The Rise and Fall of Industrial Research Labs(cacm.acm.org)
cacm.acm.org
The Rise and Fall of Industrial Research Labs
http://cacm.acm.org/magazines/2015/1/181626-the-rise-and-fall-of-industrial-research-labs/fulltext
6 comments
I agree that MSR is probably the best CS industrial lab out there right now, but I think Moshe's point is that even the rest of the MSR is probably a bit wary right now given what (and how it) happened to MSR SV. I don't think HP, IBM, and Bell Labs (with a few specific and localized exceptions) are doing basic research as they were doing a few years ago (or in Bell Lab's case, a few decades ago). Maybe you are right about Google, but generally speaking, there are very few papers from Google Research (relative to their strength).
Agree. It matters because nothing undermines scientific integrity and productivity like financial insecurity.
There's also many other MSR labs in many other places: China, Cambridge, Israel, Egypt, Munich, others. Redmond is the just the biggest.
I believe that the SV cut was the right choice vs. "spreading the pain around", but we'll see how it shapes up in the end.
I believe that the SV cut was the right choice vs. "spreading the pain around", but we'll see how it shapes up in the end.
And not all RnD labs are CS the same retreat from RnD can be seen in a branches of STEM unfortunately.
I think it's more that the low hanging fruit has been plucked than that there is a 'fall' of industrial research labs. Just like with technology in the early 20th century there was a golden age where advances were relatively quick and dramatic because the fields were brand new. Once the field matures it takes more effort to get up to speed on what has already been learned and evolutionary progress rather than revolutionary takes over.
And so far we seem to be doing pretty good with that level of progress. Any faster than this and I'm not sure we'd be able to properly integrate technology into our culture and society before it had become obsolete already.
Just look at for instance the succession of audio recording and distribution methods to get an idea of that: the gramophone record lasted for many decades, CDs succeeded them and lasted for a couple of decades, digital formats are dying out about as fast as they are being created (with the exception of mp3).
At some point you're going to have problems of interoperability simply because of the speed of progress (we're seeing something quite close to that on the browser front right now).
And so far we seem to be doing pretty good with that level of progress. Any faster than this and I'm not sure we'd be able to properly integrate technology into our culture and society before it had become obsolete already.
Just look at for instance the succession of audio recording and distribution methods to get an idea of that: the gramophone record lasted for many decades, CDs succeeded them and lasted for a couple of decades, digital formats are dying out about as fast as they are being created (with the exception of mp3).
At some point you're going to have problems of interoperability simply because of the speed of progress (we're seeing something quite close to that on the browser front right now).
To explain this acceleration of technology, let's remind the exponential growth of population: There are more people on Earth than dead population since the birth of times
So there's as much potential for science advance in 1 lifespan as since the wake of Humanity.
So there's as much potential for science advance in 1 lifespan as since the wake of Humanity.
>To explain this acceleration of technology, let's remind the exponential growth of population: There are more people on Earth than dead population since the birth of times
Actually there have been around 100 billion people born over this history of human kind. Given a bit over 7 billion people are still alive that suggests most people ever born are dead.
As an aside I have heard this suggested as an argument that the probability of death is not one. If only ~93% of all the people born have ever died then you can't be certain that everyone born will die.
Actually there have been around 100 billion people born over this history of human kind. Given a bit over 7 billion people are still alive that suggests most people ever born are dead.
As an aside I have heard this suggested as an argument that the probability of death is not one. If only ~93% of all the people born have ever died then you can't be certain that everyone born will die.
I beg to differ, based on experience. The low-hanging fruit is gone only for the paths mainstream research take. There were some great discoveries recently but since they don't resonate with the "in" crowd it gets mostly ignored.
There is a lot of low-hanging fruit in genetic engineering and robotics!
Re: robotics- you mean low-hanging fruit research-wise or application-wise?
I'm not closely following latest research in this area but I get the feeling that most of the focus is geared towards autonomous cooperation and nano-scale. What are other directions?
I'm not closely following latest research in this area but I get the feeling that most of the focus is geared towards autonomous cooperation and nano-scale. What are other directions?
Application-wise :) Mostly I'm thinking that with cheap 3D-scanning sensors and GPUs, it's finally the beginning of the era for domestic and consumer robots to be viable.
I'm usually on the side of pointing out the squishyness in some numbers, like US productivity and GDP. But we seem not to be in a crisis or even decline in corporate research. Microsoft and Google and, to a limited extent Facebook have taken over from IBM and Xerox and Bell Labs.
The Bell Labs business model was highly artificial, and that artificiality was used as an argument not to break up AT&T. Current structures are more sustainable.
Some ways of structuring corporate research are new: At Google, and at startups funded by Founders Fund, you will find R&D that's been selected for impact.
The Bell Labs business model was highly artificial, and that artificiality was used as an argument not to break up AT&T. Current structures are more sustainable.
Some ways of structuring corporate research are new: At Google, and at startups funded by Founders Fund, you will find R&D that's been selected for impact.
Bell Labs got us microwave communications, lasers, transistors, Unix, C, awk, Plan 9, fiber optics, and a bunch of other things.
The new "sustainable" structures, frankly, aren't even a substitute.
The new "sustainable" structures, frankly, aren't even a substitute.
http://cacm.acm.org/magazines/2015/1/181626-the-rise-and-fal...
For those of us who don't want to have to read the mobile version.
For those of us who don't want to have to read the mobile version.
Why fritter away all that money on fundamental scientific research when you can spend 2.5 billion for fucking Minecraft?
Being in technology, you have a front-row seat to the concept of elective decline, or the idea that civilization's decline is often a collective but intentional choice, rather than the result of external stress. It seems sad, to a historian, that a great empire would "collapse" (an affair that usually happen imperceptibly over hundreds of years) but the reality is that it's the unintended product of millions of individuals making self-interested and possibly self-beneficial decisions... that end up not keeping up or advancing the civilization. At some point, they just decide that it's not worth it to be civilized anymore, and investment ceases.
Since the 1970s, we've seen elective decline in the U.S., in science, and in technology. Abstractly people want scientific progress, but no one wants to pay for it, and people within the masses would rather be guided by their resentment of academics than fight back when academic or research jobs get cut. When state legislatures cut funding for public universities, the hoi polloi don't care because their resentment for professors is stronger than their sense of a need to keep up the society.
One might hope for Silicon Valley to be better, and look to it for leadership, and it may have kept its integrity for longer, but the current "M&A has replaced R&D" era is just fucking disgusting. It's easy to focus the hatred on a few unlikeable celebrity founders (and I've done my share of that) but the truth is that the problem is really deep and probably unalterable. We have a front-row seat, if we work in science and technology in the U.S., for elective decline-- why it happens, the individual actors who push it forward (not wanting decline, but valuing self-interest more), and the often one-way erosion of trust that tends to make it irreversible-- but we have no power to change it. And, just as one might read about a civilization that collapsed 3,000 years ago and think, "They would have been fine if they just <X>", we can easily come up with solutions that will work but never see implementation, because (as seems to be a constant of human organizations) the wrong people are in charge.
Since the 1970s, we've seen elective decline in the U.S., in science, and in technology. Abstractly people want scientific progress, but no one wants to pay for it, and people within the masses would rather be guided by their resentment of academics than fight back when academic or research jobs get cut. When state legislatures cut funding for public universities, the hoi polloi don't care because their resentment for professors is stronger than their sense of a need to keep up the society.
One might hope for Silicon Valley to be better, and look to it for leadership, and it may have kept its integrity for longer, but the current "M&A has replaced R&D" era is just fucking disgusting. It's easy to focus the hatred on a few unlikeable celebrity founders (and I've done my share of that) but the truth is that the problem is really deep and probably unalterable. We have a front-row seat, if we work in science and technology in the U.S., for elective decline-- why it happens, the individual actors who push it forward (not wanting decline, but valuing self-interest more), and the often one-way erosion of trust that tends to make it irreversible-- but we have no power to change it. And, just as one might read about a civilization that collapsed 3,000 years ago and think, "They would have been fine if they just <X>", we can easily come up with solutions that will work but never see implementation, because (as seems to be a constant of human organizations) the wrong people are in charge.
What do you mean by a decline in scientific progress? The rate at which science is done? Per capita? Do you perceive some kind of long-term trend? Spending per capita on R&D by the U.S. was moving upwards last I checked, maybe that was just in the Bush administration, but I'm missing the trendline here. And what sort of oh-so-important research do you think we're missing out on that could be better parallelized with more funding?
What does "civilization" have to do with some trend in U.S. science funding, that probably doesn't exist? If you think the U.S. is in technological decline... how come oil production has drastically increased? Why are cars so much safer now? If people don't want to spend money on science and technology, why do we spend more on education than any other nation? Why should we infer anything from state legislatures' funding of universities, from which the best and brightest move out of state, which make more sense to fund nationally?
I'm glad you've heard of words like "civilization," "civilized," "hoi polloi," and "fucking." Next time, try saying reality-based things with them instead of antiplatitudes pulled from thin-air.
What does "civilization" have to do with some trend in U.S. science funding, that probably doesn't exist? If you think the U.S. is in technological decline... how come oil production has drastically increased? Why are cars so much safer now? If people don't want to spend money on science and technology, why do we spend more on education than any other nation? Why should we infer anything from state legislatures' funding of universities, from which the best and brightest move out of state, which make more sense to fund nationally?
I'm glad you've heard of words like "civilization," "civilized," "hoi polloi," and "fucking." Next time, try saying reality-based things with them instead of antiplatitudes pulled from thin-air.
> If you think the U.S. is in technological decline... how come oil production has drastically increased?
I read a fair amount about energy issues, but mainly for my own understanding and less so to be able to readily explain it, so forgive me if I get something wrong and for the lack of citations, but I believe it's basically that the increase in oil production is pretty much completely due to unconventional sources like shale oil (and fracking for natural gas) which have a much lower Energy Return on Energy Invested (EROEI), which means they're only profitable when the price of oil is very high.
What expecting these unconventional sources to save the day fails to account for is that they are still very finite, and more importantly, that when the price of oil gets high enough, it hurts the economy, which then decreases demand for oil, which causes the price to drop, which makes it unprofitable to operate and/or develop new unconventional wells until the price rises again due to low supply. This cycle is not really compatible with a healthy economy. We are seeing the beginning of this as we speak, with oil companies laying off people because shale oil isn't profitable at current prices.
Also, the technology for shale oil and fracking has been around for a long time, so it's not exactly technological innovation, just a new source that became profitable to exploit as easily accessible oil becomes scarcer and prices are high.
> And what sort of oh-so-important research do you think we're missing out on that could be better parallelized with more funding?
Maybe space travel?
Or viable renewable energy that could actually sustain civilization?
http://energyskeptic.com/2014/science-no-single-or-combinati... http://energyskeptic.com/2014/why-fusion-will-never-work-and...
I read a fair amount about energy issues, but mainly for my own understanding and less so to be able to readily explain it, so forgive me if I get something wrong and for the lack of citations, but I believe it's basically that the increase in oil production is pretty much completely due to unconventional sources like shale oil (and fracking for natural gas) which have a much lower Energy Return on Energy Invested (EROEI), which means they're only profitable when the price of oil is very high.
What expecting these unconventional sources to save the day fails to account for is that they are still very finite, and more importantly, that when the price of oil gets high enough, it hurts the economy, which then decreases demand for oil, which causes the price to drop, which makes it unprofitable to operate and/or develop new unconventional wells until the price rises again due to low supply. This cycle is not really compatible with a healthy economy. We are seeing the beginning of this as we speak, with oil companies laying off people because shale oil isn't profitable at current prices.
Also, the technology for shale oil and fracking has been around for a long time, so it's not exactly technological innovation, just a new source that became profitable to exploit as easily accessible oil becomes scarcer and prices are high.
> And what sort of oh-so-important research do you think we're missing out on that could be better parallelized with more funding?
Maybe space travel?
Or viable renewable energy that could actually sustain civilization?
http://energyskeptic.com/2014/science-no-single-or-combinati... http://energyskeptic.com/2014/why-fusion-will-never-work-and...
That cycle is how capitalism works, for any good which requires a large capital investment for financial returns that drop whenever there's excessive competition in the market. Historically, we've seen it with textiles, railroads, steamboats, automobiles, office parks, farm crops, telecoms, and airlines. If we actually got space travel and viable renewable energy, we'd almost certainly see it with them as well.
No, it's not pleasant for the people who are laid off or businesses that fail. But basically the entire financial industry has evolved to smooth out the peaks and valleys of this cycle, providing the capital to get the boom started and then recycling the carcasses of overcapacity into productive uses when the boom ends. It's business as usual for a capitalist economy.
No, it's not pleasant for the people who are laid off or businesses that fail. But basically the entire financial industry has evolved to smooth out the peaks and valleys of this cycle, providing the capital to get the boom started and then recycling the carcasses of overcapacity into productive uses when the boom ends. It's business as usual for a capitalist economy.
Yes, its pretty sad to see the demise of industrial labs doing more fundamental research, but this might be related to the demise of the big corporations in general. The HN community understandably hates the bigger companies for being monopolies and stifling innovation and whatnot, but there are two sides to the story. A large corporation, with bigger revenues, can think longer term, and invest in basic research so as to promote efficiency and/or develop technologies to give it a competetive edge. I believe Bell Labs was invented for this reason: the exact story escapes me, but they asked a scientist to help them out with one problem, and then kinda realised it might not be a bad idea to have a bunch of these scientists around, just doing their research, to perhaps be occasionally available to help the engineers with the problems they face.
I do see a lot more cooperation between the Industry and University though. And as we enter an era of smaller corporations, the burden of industrial research might shift to consulting/academic sponsorship programs instead.
I do see a lot more cooperation between the Industry and University though. And as we enter an era of smaller corporations, the burden of industrial research might shift to consulting/academic sponsorship programs instead.
I work for a megacorp. We spend a great deal of cash on self-funded projects to move the state of the art forward in our division's particular niche (avionics). We released a product ~6 years ago that fundamentally changed the computing architecture for modern airplanes (in this case, 787). I work with extraordinarily brilliant people who are given many opportunities to have their cutting edge ideas heard, funded, and executed. While we don't do a lot of basic scientific research, we do plenty of technological research on the absolute cutting edge of our discipline. From what I know of our other industrial divisions, they similarly do cutting edge R&D in their respective fields. I'd hardly say that my megacorp is stifling innovation (quite the opposite, actually). We do have tiny cubes, though :) (but hey - I agree that the money is better spent on R&D, so I have no complaints). I know that not all megacorps and their jobs are quite so awesome, but I can vouch for the fact that some are. As a side benefit, the airframer duopoly prevents distracting patent wars among subcontractors very effectively, leaving us with more resources to work on fun problems.
And why do normal, intelligent people resent academics? Articles like this https://news.ycombinator.com/item?id=8796779
Yes, the Silicon Valley branch has shut down, but MSR's main office in Washington is still open and doing great research. It's probably the best industrial lab that's still running.
I also have interned at PARC (no longer "Xerox PARC", but still kicking). I was given remarkable freedom both by my direct supervisor and the division manager.
Down the road, HP labs was still going (though a few of their staff had migrated to PARC). Likewise, IBM and Bell Labs still are going.
And Google seems to be really spinning up their research. I won't comment on specifics in a public forum, but I've noticed signals that they're taking basic research seriously. (They used to be very short term, product oriented)