Hacker News Reader

Comment by yodon | Hacker News Reader

By way of comparison, Microsoft is the single largest funder of academic CS research in the world, significantly larger than the NSF or US government, and unlike competing companies like Google and Apple they encourage academic researchers to publish papers on the work they fund to the point where if you don’t publish your work you don’t get more funding from them. Microsoft also doesn’t file patents on the academic research they fund.

But we all know Microsoft is the one that’s evil, because they bought GitHub unlike Google which is merely trying to stifle all use of and research in the most important arena in CS.

View on HN

Why do you think Google doesn't encourage publishing papers? Google has a very big amount of papers they publish ... as an example, https://ai.googleblog.com/2017/12/google-at-nips-2017.html

Nearly all the authors in that list are google employees, not independent academic researchers.

That blog post is about Google’s employees publishing papers on work google owns, the same sort of work that google is attempting to secure patents on. This is very different from supporting independent academic research.

Supporting independent academic research like https://ai.googleblog.com/2018/03/google-faculty-research-aw...?

Everything I’ve heard from academics is that the terms that come with funding from Microsoft are much better and more flexible for the academic than the terms that come with funding from Google, and there is certainly a great deal more funding available to academics from Microsoft than Google.

That said, given that my point was about the absurdity in difference in responses to Microsoft’s actions and Google’s actions, I’m particularly amused(?) that here we are, in a page about Google applying for patents that would cover a tremendous amount of essential CS research, and the primary thrust of your points here seems to be “Dude! Google’s Awesome! They’re so supportive of research!” which is pretty much exactly the response I was trying to highlight. Don’t be evil was a wonderful way to start their branding. It doesn’t matter that’s not the company they are anymore, it’s still how people think of them.

You are trying to frame people's responses into a "Google good, Microsoft bad" response, and I do not think that is fair to them at all. They are disagreeing with your idea that Google is non-existent on the academic research front, and Microsoft is doing way more than them. And frankly, as someone in academia, this does not pass my smell test either.

I’m baffled where you get the idea that I think google is non-existent on the academic front. Yes, Apple is not-existent, but that’s certainly not something I believe or would say about google, because it’s not the case. I did say Microsoft is the #1 funder of academic research in the world, which I believe is still true and which no one seems to be challenging [edit: it looks like there is some discussion of whether Microsoft’s funding is bigger or smaller than the government’s contribution, but all within a context of Microsoft certainly being the biggest NGO funding source and possibly being the biggest overall]

Again, I think it’s fascinating how in commenting on a post about google trying to file patents that would completely control the most important new field in CS, someone who correctly points out that Google is not #1 in funding academic researchers is branded as calling Google “non-existent on the academic front”.

This is precisely the sort of unproductive back and forth argument that the HN software is designed to discourage and punish, so I’m going to try to back off here and say hopefully we can agree to disagree, particularly since it doesn’t sound to me like we actually disagree.

I admit that I exaggerated with my wording that you think Google is non-existent on the academic front. It's unfortunate that your response focuses almost entirely on this one aspect, and not the message my post attempted to convey.

> someone who correctly points out that Google is not #1 in funding academic researchers

This type of wording is unnecessary. You are declaring that you are correct, while at the same time not providing supporting evidence when requested.

> This is precisely the sort of unproductive back and forth argument that the HN software is designed to discourage and punish

Welp. I am sorry this had such a negative effect on you. We can certainly agree to disagree. Cheers.

Depends on what you consider "bad" versus "good".

But Microsoft getting patents to use for royalties or as a weapon to me is not as good as the Google approach.

Google does not use patents to get royalties but instead gives away tons of IP to help everyone or the entire industry move forward.

Since we asked you to stop using HN to prosecute agendas about Google and you've posted dozens of comments since then doing exactly what we asked you not to do, I've banned this account.

Obviously there's nothing wrong with saying nice or critical things about $Bigco on Hacker News, be it Google or anyone else. But when it becomes both predictable and proliferating, it stops serving intellectual curiosity and that violates the purpose of this site. You crossed that line a long time ago, unfortunately.

https://news.ycombinator.com/newsguidelines.html

I'm not saying "Google's Awesome!". My point is that the view you presented is far more biased than what I get from my colleagues and friends that stayed in academia. People running around like headless chickens in the "MS buys GitHub" thread was unfounded, and it is unfounded here.

If you look at Google's Faculty Research Award page (https://ai.google/research/outreach/faculty-research-awards/), the awards are described as "unrestricted gifts." I'm not sure how it gets more flexible than that.

Disclaimer: I work for Google.

Yodon, The difference is Google gets the patents so nobody else can and does NOT use as a weapon or for royalities.

Microsoft just takes a different approach. They use the patents for royalities and how they make over a billion off of Android phones a year.

Google has tons of patents and even buys them up like what they did with Moto. Yet they do not charge royalities for using. They instead use for cross license agreements.

Just a very different culture from Microsoft.

But one of my favorite examples of Google and patents is VP8.

We had Mpeg-LA extorting license fees out of us for Mpeg2. Google said enough and created VP8 and gave away for free for anyone to use.

But then in addition also offered patent infringement protection for anyone using VP8. That is just not an approach you would see from Microsoft.

"It doesn’t matter that’s not the company they are anymore "

What has changed? I would say they are giving back more now than ever before. They use to give their "secrets" in papers like Map/Reduce, GFS, and so many others.

But now they actually give the software away with their "secrets". Perfect example is Kubernettes. They build Borg and learned from it and then with everything they learned now give away K8s.

Which is now used by Microsoft, Amazon and pretty much the entire industry. They do not charge a cent and help their competitors better compete against them.

They did it because it helps the entire industry move forward. That is just not the culture at Microsoft.

Google continues to give away just tons and tons of IP.

But you do have me curious. How do you think they have changed in terms of the technology community?

BTW, another example is giving away SPDY which became http2 which helped everyone including competitors.

Or all the money they spend finding all the major security flaws including Shellshock Meltdown, Cloudbleed, HeartBleed, Spectre and many others. Cloudbleed was with a competitor and helped them solve the issue.

Google finds many Microsoft vulnerabilities and tells them about them helping them make their products more secure.

I was going to link to exactly the same thing. Only considering these 2017 faculty research awards, that is over $22 million awarded to academic research. Not to mention Google frequently uses the model where professors work at google for a summer / semester / year on a funded project, but are not permanent employees.

Even if it was, publishing paper is still contributing to academic research. How is contributing yourself somehow worse than just putting your stamp on other people's work?

What is your source that Microsoft funds more CS research than the US Government? Let's ignore any DoE, DARPA, or NIH sources of funding, and just focus on the NSF. The computer science portion of their budget is almost one billion dollars.

> What is your source that Microsoft funds more CS research than the US Government?

Microsoft reported spending $14 billion on R&D in 2017. That year, NSF's CISE budget was somewhere around $950 million.

The comparison is apples and oranges, because NSF is funding public research and a lot of basic research, whereas one must assume the vast majority of that $14B at Microsoft is spent on the D portion of R&D.

Still, though, MSR is one of the more prolific research institutions in CS. If they were a university, they'd probably be top-ranked.

Still, though, MSR is one of the more prolific research institutions in CS. If they were a university, they'd probably be top-ranked.

I definitely agree. They have some top notch researchers, certainly.

I also agree that the budget comparison is apples and oranges, so I'm not quite sure why you brought it up. The OP made a grandiose claim about Microsoft's academic research funding, and made it quite clear with his discussion of Google that corporate research does not count. So I want a source for the amount of money Microsoft spends on pure academic research without the corporate ties.

Personally, I think if Microsoft spent anywhere close to a billion dollars on pure academic research, the funding situation in CS academia would be completely different. Microsoft grants would be pursued just as hard as NSF grants, because they would be such a major funding source. But that is not the case. Corporate grants are afterthoughts compared to the NSF for academics. Additionally, if Microsoft were plunking such a large amount of money into academic research, I assume they would at the very least publish that figure in a PR piece to cash in on some public good will.

> I also agree that the budget comparison is apples and oranges, so I'm not quite sure why you brought it up.

It's annoying that MSR's numbers aren't broken out from the R&D total. All we can do is speculate what portion of that goes to MSR et al, really.

But it's not surprising. There are all sorts of incentives for not breaking out those numbers. Still, the numbers are helpful. $14B is a lot of money. Even a small slice of that makes MSR one of the most prominent funders of CS research.

> Personally, I think if Microsoft spent anywhere close to a billion dollars on pure academic research... Microsoft grants would be pursued just as hard as NSF grants, because they would be such a major funding source.

I disagree with your perspective throughout this paragraph.

Funding grants for university faculty is NOT the only valid model for funding pure research. Therefore, we could very easily expect MSR to contribute a significant fraction of total US CS research funding without decreasing the competitiveness of the grant landscape for professors at universities.

To me, the distinction between "basic research" and "everything else" seems more important than the particular funding model used. So the criticism that Google has a bad track record with long-running research agendas is highly relevant, whereas 'how much money goes to academics in universities' is far less relevant.

> Additionally, if Microsoft were plunking such a large amount of money into academic research, I assume they would at the very least publish that figure in a PR piece to cash in on some public good will.

But of course, they already do that: "$14 Billion!"

> $14B is a lot of money. Even a small slice of that makes MSR one of the most prominent funders of CS research.

You are moving the goalposts. I am not saying Microsoft is not a prominent funder of CS research. I am challenging the notion they spend more than the NSF (or for an even stronger claim, the entire US Gov) on non-corporate CS research.

> But it's not surprising. There are all sorts of incentives for not breaking out those numbers. Still, the numbers are helpful. $14B is a lot of money. Even a small slice of that makes MSR one of the most prominent funders of CS research.

No, those numbers are not helpful. We have absolutely no idea what percent of that goes to academic CS research. It could be 10%, that would be incredible. It could be 1%, that is still great. It could be 0.01%, that isn't very notable.

> Funding grants for university faculty is NOT the only valid model for funding pure research. Therefore, we could very easily expect MSR to contribute a significant fraction of total US CS research funding without decreasing the competitiveness of the grant landscape for professors at universities. To me, the distinction between "basic research" and "everything else" seems more important than the particular funding model used. So the criticism that Google has a bad track record with long-running research agendas is highly relevant, whereas 'how much money goes to academics in universities' is far less relevant.

You are again moving the goalposts. The OP specifically said Microsoft was the largest funder of academic CS research.

> No, those numbers are not helpful.

I think those numbers are useful. Microsoft spends $14B on CS R&D. We can go back and forth about the R and the D, but that's still a freaking huge number.

> The OP specifically said Microsoft was the largest funder of academic CS research.

Needless to say, I'm not OP, and I never made any of the claims OP made.

The only purpose of my original post was to try and share what numbers we do have on this question.

And I stand by reply calling BS on the distinction between research based upon whether money is flowing into professors' research groups. Some of the world's greatest patent lawyers are retained by universities, after all.

> unlike competing companies like Google and Apple they encourage academic researchers to publish papers on the work they fund

First, you cannot compare Google and Apple when it comes to academic research. Apple barely does any, Google is usually the largest source of papers in all major ML conferences (when it's not FB): NIPS, ICLR, ICML, etc.

I think you are confusing academic research and corporate research. Google has a huge stable of corporate researchers who publish papers in academic journals. Microsoft funds actual academic research.

And yes, my inclusion of Apple in that list was fully aware Apple doesn’t do significant funding of academic research. That’s the point.

To be fair the boundary between academic and industry research is quite gray now - lots of collaboration, many PhDs doing internships, etc. Still, it's true MS is not given enough credit.

So just because Google hires researcher and gives them resources, their publication is somehow worth less than Microsoft paying outside researchers?

Why is one somehow superior than the other? Both are publishing papers and pushing academia forward, why does it matter where the researcher are doing their work?

Frankly, when I read a machine learning paper without looking at the list of authors, it is impossible to tell whether it is written by "corporate researchers" or "actual academic researchers". I'm not sure the distinction you're making is really helpful. As long as the research is published openly, peer-reviewed, discussed in conferences, whether the researchers work for the company or for a university does not make much of a difference.

Most of these "corporate researchers" were top of their field in academia before joining Google, Facebook, Microsoft. Furthermore, Google also funds academic research [0]... I'm not even sure where you're getting all your information at this point.

[0] https://ai.google/research/outreach/faculty-research-awards/...

Microsoft is also one of the largest patent royalty extortionists in the industry.

Can you imagine if Microsoft had these patents? Every single internet-connected device with "AI capabilities" would be paying Microsoft a royalty within 3 years after the company would obtain the patents.

As a reminder, Microsoft made a few billion dollars from Android for patents.

https://www.forbes.com/sites/ewanspence/2015/11/01/microsoft...

Microsoft shakes down tens of companies with bogus patents, Google does not. That is one comparison you are forgetting.

Researchers at MSR can't publish everything they work on, you know. In a sense, MSR is also a patent mill. Sure they fund research and a good deal of it is published, but it's not simply out of the goodness of their hearts.

The comment was about academic researchers being encouraged to publish, not MSR researchers. People who work for MSR are Microsoft employees and therefore are corporate researchers not academic researchers.

Google has just given back so much more than Microsoft. Great example is Google dominates with Android and Chrome the client and server and creates SPDY and gives it away. Heck it was changed and they even switch from SPDY to http2.

Or a big one is MpegLA was commiting basically extortion and Google created VP8 and gave away for free. But then did a crazy thing of even giving people that use for free protection from patent infringement. Who does that?

That is why we do not pay what we had to for mpeg2 licenses. Just not aware of similar from MS?

Or a big one is Google finding Shellshock, Spectre, Cloudbleed, Heartbleed, Meltdown and others. MS has so much to lose and does not find anywhere near what Google does to help the entire industry. Plus Google even shares mitigations.

Instead Google finds MS vulnerabilities and gives them 90 days and MS b*thches about it instead of thanking.

But then all the papers of things that are now just defacto standards. Like GFS, Map/reduce, Bigtable, Borg with K8s and just so many more.

Some are ways we now do things.

Then all the source code they give away.

The big difference is Microsoft uses patents to get royalities out of people and use as a weapon in slowing down the industry versus helping push the industry forward.

Google does not play that game. They have tons of patents and purchased patents for cross license but they do not get paid anything in royalities. Instead they give away IP for everyone to use.

That is why Microsoft gets paid more for every Android handset through royalaties from patents than Google does.

Just a different approach between the two companies.

Reply on news.ycombinator.com