Microsoft Accused of Squandering Billions On R&D 580
Julie188 writes "Even as Microsoft celebrates its 10,000th patent, angry shareholders are starting to speak out against what they say is the squandering of billions of dollars on pointless R&D projects. The 10,000th patent covers a technology that allows a device to associate data with objects placed on its surface, and is likely eventually to become part of the Surface table PC. But shareholders are fed up with the $8 billion annually spent. Said one, 'I believe Bill Gates is a charlatan because what he has said, implied, promised to shareholders and stakeholders and all of these visionary things that he mumbles and jumbles about and doesn't make reality of. MS is spending billions of dollars on R&D. Where is the return on investment?' In contrast, Apple had almost the same revenue gains as Microsoft while spending one-tenth as much."
Bill Gates? (Score:5, Informative)
Re:Bill Gates? (Score:5, Informative)
He's still chairman of the board.
Re:A dangerous precedent (Score:4, Informative)
Bell Labs, Xerox PARC, IBM's various research labs, 3M's research and others have all generated wonderful new things from their basic research.
And yet for all the raves these research groups generate, it very seldom turned into successful product launches for the parent company. Xerox is famous for inventing lots of cool technology that became successes for other companies. Bell Labs had a fearsome reputation, but much of its output never ended up in BellCorp products - otherwise we would still be talking about AT&T as a dominant Unix vendor.
3M is a better example, but most of their projects are closer to home - production engineers working on product ideas of their own, rather than basic research.
MS Research may do great things, but few companies are willing to take the schedule and financial risk that goes along with productizing a new technology. Making the jump for R to D is difficult for a company that wants to know a schedule, budget, ship date and ROI within plus/minus 10%.
Re:They aren't investors (Score:5, Informative)
C# and .NET were born out of the COOL project. COOL was a engineering response to Sun's lawsuit over Microsoft's attempts to extend Java in incompatible ways. Microsoft Research was never involved in the development of .NET.
Sometimes R&D isn't R&D (Score:5, Informative)
Depending on how Microsoft classifies it's workforce, this may be a simple labeling issue, Many companies call future development work R&D for tax purposes. I believe you can deduct or amortize part of your R&D budget. So, Windows 8 may very well be "R&D".
a lot of .NET development has been (Score:5, Informative)
A bunch of the .NET languages, runtimes, and compiler features originated in or were developed closely with Microsoft Research, and some parts (like F#) were almost wholly developed there.
Although it's not very much liked by Slashdotters, Songsmith [microsoft.com] has also been relatively successful. Kodu [microsoft.com] is also getting a reasonable amount of press, and helping to solidify XNA's lead in the education-via-games space.
More generally, they develop prototypes of a lot of ideas that get reimplemented by the "product" side of the company. For example, MSR has been experimenting with adding machine-learning and data-mining features to MS desktop products for years, something that the product group is now starting to do with Excel. Those sorts of things are harder to quantify of course--- did the MSR experiments in that area help the product team at all? Would they have done the same anyway? Hard to say, but in general I think the advantages of having an R&D division in your company are undercounted in these "soft gains" ways, which is one reason that once companies downside their R&D divisions, the product groups stop producing as many new things as well.
An ounce of truth, but the wrong argument (Score:5, Informative)
As a [remaining -- for now] Microsoft employee, I can tell you that there is lots of stuff going on here that gets cancelled. Things do not always pan out.
There are probably projects and people that could be cut. MS could probably be more efficient.
Generally, I've seen good technology and near-finished products get killed for political reasons. That work tends not to be completely lost, however. Near-produts tend to have their interesting technologies teased apart, refactored, and re-incorporated into future MS offerings.
However, much as I malign them, I trust the various managers within MSFT to make R&D and strategy decisions over some dipshit that owns 200 shares of MSFT and is irate that he's not seeing '95->'99 era stock price appreciation.
The MSFT stock has been garbage for a long time -- and I am sure I own more of it than the average complainer. Microsoft has always spent money all over the place because real progress takes investment. The company continues to be highly profitable and doesn't appear to need micromanagement from people looking to get rich via stock speculation.
I haven't carefully analyzed the ramifications, but I am at least emotionally drawn towards the idea of MSFT rebuying _all_ of its public stock and telling the market to FOAD.
Last I checked our market cap was down in the $200B range, so I don't think that's a plausible option, given our cash position.
Re:Death march (Score:1, Informative)
The Static Driver Verifier. Okay, so it's given away free with the DDK, but it indirectly helps them since driver quality is now by far the main stability problem Windows has.
It's also not new, as static analysis has existed in various forms for quite a while (lint is a form of static analysis). I'm not saying it isn't a good product, it just isn't something that counts as R&D seeing as the research has already been done so all they needed to do was actually implement a version for their driver model (they also could have done one better and implemented a generic version). My statement still stands, the MS R&D department doesn't really produce anything useful that isn't simply a re-implementation of something someone else has already done. To date the most innovative things I've seen them do anything at all on are MS surface (which isn't really new, although the way it's implemented is certainly one of the cleaner approaches) and that god awful abortion of technology that should have never been Songsmith.
I don't really think MS is better not having R&D, but I feel like what passes for R&D at MS currently falls well short of the mark to the point that that money probably would have been better spent buying a company that actually is doing some real R&D.
4 MSR-initiated products off the top of my head (Score:5, Informative)
- Parallel Extensions to .NET
- Surface
- Photosynth
- WorldWide Telescope
I don't know if Parallel Extensions is worth $8 billion, but it's a huge deal and the cornerstone of the ManyCore/Multicore work MS is doing. It's pretty freaking cool. (And the Mono folks have already implemented it...)
Re:Budget (Score:3, Informative)
A good example of something that has crossed over from research into reality is the Generics in .Net, that started out as a compiler mod by some guys in MS Research.
Re:I Think I See The Problem (Score:3, Informative)
"they're forced, by legal obligation, to work on making stock prices as high as possible."
that is false.
Re:Death march (Score:3, Informative)
the research has already been done so all they needed to do was actually implement a version for their driver model
You are seriously underestimating the complexity of the task. In any case, essentially everything has been thought of before. Your bringing up lint is simply stupid: you are either saying that everything that's been done (by MS and others) in the area of static analysis is a knock off lint, which is simply an ignorant thing to say, or you are saying that the work done by Microsoft is *really* a knock-off from lint, which is false.
Re:Death march (Score:5, Informative)
It's also not new, as static analysis has existed in various forms for quite a while (lint is a form of static analysis).
The work that the SDV is based off of is called SLAM, and it was as much an advance to the field of static analysis as anything people do today is.
Take a look at the publication list [microsoft.com] from the SLAM project. The research that has gone into it has seen publication in POPL twice (along with PLDI one of the two top-tier conferences in PL), CAV three times (also extremely good), and many other venues.
The BLAST project, which is in some sense a successor to SLAM (not at MSR work), has seen quite a bit of additional publications.
You quite clearly don't know what you're talking about; PL is my research area, so I somewhat do.
Microsoft Research is one of only a couple industry research labs that publishes research of similar quality and quantity to a good research university (another is IBM; Google definitely doesn't). I am much less opposed to MS than most people at /., but I will steadfastly defend MSR.
research.microsoft.com (Score:4, Informative)
It may not be popular or known to common users, but Microsoft Research [microsoft.com] is actually fairly well known for its work and yields plenty contributions to scientific publications - so it isn't like they aren't doing anything [microsoft.com]. Here are some [microsoft.com] random [microsoft.com] pages [microsoft.com] from the site.
If anything, it's surprising that more of it doesn't bubble up into consumer products. Maybe it's simply mismanaged or mistrusted by the management?
Re:A dangerous precedent (Score:5, Informative)
Xerox is a good example of this. However, the other two are less good. You mention that 3M doesn't have a particularly large research arm separate from their manufacturing R&D. As for Bell Labs, remember that at the time it was truly ferocious, it wasn't allowed to do much with their technology because of the company's regulated monopoly status. They could develop UNIX for internal use all they wanted (and transistors and routing algorithms and...), but they couldn't actually sell it outside the Bell System. And by the time they could sell it externally, it wasn't like they had anyone left who could have productized or sold it for them.
In reality, corporate R&D has been dying for the last thirty years, except in the military space. It's a shame because the investors are simply eating the seed corn from which new products could have sprung.
Re:They aren't investors (Score:5, Informative)
Well, here's an anecdote.
6 or 7 years ago, when I was a low[er] level QA person at MSFT, I had a recurring meeting with someone from MSR because my division was using the new binary analysis and instrumentation tools that they had cooked up. I was one of the people implementing that toolchain in our production and testing process.
Now every product and every team at Microsoft uses that toolset.
Every year, MSR holds "Techfest", which is kind of like the science fair, except all of the experiments are awesome. MSR folks setup boothes/demos etc to show off what they've been up to. Normal MS employees attend this thing to allow for exactly the sort of informal, node-to-node idea exchange that ends up building the bridges from academia to engineering that you posit must not exist. And that is just one mechanism -- one that is accessible to low-level people in product groups for them to learn _what_ interesting things are happening, and who is doing them, and how to stay abreast of what's going on there.
I had an email conversation last month with someone at MSR who does visualization reseach about the publicly-downloadable visualization controls. I'm using them in one of my internal reporting tools and have some feature asks and was explaining some of the problems I'm having with the currently released bits. They've got new stuff they've been working on that will probably help me out when it's ready, and now they're aware of one more "real-world" use case for visualuzations of the type they're working on.
I'm a nobody, leaf-node QA engineer. And I've had interactions with MSR that have made my job better and easier, and the products I've worked on better.
Re:Ballmer's Xbox Fiasco, Search Insanity, And Oth (Score:2, Informative)
It looks like the 360 is pretty far ahead in the U.S. market and about 2 million ahead in the other markets, and then around 2 million behind in Japan. Of course, the Wii is dominating everything.
Re:Death march (Score:1, Informative)
The work done by Microsoft on Static Driver Verifier isn't new...
Yes, it was.
SLAM was not simply an implementation of existing techniques. Comparing SLAM to Lint is like comparing a safety belt to a life support system. If you really want me to get into what they do differently I can, but that would basically turn into a description of what SLAM actually does. Just for starters, you can't give Lint a description of what a program must or must not do; but this is precisely one of the inputs to SLAM.
SLAM was basically the first remotely effective software model checker for infinite state systems. (My office mate, who knows rather more about software model checking than I do and just submitted a paper to CAV a couple weeks ago, signed off on the accuracy of this system.)
It was also hugely influential in future research outside of MSR in the years since it's publication. It laid essential groundwork for BLAST from Berkeley(3 CAV papers, 1 PLDI, 1 ICSE (with FSE, one of the two top-tier software engineering conferences), 2 POPL), Magic from CMU (1 ICSE paper), Yogi from MSR (1 FSE paper), a couple other projects, and probably a handful of PhD theses.
Again, you quite clearly have no idea what you're talking about.
Re:Bill Gates? (Score:5, Informative)
Re:a lot of .NET development has been (Score:2, Informative)
They spend a lot on R&D, and they actually innovate quite a bit. I think Apple is more trendy right now and that is the reason they're making so much money with little investment. They're investing more into advertising, branding, and image than they are into technology. It seems that Microsoft is making a little more effort to do this these days, but it will be difficult for them to totally turn things around in this respect.
In terms of technology and innovation though, Apple takes what it can from open source, and contributes back exactly as much as they're legally obligated to. They didn't design the operating system, and a lot of their original system APIs come from the NeXTStep heritage (and thus are implemented in Objective-C, which is not hugely popular but was probably the fastest way for them to take NeXT's stuff and start selling it).
This isn't really accurate. You do realize that the list of original inventions by Apple is huge. Those of us who have been using Apple stuff since the early 80s really appreciate this in a way that someone who started paying attention to Apple recently can't. In fact in the 90s, Apple was teeming with great "Apple first" technology, but nobody knew about it or used it because they couldn't market it effectively. Newton anyone?
In fact, the long running joke is that Apple *is* MS R&D dept.
Apple does use and contribute to OSS, but thats honestly the whole point of OSS. You can't knock them for that. As far as them not inventing the technology in OSX, thats not correct either. Its not like OSX was available in the OSS community and they just put it in a box. They have spent years developing the technology that sets OSX apart from other UNIX based systems. Quicktime, Core Audio, Quartz, the whole OSX GUI, the search technology that is now Spotlight, etc.. And even the tech they got from NeXT technically is "Apple" because Apple bought NeXT and all of their employees and IP. So the Mach kernel, Cocoa, WebObjects, etc.. are all now "Apple". OSX is a mixture of original "Apple" and "NeXT", but its all Apple now, and there is nothing else quite like it on the market...
Just because Apple is "trendy" now doesn't mean they aren't and haven't been innovative. These things aren't mutually exclusive. And in fact, you'd want the most innovative companies to be the most popular. Its a shame it doesn't happen that way more often.
Re:Milk The Idiots Out Of Millions In Online Fees (Score:4, Informative)
You honestly think that 360s are outselling Wiis here? As somebody who actually works in a store selling video game consoles, I assure you that is not the case. If we could keep Wiis in stock for more than a day or two at a time I'm sure they'd sell even better, but as it is I've probably seen as many Wiis sold as PS3s and 360s combined.
Re:They aren't investors (Score:2, Informative)
I'm a programming languages grad student at the University of Washington, which is a 15 minute drive from MS. I'm not on the inside, but I see a *lot* of MSR talks. Nearly all the PL talks which I see describe testing and analysis tools MSR developed and nearly all of them end with something along the lines of "now it is mandatory that all checked in code be tested by our system." It seems MSR produces a *huge* number of in house tools for MS.
Re:Bill Gates? (Score:3, Informative)
There is a misunderstanding exhibited by many posts here that "R&D" is about "experimentation".
Only the "R" part of R&D is to do with experimentation, and that is what MSR does. The "D" component is product development -- the thousands of developers, testers and program managers, plus all the staff around them, that do the development of the many future products coming down the line, as well as the maintenance on products already shipped. The focus of development is basically to ship something.
MSR is not about shipping products, but about researching new technolgies in general; technologies that could either lead to future product features one day or whole new product categories -- but only after actual development (in a product development group outside of MSR) takes place. MSR does not work on product development itself except in rare and small cases where research can directly spin out into a product that generally doesn't fit into an established product group.
The dollar figures MS posts for R&D do not break out the proportion that is R and the proportion that is D. Since there are less than 1000 researchers in MSR, but multiple 1000s in development, "you do the math."
BTW, regarding tech transfer, MSR has documented its tech transfers for a number of years now. For example, the most recent list is here: http://research.microsoft.com/en-us/about/techtransfer2007.aspx [microsoft.com]