs before getting tenure and it allows you to move around the world if you like. Academia is only permanent if you get tenure, and even then it's still dependent on your research output and teaching performance. If you got a job in a grad scheme, you'd have a more stable job after finishing your PhD.
Getting tenure is hard. This is absolutely not the normal route for PhDs, as much as they think it is. There are far fewer permanent positions than there are postdocs, so it's not uncommon for people to move abroad just to get a stable job.
60k is a senior software engineer salary (especially outside of London)
A typical professorial scale in the UK (most universities publish this information publicly) ranges from £60-120k.
You have to realise two things. First, that FAANG are crazy outliers. Second, that UK universities are publicly funded and with few exceptions, justifying six figure salaries from the taxpayer is difficult. A lot of professors make good money on the side by doing consulting.
I believe this is different in the US, as a lot of universities (like Stanford) have huge private endowments. I'm not sure how it works at places like Oxford where the university owns half the centre of town.
A senior engineer in London or a tech hub like Cambridge might make six figures, but that's after several years probably. For engineers, there are much higher salaries to be found elsewhere in Europe in places like Munich.
If you want to build machines that monitor people and sell them more ads faster, go for it. If you want to find problem where you can take a working-class job, model the man or woman who does it, and build a net to put them out of a job without compensation, be my guest.
A major divisive narrative I see in ML/AI is whether philosophical questions as to the ends of the work are worthy of time.
This is starkest in the debate about providing ML based solutions to governments. Some group says "we're not on board philosophically" the others shrug and move on.
I don't think this is much different than any other engineering/science discipline honestly.
So I think DeDeo, as part of academia is trying to say: Academia has knowledge as it's ultimate goal, corporations just have profit and product. So therefore:
"But if you want, at some point in your flourishing career, with your mind and your soul, to join the two-thousand year old parade of intellectual progress, you are not going to do it at Google."
The strongly negative response, I think is in part because it's derisive of corporate work, but also because there seems to be a bias for the idea that "I'm not interested in the philosophy around why I am doing this work"
Scientific results can come about through many avenues. Sometimes it’s just pure curiousity, and sometimes it’s because you’re trying to solve a practical problem and end up doing research that yields breakthroughs.
And each environment has their negatives, it’s not like academia doesn’t have toxic politics, slaving of grad students, publish-or-perish, the race to get grants, etc
I mean he may bring up some legitimate issues but his whole ‘I met a few people and let me extrapolate this to the industry as a whole’ is itself anti scientific.
https://medium.com/machine-learning-in-practice/nips-accepte...