Slashdot Mirror
Industrial Labs that Still Do Fundamental Research
An anonymous reader asks: "I am a graduate student of Mechanical Engineering at a reputed University in the United States. I have had a lot of fun working towards my PhD. I have published papers and done exciting research. I should be finishing up in the next few months or so, but I would like to continue doing the same kind of work that I am doing now. One option would be to take up a post-doctoral research appointment and find myself a faculty position. I am somehow not attracted to this option because of the tenure and grant pressure. My ideal job would be in something like the Bell Labs of yester-years. Do you know of labs that have that kind of environment? National labs are supposed to have such an atmosphere, but my stint in one of them makes me think otherwise. Google does seem to have such an environment but I am not a CS person. Does Slashdot know of labs where basic research in applied engineering is still done in the US, without the pressure of money and immediate results?"
I don't know, of course, but you shouldn't be surprised at all if there are absolutely no privately-owned (like Bell Labs was. Not talking about private universities here) pure research labs in the U.S. anymore.
The U.S. is run almost entirely by bureaucrats, lawyers, and accountants now. Such people have no interest in anything beyond next quarter's profits and their own stock options. Why would they care about something so "unprofitable" as pure, undirected research?
Worse, I think the rest of the world is following suit. But I could be wrong about that, too.
Either way, it's quite depressing. Actually, most of the current trends are quite depressing. I should probably stop thinking about them, and probably would if it weren't so useful to have some idea of what to expect...
Use 'slashdot stuff' in the subject line in any email you send me if you want to get past the spam filter.
... can be an interesting place to work. Very much depends what you would get to work on, though. I guess presure on results out is almost always there in industral labs. But still, an interesting problem to pursue for few years can grant you the illusion you seek.
http://ge.com/research/
I would like to die like my grandfather did - sleeping. And not screaming in terror, like his passengers.
You mention that you would like to work for the Bell-labs of old. What makes you think you need a CS degree?
In my limited experience, research labs for technology companies (like IBM, HP and Sun) employ a very diverse group of people from multiple disciplines. The common trait of these people is that they are interested in researching computers, without necessarily having a CS degree. In some ways having a CS degree might not help if you want to do radically innovative stuff (one never knows). I cannot comment on the likes of Google, Ebay or Amazon, but I am sure they have a lot of smart people working on their computing problems that do not have CS degrees. Consider this, if you work for Amazon and research interface design to guide customer decision making, I would *hope* you don't have a CS degree...
If your engineering degree will give you access to any of the research labs, I don't know. Part of it is luck of the draw - having some skills they want. The other part is pure brain power, e.g. are you smart enough to cope and flexible enough to adapt.
If you want to work at a research lab, be prepared to present yourself as a capable candidate.
Really, a lot of advisors do consulting with their associated industry, or were once in such a research lab you are looking for. If that doesn't pan out, e-mail some other professors in the department whom you know. You'll find someone who knows the scene. Another option is to use CiteSeer or Google Scholar to search for papers in areas that you are interested in, and skim them for any that are published by private company labs, and apply there.
If you first do a year or two or real work in real industry, then go back to academia or fundamental research, you're more likely to have a far better appreciation of the industry and more likely to make valuable contributions.
Engineering is the art of compromise.
The various national laboratories and other FFRDCs vary widely in their environment. I wouldn't necessarily write off all of them based on your experience at one. They have the large benefit of having research in their particular field being a core part of their charter, and government funding to boot.
If you are looking for the kind of place that Xerox used to be, especially as a way to avoid the mindnumbing grind of chasing grants and spending your life in what amounts to temp job, forget it.
First, even at the "golden years" of blue-sky research, the only ones that had a permanent position were people that had already proven themselves by a long grind in the post-doc mill and found to be exceptional. Going from your thesis to a steady research job in a place like that didn't happen even then.
There are places like that today - around here we have NICT and ATR in southern Kyoto, for example. But there too, much of the research is implicitly or explicitly aimed at resulting in something useful, and you are no more free of the grant process than at a university. The people with a permanent position are again few and far between; the head researchers overseeing the groups of post-docs and visiting researchers having some temporary grant.
Really, the difference between university research and research institute or large-company research is in my experience mainly in the need to teach (and the opportunity for a semi-steady income) at a university on one hand; and the greater financial resources for equipment and travel at institutes on the other.
I know of only two ways to get to do free research without the teeth-grinding pain of grant-chasing and temporary job upon temporary job:
* Get a steady part-time job you can live on, and do research in your spare time. Teaching is not a bad option if you're reasonably good at it; you have access to the university, with seminars, labs and people, and teaching your subject forces you to pay attention to areas you perhaps would tend to ignore if left to your own devices.
* Make a fortune, retire and do research as a hobby, perhaps form and finance a small group with a couple of colleagues you like and work well with. Hey, we can all dream, right?
Trust the Computer. The Computer is your friend.
If you're doing applied research, you won't have to pursue the money, but you'll have to produce concrete results, on time, on work that's assigned to you.
I'm surprised you're nearly at your Ph.D. and this has not been made clear to you. You really, really need to be having this conversation with your advisor and other faculty (or senior researchers) within your department. Start with your committee--they know you and your work (hopefully!)
In my own experience as an undergraduate student in psychology at UC Davis, quite a few professors make regular use of actual research papers in place of textbooks. I think it is a great technique since it exposes me to both classic studies and cutting edge research. Furthermore, it allows me to judge the research on its merits. Textbooks to often just cite the results of a research paper, which amounts to a bunch of uncritical fact learning. Another challenge of reading cutting edge research is that a lot of prior knowledge is required to comprehend it.
A Good Troll is better than a Bad Human.
I read this as saying you'd like a great job without pressure. And maybe a pony as well. It may be worth noting that the people at Bell Labs of yesteryear were generally people who would cruise through tenure and get plentiful grant funding consistently. A place with opportunities to do interesting, independant research of your choosing requires a great deal of ability and drive, whether it is academic, private or governmental. If you don't want to work too hard, fine, but don't expect a dream job without fanatastic commitment and drive.
"National labs are supposed to have such an atmosphere, but my stint in one of them makes me think otherwise. " ...labs where basic research in applied engineering is still done in the US, without the pressure of money and immediate results?"
If the national lab environment wasn't for you... the corporate environment may be even worse. As a PhD in chemical engineering working at an R&D lab in one of the biggest 'tech companies' in the US that still does physical sciences reseach... I can say this from first hand experience. And, by the way, we employ a LOT of PhD mechanical engineers (mostly with materials science expertise).
At one of the conferences I've attended, I talked with one of the pioneers in my area of research (organic electronics) that works at TJ Watson Lab. Even he complains at how 'managed' the research is at Watson. Actually - his particular project got shelved. All my friends (other PhDs) at Watson do seem to have this cloud of doubt looming over their head regarding the longevity of their positions.
"
As you know - physical sciences research (of which I suspect you are a part of) is extremely expensive. (~$4000 barely gets me an electronic weighing balance that allows me to weigh out the chemicals that I use, much less do anything with it) Someone's got to pay for this. The return on investment for research has gained huge scrutiny in the past several years since it's typically so bad. Many company's don't have such efforts (e.g. Apple, Dell) and are still successful as they concentrate on industrial design and business execution. They simply BUY this technology from smaller companies (or acquire them). And as far as working for those 'smaller companies'... this is even more stressful since it is really sink or swim.. so the 'pressure of money and immediate results' is even greater.
My best advice is this... on your interviews - ask as many questions as possible to learn about how serious the company is in making the appropriate investments for whatever project they are hiring you for. Talk to your would-be peers and ask them frank questions about the work environment.
Lastly - one of my close collegues at work left a senior scientist position at a national lab to work where we do now and he regrets it deeply. If you are really, truly into research and learning the nature of things, and have low tolerance for corporate bullshit - then stay in academina/national lab. If you can stomach it - as I can - there are definitely perks to working for a big company's reseach lab (e.g. the pockets are deep).
"Years ago there were more far thinking companies like Xerox, HP, SGI and Bell Labs, but they got lazy and were under more pressure from shareholders to focus more on short term profits and less on long term viability of the company."
I would actually say a lot of the R&D a lot of companies did back in the day did not help them as much as it should have. They would invent great things, but some other company would usually profit off of it. R&D is expensive and needs to be well-justified.
Today, computer science has plenty of R&D in industry, but mechanical engineering has to turn to defense simply because of the huge cost in making anything interesting. The technology that goes into modern warfare will trickle down into society in several years, similar to the way NASA worked 30 years ago. It's not an entirely terrible system, because no one but defense is really willing to spend the amount of money and defense is pretty universally agreed on as neccesary.
Generally, a person that finished the first (or sometimes second) year of grad school is on the level they should have been when granted a bachelor. Really !
If this sounds strange consider that people who graduate with Physics major often have never been exposed to Green's function. How can you practice physics in a modern world without knowing this ? This is highschool level.
What should happen is that instead of doing a curve and giving the top 1/3 A's students should be checked (thoroughly) for knowledge of basic concepts in the course and failed if they miss a single one. And the A's should go to those who know the material through and through.
I would actually say a lot of the R&D a lot of companies did back in the day did not help them as much as it should have. They would invent great things, but some other company would usually profit off of it. R&D is expensive and needs to be well-justified.
...... I could go on and on and on.
Your memory is not as long as mine then. HP became huge by investing in R&D. Apple and Adobe arguably became who they are because of investments in fonts and laser printers (not to mention software and industrial design). Yahoo and Google are who they are almost exclusively because of R&D. Before that we can certainly look back to GE, Siemens, Boeing, Corning etc...etc...etc.... All of these companies profited quite handsomely because of R&D, but I suspect you are thinking of companies who at some point in their management cycle started to focus on the short term rather than the long term and it cost 'em.
Today, computer science has plenty of R&D in industry, but mechanical engineering has to turn to defense simply because of the huge cost in making anything interesting.
Oh, please. I can think of a ton of things that do not cost a tremendous amount of money to engineer, yet are big money makers in their individual markets. Think glass and composites for a variety of things from buildings to aircraft to bicycles to skateboards. Think ceramics for many of the same structural applications and more (acoustics and many others). Think automobiles or hell, even bicycles. The last downhill mountain bike race (linked here) I attended had Honda downhill mountain bikes with automatic transmissions. Think applications in home construction. Think about
The technology that goes into modern warfare will trickle down into society in several years, similar to the way NASA worked 30 years ago. o. It's not an entirely terrible system, because no one but defense is really willing to spend the amount of money and defense is pretty universally agreed on as neccesary.
I have no doubt about that, but after working with some folks in defense, I can tell you it is an inefficient system littered with middle managers and other parasites that each need the hard work of others to justify an existence. Furthermore, completely idiosyncratic and political decisions go into many defense related projects that end up on the cutting room floor for reasons completely unrelated to the performance of the defense project. Read about the XM-8 rifle system to understand what I mean. The dollars that go into black projects invest in technologies that are tied up for years, sometimes decades before ever being made available to the general public and often result in environmental and economic consequences that would be better managed in open, competitive environments. All told, I would much rather see those dollars go into education, basic science and open competition for even defense related projects.
Visit Jonesblog and say hello.
OK, a couple of notes here:
1) Teaching undergraduate classes has (with a few exceptions due to notable students) essentially *never* directly contributed to original research. (I am speaking here for physics. In psychology, undergrads contribute to research - as lab rats.) Undergraduates do do their own work, but it's generally independent of classes. I don't think this has changed. In fact, I think the fraction of undergraduates involved in thesis-like projects has probably increased. (Ask a physics major of thirty years ago: how many of your graduating class did independent research? With my class, it was nearly 100%)
The benefit of teaching undergrad classes is that it forces you to communicate, and often will make you look at your subject in a different way.
2) Undergraduate classes are very much advanced from where they were - the topics change with the times. Fifty years ago, quantum mechanics was a course for advanced graduate students only (like quantum field theory or string theory is today) but as an undergrad, I took three semesters of QM.
You should read a couple of books. 1) The World is Flat. 2) The Pentagon's New Map.
You are correct in thinking that the race for the Cold War is over. However, what you need to consider is that we are now in a global market competition for goods and services that will require innovation to stay on top. In fact, it could be argued that the Cold War was in reality an economic war that Communism lost (is still losing) because they cannot maintain the technology and information lead. Their infrastructure simply could not compete.
So, getting back on point: If we focus just on applied engineering, we will end up being the country where work is simply outsourced to because of cheaper labor. This is already happening to a great extent with the European and Japanese automobile companies who are building more of their products here because Americans work for less money than their counterparts in Germany and Japan. So, if you paid attention in history, economics and world history you would find that history has shown that those countries that define and maintain the technological edge will lead economically. Those countries that cease or fail to invest in long term strategies and educational investment wither away or at least fade to some extent behind another group/country that invests more in "brains".
Visit Jonesblog and say hello.
I personally think that schools enforce the idea of cram for a test and subsequently drop the knowledge learned. Everything we learn now adays is extreamly compartmentalized. First test learn everything about Sequence and Series. Secound Test Learn how to do Volumes of Nth order objects. Third and Final Test for the semester learn how to do various Large Matrix operations. Is anything I learn in x part of the semester used in any other part... or in any other class in the future? No. So we get exposed to an idea for 3-6 weeks, try it out and never use or practice it again.
I can assure you that every Physics major has been exposed to the Green's function however it was probably for 2 weeks which they were tested on once, and never used again because the way college is taught now adays. Even though it is horrible that they dont 'know' the green function the point of college is to give you a wide breath of knowledge so that hopefully when you are faced with a situation you pause for a moment and think... hmmm I think there might be a way to solve this problem let me look at a math book and/or google it and arrive at an answer that solves your problem.
Expect if you find a job where there is little connection to revenue and performance, the job will vanish due to failure of the company. As the anonymous reader self says: "..Like bell labs of YESTERYEAR".
I have worked as a R&D project manager for companies with these ivory towers of researches. While I need the algorithm next month, they usually propose to create some two year research project with some unclear goal.
I believe they should deliver or disappear.
You should just embrace the need to deliver, and have a lot of fun doing it.
"Fix it"
Being able to do great research in a stimulating environment is not about technology in the first place, nor is it about the formal type of structure you are conducting that research in. As one post said it below, it is about people. It **is** possible to find a privately owned corporation, even small or medium-sized, that will let you do exactly such a thing, simply because there is a good human contact between you and the management / executives. It's about trust. I am speaking from my own experience: a French ( now ) 800-employee tech company let me set up their research department. From scratch, and on a low budget. It was until now the greatest time in my life, professionally speaking. The morale of this story: think outside of the box. Go look for what you want in usual places taking the usual means - and what you'll find will be predictable. Go look for what you want in unusual places, take unusual means - and you'll find unusual things.
Non sunt multiplicanda entia praeter necessitatem - necnon voluptatem
( Ockam's razor, hack #3254.1 )
Religous speak to God. Insane are spoken to by God. When all shut up, one can finally hear Shostakovich in peace
I think you miss a key part of research... The scope of it. 50 years ago the scope of engineering research was a lot more broad than it was today. In the computer industry, we can go back 30 years and see it. A computer architect 30 years ago likely knew more about electrical signals than todays. However the work on it could be started right after finishing an undergraduate degree. After an undergraduate degree you could undestand how transistors work, understand boolean logic, and put together a few thousand transistors to create a processor. State of the art research would have been in ALU design, datapaths, etc. Advance a few years and the research is focusing more on pipelining processors, parallelization, branch prediction etc. Granted much of this work had been done priorly with supercomputers, but going back even further similar progress had been made.
A little further down the road more work is being done in cache structures and deep pipelining, multiprocessor memory concurrency etc. The amount of knowledge needed is immense. In order to work on any of these features researchers needed a background in electrical engineering (although, they have cut back some of the detailed analog work necessary), they need to understand the workings of boolean logic, take the basic circuits courses, understand computer programming, know how to fully design "simple" processors, know how all the advanced features in the processor work, and then concentrate on a single component, and try to improve upon it. Much of this knowledge isn't obtained until their graduate career. Only then can they start reading papers on their specialization, and later they can hopefully contribute papers to the field.
The low hanging fruit just isn't as available in well established fields. Granted there are fields (even within computer engineering thankfully) where little work has been done, and large gains can be had. However even these fields all require significant background knowledge of all the complex systems involved.
Phil
Since the above are all illegal, especially in scientific quantities and are forbidden from schools and other training facilities, not much happens in terms of industrial science these days. I'm happily reinventing the wheel; how many mechanical engineering grads can build a wheel without going to a store?
>> Defense Labs and National Labs: the political forces are too strong for blue-sky research to happen there.
Definately take the politics out - I once worked in defence research lab, specialising in weapons technology. My pet area is killing groups of people as quickly as possible (outdoor specialist). My team came up with some breakthrough ideas, but the g-men said it was too abstract, too blue sky, too arty-farty.
It pretty much came down to "it can kill lots of people, but unless it can start production in my state next quarter and be killing brown people within the year, it's a no-go.", my favourite excuse (shot down because the office favourite's conventional design had a cool looking model): "Your laser is great, the people are out of the way, but now the oil fields on fire.".
Get politics out of war!
I currently work in academia but with a lot of very strong ties to industrial research labs, big ones ( TJ watson, Sony, Fujitsu, Intel ). The industrial guys still do a lot of very cool research, only some of which is directly targeted at immediateyl relevant problems. However, I think that the type of research that you are looking for really doesn't exist any more, nor am I sure that it ever did. The bell/Xerox/Skunkworks type of facilities of yore are surrounded by a lot of mythology. The truth is that think tanks were always expected to produce something, and frequently did since they hired some extremely smart people. Companies always expect a return on their investment.
For example, TJ watson labs do what IBM considers to be blue sky work i.e. there is not necessarily an immediate commercial benefit from the work. The groups compete for funding internally and they have a project peer review process which is very similar to what you will find in academia. There aren't really any places left where research is not targeted. The point is that the differences between academia and industry are quite small.
Over the last 20 years of so I would suggest that the idea of 'Blue Sky' research has changed a bit. 'Blue sky' as a term can be a little misleading since it suggests someone walking into the room and saying "Here is an unlimited pot of money. Go invent something REALLY cool!". I don't think this has ever happened. A research project is almost always restricted in that you work in a specific field (semiconductors in my case). The chances of me landing a job in say psychology are pretty slim unless I went back and did quite a lot of re-training. Researchers, since research is expensive, have to be experts in their field before you entrust large amounts of money to them and that is true in both academia AND industry. So pretty much everyone is going to require either: A LOT of experience (8-10 years plus) in the field, or a solid postgraduate training in a related field, before they consider you for a position. A PhD is a good way of racking up the experience points in a more compressed fashion than if you go straight into industry. You HAVE to specialise to some degree, unfortunately.
One thing I would say is that Academia offers a lot in terms of the lifestyle. I travel all over the world every year, meet some very nice people, have very good dinners and someone else pays for it. Also don't underestimate the benefits of the flexibility which you get in academia. I don't have to get up early in the morning and get much better holiday time than in industry! Unfortunately the pay sucks :(.
Hmmmmmm
Put politicians into war (right there on the battlefield). That will make for a lot fewer wars.
Censorship is telling a man he can't have a steak just because a baby can't chew it. --Mark Twain
Physics major often have never been exposed to Green's function. ... This is highschool level.
...
"Green's function is a type of function used to solve inhomogeneous differential equations subject to boundary conditions. Technically, a Green's function of a linear operator L acting on distributions over a manifold M, at a point x0, is any solution of (Lf)(x) = (x x0), where is the Dirac delta function." - http://en.wikipedia.org/wiki/Green's_function
I don't know what highschool you went to
A song for you.
Happy people make bad consumers.