M$FT

Microsoft spent $4.4 billion on research and development in the year ending June 30, 2001. $6.3 billion on R&D in the year ending June 30, 2002, $6.6 billion on R&D in the year ending June 30, 2003, and $7.8 billion on R&D in the year ending June 30, 2004.

Just what, exactly, is it doing? By and large, Word and Excel are the same programs they were when Wordperfect, Lotus, and Borland went down in the middle of the 1990s. Explorer is the same program it was when Navigator went down in the late 1990s. The mail and presentation programs are nice (perhaps), but. And there have been a lot of improvements to Windows: it doesn't crash any hour anymore, and the flaws that now have the Berkeley sysadmins on a search-and-destroy mission looking for installations of Windows 3.1, 95, and 98 by and large no longer exist.

What will users have to show for R&D expenditures that may crack $9 billion this fiscal year? What will shareholders have to show for this $9 billion. I know, they say "10% sales growth." But what would sales growth be if R&D were cut back to, say, $1 billion?

If anyone understands Microsoft as a business enterprise, I'd like to hear from them...

---

MICROSOFT FINANCIALS

All amounts in millions except per share amounts.
	2003	2002	2001	2000
	06/30/2004	06/30/2003	06/30/2002	06/30/2001
Net Sales	36,835.0	32,187.0	28,365.0	25,296.0
Cost Of Goods Sold	6,716.0	6,059.0	5,699.0	3,455.0
SG and A Expenses	13,306.0	9,988.0	8,095.0	5,742.0
R and D Expenditures	7,779.0	6,595.0	6,299.0	4,379.0
Income Before Depreciation and Amortization	9,034.0	9,545.0	8,272.0	11,720.0
Non-Operating Income	3,162.0	1,509.0	-397.0	-195.0
Other Income	-	-	-	-
Income Before Tax	12,196.0	11,054.0	7,875.0	11,525.0
Provision For Income Taxes	4,028.0	3,523.0	2,520.0	3,804.0
Income After Tax	8,168.0	7,531.0	5,355.0	7,721.0
Minority Interest	-	-	-	-
Net Income Before Extra Items	8,168.0	7,531.0	5,355.0	7,721.0
Discontinued Operations	-	-	-	-375.0
Net Income	8,168.0	7,531.0	5,355.0	7,346.0

Comments

[Unstable Isotope]

Microsoft is suffering from what all big business suffer from - the difficulty of doing innovation. Once a company has a big product, it tends to cannibalize competing products. The only way Microsoft is innovating now is by buying smaller, innovative companies.

[wolf]

For that kind of money, one would think that Excel could be expanded to more than 256 columns. The problem is that it would break all previous versions. Couldn't some of that marketing cash be used to come up with a new name? Doesn't MS want NEW products?

[Dan Ryan]

Two words, 'security patches'.

Seriously, I have no idea either. But I suspect a lot of the money went on the more marginally profitable, less mainstream products. Office does appear to be a nearly pure cash cow now.

[Laura]

I don't know about how Microsoft operates, but I have just this week chosen alternatives to common MS applications. Why did I choose? It's a mix of impatience (Outlook took FOREVER and then it would freeze), curiosity, rebellion, and a sense that MS isn't all that interested in making my computer experience better.

For example, we got a new PC here at home last week. We had Office on our old PC, and I used it a lot. This new PC came loaded with MS Works and a 60-day trial of Office. Well bust my buckles! Why? So I can shell out the bucks to buy Office, because MS-Works is crap? Microsoft made me mad; I can do better.

So I did: I've been using Firefox happily for a while, so I widened my horizons. Even though I still us Windows XP as an OS, I am using these alternatives:

IE ......... Mozilla Firefox
Outlook..... Mozilla Thunderbird
MS Office... OpenOffice.org 1.1.4

and I'm lovin' every minute of it!

My experience, and your post, makes me wonder what they are researching... how to widen the audience for their inferior MS Works?

How to make a new industry for post-market patches?

[Noe]

Microsoft's R&D expenses are not comparable to what is in an UCB NSF grant. MS includes what I would consider product promotions, and marketing activites. Consider all the development expenses that went into MS Chicago aka Win95.

MS has hired some very good researches such as DB expert Jim Grey (MS Terra Server) and others who use MS products to conduct research and hence try to influence the next generation of programmers/researchers. MS's Jim Grey is now working on implementing National Virtual Observatory, using MS software.

[Shai]

well they do a lot of basic research:

http://research.microsoft.com/research/default.aspx

and you also forgot that they also design programming languages, programming environments, relational databases, etc. and embedded/mobile/media center iterations of products

[chris m.]

does R&D get special tax treatment?

[knobboy]

Some things:

1. .NET
2. SQLServer database
3. Windows Enterprise server (NT4.0)
4. New OS - 2000, XP, Longhorn (whenever that thing gets done)
5. Embedded stuff
6. Mobile stuff
7. Tablet
8. MediaPlayer
9. All the Office products have been completely rebuilt a couple, three times since the mid-90s (so much so that older versions won't run in newer environents.
11. Lot of money on development tools (VC++, VB, C#, etc.)
12. Done a lot of Mac work - "MacTopia" - that's what they call it.
13. Rewritten Office to run on Linux/Unix, but you aint' gonna hear 'em say that.
14. MSN
15. Xbox

Which is not to say it's been quality stuff. But yeah, they've been busy.

[a]

M$ == Evil. That said, once your software is designed by the courts and goverment agencies, it should be no surprise that each line of code becomes as costly as a line in the legal brief.

[Bruce Wilder]

Microsoft has a monopoly. In the classic model of monopoly, the monopoly raises prices, by withholding production (i.e. reducing supply). Microsoft's monopoly is based on supplying the product at a very low price, flooding the market to eliminate competitors. Since the costs of production are mostly sunk costs, a low price x huge volume equals large economic profits. To put it another way, they make money by underspending on Windows development and maintenance. Unless a competitive offering rears its ugly head, they do nothing to improve Windows, because doing so would cut into profits. When a competitive offering does appear on the horizon, they improve Windows enough to destroy the alternative, maintaining the monopoly and minimizing the expectations of reward, which might lead to another competitive offering.

I did not realize that this was a mystery to anyone, with the possible exception of Schmalensee.

[Michael Cain]

Keep in mind that R&D is what the IRS and the tax code says it is, which is not necessarily the same as what one might expect it to be. In a previous job at a large cable company, I built and applied software that (1) measured high-speed data network performance for critical service components (eg, DNS lookups) from the end-user's perspective, (2) monitored end-user interaction with the set-top box software, (3) demonstrated the feasibility of providing certain architectural features in the network to support third-party devices. None of them led to "new" products. (1) led to changes in contracts with vendors. (2) helped identify problems with specific features in devices under consideration. (3) was an input to certain decisions about architecture that would be invisible to the end-user no matter which way the decision went. Two of the three produced patent applications. All of them (including salaries for people using (1) and (2) for usability studies) counted towards R&D tax credits at the time.

We know that MS has poured a lot of money into technologies like voice recognition, digital rights management, video compression, file systems, and search. A department with 30 people in it doing small changes (but thoroughly tested and documented) for the search engine for MSN, at $150K fully-loaded per head, is $4.5M per year. Ditto for a group doing development and comprehensive evaluation of a new video compression algorithm that doesn't make it into a product. And for a group that builds and maintains tools for managing MSN server complexes. Those are jobs that most people would agree are R&D -- but there will be a lot of other jobs for which the R&D definition can be stretched, and will be, for various accounting and financial purposes.

[arlie]

Yes, key, there is R&D tax credit. I don't know the current details of it, but reading through the tax code might clue as to why they've decided to increase the budget for it....that and understanding the pressure they are under to get into the next thing first.

There's a lot going on at MS Research, and right it's not about adding more features to Office. They have an annual exhibition (I don't think it's open to the public), but there are some details on Slashdot--here for instance: http://slashdot.org/article.pl?sid=05/03/06/1758225&from=rss
...and there are quite a few people that blog who happen to be at MS Research.

[ken melvin]

I think Microsoft was hearing footsteps. When they were being virused right and left, Linux, Apple and Anybody started to look good. Gates Inc. could see themselves losing it. Too, after the lawsuits, they can't, with impunity, pull the stunts they pulled with the FDA, US Appeals Courts, etc. I assume they are still working the obsolescence schtick.

[Cranky Observer]

> MS Office... OpenOffice.org 1.1.4

Keep an eye out for OpenOffice 2.0; it is in final beta and is said to be a great improvement over 1.x (which wasn't bad).

> Microsoft has a monopoly. In the classic
> model of monopoly, the monopoly raises
> prices, by withholding production (

While I belive the genesis of Micro$oft's R&D group was well-intentioned, for the last four years it has been engaged in piling up patents and copyrights for what Gates (and many others) view as the upcoming war over "intellectual property". The news that Hatch has been brought back to chair a special subcommittee of Judiciary on "intellectual property" pretty much cements this view. Along with Microsoft's work on "trusted" computing (where "trusted" means "trusted by Microsft and the MPAA", or in other words not trustworthy to the individual citizen).

> Some things:
>
> 1. .NET
> 2. SQLServer database
> [...]

Essentially nothing on that list that hasn't been available since 1975 or so, usually in a form far superior to Microsoft's tone-deaf reimplementation.

Cranky

[J.Asherman]

VIDEO GAMES XBOX.
3D virtual reality video games suck up money.
Further they are constantly redesigning the hardware.
And they are hard at work developing other products to replace competitors (and non-competitors) perfectly good products while letting their own stuff be a mess as all of you who use windows know.
The Mac is getting messy again too.

[Les Brunswick]

One of the reasons Microsoft's research expenditures are so high is that it hires up as many of the top minds as possible, just to keep other companies from hiring them. The bright fellows are then put to work at various projects that usually have no commercial future, but do keep them busy.

Also, Microsoft is desperately trying to come up with another monopoly to interlock with its present two ones. Its present monopolies are growing only slowly, and they will be undermined by open source if Microsoft doesn't create a new one. Expect lots of illegal maneuvers along the way, as in the past.

[a]

Interestingly,

Year end 6/30/2001 6/30/2002 6/30/2003 6/30/2004
R&D expense 4,379.0 6,299.0 6,595.0 7,779.0
Headcount 48,030 50,621 54,468 57,086
Per-developer 180K 250K 242K 272K

Headcount from http://www.microsoft.com/presspass/inside_ms.asp

Note that M$ does not break headcount by the year but in 2004 it was roughly 50/50 between sales, support and infra vs. dev. Assuming that held for the last 5 years, we have per-developer R&D.

Clearly, something happened in 2002. Again, from http://www.microsoft.com/presspass/inside_ms.asp: Fiscal year 2002 & 2003 results have been restated to reflect the retroactive adoption of SFAS 123, Accounting for Stock Based Compensation.

[jim]

Yeah, this is an accounting effect. Things you spend money on have to be shoehorned into one of three categories: Cost of goods sold, SG&A, R&D. For a software company, it's hard to fit stuff into cost of goods sold. Testers, maybe. You spent most of the money developing the goods you sold long before they shipped. SG&A covers marketers, lawyers, managers: "suits". The technical workforce basically has to go under R&D. Whether what they're doing is R&D or not.

[dsquared]

No, I don't agree. Try cutting & pasting extracts of Excel spreadsheets or charts into Word and Powerpoint in the XP versions. Then try to achieve anything like the same effect with any product available in 1997. Getting OLE to work is not a trival task.

[MrLipid]

One would think after all that R&D that a solution would have been found to the mysterious "Why can't Word properly display Page X of Y?" problem.

XyWrite III+ had a handle on that over 20 years ago.

[ecoast]

Two points.

1. Yes, it is an accounting gimmick, as others have pointed out.

2. No matter how much money MS spends on R&D, it really has no where to go. It is taking huge write-offs with its
corporate ERP packages Great Plains and Navision, basically junking those products and rewriting new software into that space, which will have moved away from MS, when the package is ready. Database SQL Server will never come close to Oracle or DB2. Games, wireless,
whatever it tries, there is no "windows' like monopoly in any of those and MS is far behind others.

The fact is technology is moving away from general purpose PCs and it is all special purpose, task-oriented, embedded systems now. MS has no feel for that technology or market. So it is stuck with coming up with newer and newer versions of its Office package to the point that those tools become unusable.

[Ian Welsh]

The business model is simple enough - because they have network externalities (otherwise known as "because everyone uses microsoft everyone uses microsoft") they are able to charge a premium for their products and almost everyone buys them.

This is a specialized case of rent.

http://www.bopnews.com/archives/002913.html#2913

[MrM]

I would not be too suprised if MSFT were betting (and investing) big on playing an important role in the upcoming utility computing battle - hence investment in fundamental research, High Performance Computing, Longhorn, Dynamic Systems Initiative, Business Solutions, etc.

I also got to believe that its Hardware R&D dollars go into more places than just Xbox, mice and keyboards.

[Tom Marney]

Two words: capital glut.

Over at Political Animal (http://www.washingtonmonthly.com/archives/individual/2005_03/005886.php) there's a post about the philosophical underpinnings of California's $3 billion stem cell research initiative. What I wanna know is why Bill Gates isn't paying for this instead of California's taxpayers.

[Steve]

If Microsoft is like every other tech company I know, all of its software and hardware (they have a few) developers are involved in "R&D" using the all inclusive capital D Development.

...and I mean every one of them, from the compiler developers to the C# script writers. If the janitor works on a script for the facilities web page they would probably claim her as a developer to.

The reason is the R&D tax credit, which allows substantial tax deductions for R&D. It has been once again extended for 2005 http://itmanagement.earthweb.com/erp/article.php/3412411

I've been involved in audits for the at a couple tech companies, and it is taken very seriously and it's BIG BUCKS.

[Faisal N. Jawdat]

I presume that the research group is there to produce the next Windows and the next Office, in two senses:

First, Microsoft has to come up with reasons for you to buy the next versions of Windows and Office every two years, or those revenue streams will dry up (or at least collapse to only that which can be generated by licenses sold with new PCs).

But, if you assume that the PC markets are saturated or close to saturated across the developed world (probably not true, but close enough that a SWAG says it will hit in the next ten years), then even if MS can get people to replace their OS and word processor every two years, its growth is still limited by the growth of its two cornerstone products (whose markets are saturated and presumably only growing at the rate of population growth in the developed world).

So, second, Microsoft has to come up with new products to continue growth beyond that which it gets from Office and Windows. To do this, Microsoft is funding basic research on a massive scale -- a scale that most companies can't afford, and most universities no longer do (they keep losing their best people to industry).

[Mumon]

They want to become a consumer products company.

So they spend their R&D money on designing protocols for how consumer electronics products talk to each other.

They also spend their R&D money on security stuff, embedding communications protocols in Windows, etc.

They're generally nice people when they're not to close to Broadcom...

[David Williams]

What many of us who work primarily in the Unix world forget is that M$FT have been plowing huge amounts of R&D cash into applications, some ill-fated, in the SMB sphere. For instance, B-Central was a colossal failure which M$FT devoted years of R&D to.

From the very late 90's to the present they have been fielding applications meant to compete with the BEA's, Oracle's and SAP's of the world. The reason we forget is that, on the whole, these have not been market-sector conquering apps.

[anne]

Steve

So, the answer I gather is there is not an express research and development department spending 9 billion dollars. Rather all employee efforts at any form of product development are "research."

[Mark]

I can comment on the .NET angle from firsthand experience...

We are now in the age of the "abstract platform" - software developers write their programs to run on the Java or .NET platform rather than targeting a particular operating system like Linux, WinXP, etc.

This is an important shift, IMO. Writing software to run on Windows is now safer, easier, faster, more secure, etc. .NET closes lots of those pesky security holes, things like "buffer overrun attacks" will not happen with a .NET application. The famous "this program has performed an illegal operation and will be shutdown" should fade away as more programs are written using the .NET platfrom rather than traditional C++.

Creating, maintaining, and extending .NET is a big project. I guess my point is that huge changes are happening even though, to the end user, the flagship products Windows and Office don't look all that different.

[Steve in Seattle]

As a former MSFT R&D employee turned economist, I can tell you that MSFT's R&D money is primarily spent "chasing tail lights".

Imagine a company culture which has a lot of type-A personalities who are rewarded with praise for doing unique things. The trick is, how do you keep thinking up new things? The answer is that you watch what everyone else is doing and then try to figure out how to leverage MSFT's monopoly position to do it "better" (with better margins) than the innovative competitors.

But watching/replicating what everyone is doing is expensive. And that's why they spend so much money on R&D. The primary case study is called "NetDocs", the Internet Service version of Office. Do you remember when Internet service providers were the next great thing? After spending $4 billion, the entire project was flushed.

Since this model will only sustain itself until the Windows/Office franchises runs out of "big money", the hope is that one of the big bets -- Xbox, Windows Mobile/cell phones, etc. -- will replace Windows/Office by then.

[Kuas]

I'm reminded of stories of the US providing funding for Soviet nuclear scientist after the end of the cold war just so they didn't go to work for the wrong people.

[Michael Robinson]

One of the key benefits to the shareholders of Microsoft's lavish spending on chronically mediocre R&D efforts is that it sucks up a huge amount of talent from the available pool of top-flight applications innovators who would otherwise be working for the competition to undermine Microsoft's monopoly (see, e.g., all the the trouble two Stanford graduate students can cause when left unsupervised).

Microsoft's R&D efforts are sort of like programs that pay ex-Soviet scientists to do busy work instead of building suitcase nukes for terrorists, but with an inversion of the moral polarity.

[Ed Byrne]

I can't understand it either. Looks like they're making the same mistakes Xerox did all over again. Having said that, there's a history or large companies with an equally large R&D spend failing to innovate - but at least Microsoft is ALSO acquiring innovations through company purchases, so while the R&D budget may be failing them, the company still may produce new products.

Funny you should mention this topic though ... I wrote an entry on Microsoft wasting billions on research the other day as well! Have a look if you get a chance ... http://www.ebyrne.net/2005/0314/is-microsofts-rd-spend-a-waste-of-money/

[Tleeves]

You make 8.1 income on 36.8 revenue. Make it in cents, dollars, or billions. Then tell me what's wrong with their business plan.

[Jonathan Goldberg]

Remember the title of Bill's buddy Andy Grove's book? It's "Only the Paranoid Survive."

MSFT's view of history is that every now and then a big change comes along, similar in magnitude to introduction of the PC. When it does, there is a shake-up of the established order; e.g. from IBM ruling the world to MSFT ruling the world.

Those on top after such a change are rarely the same ones as before. MSFT intends that next time will be different. To put it another way, they intend to be PARC *and* get the products to market. That's what all that money is supposed to buy.
Will it? And if so, when? I certainly don't know.

[Richard Cownie]

>I really wish Brad would stop writing on MS as if he has a clue.
>When someone talks about MS as if it's mostly Word and Excel, you
>can tune them out.

Microsoft only makes a significant profit on two product lines:
operating systems, and MSOffice. So it isn't surprising that
an economist would tend to ignore all the other stuff. Or you
could put the question more specifically: why does Microsoft
continue to put huge amounts of R&D money into product areas
where they don't make any profit ?

The answer is that they have a stock market valuation based
on the assumption of perpetual 30% growth, and they've
approached saturation of their current markets. They hoped
to take over the lucrative enterprise server market, but
Linux (together with security issues) has blocked that.
They wanted to take a cut of every Internet payment with
MS Passport, but that plan seems to have failed. MSN is a
flop. So now they're desperately searching around for the
Next Big Thing.

[Dave Schuler]

Several good suggestions of what all of those R&D dollars are being spent on have already been made including:

- Security (patching the results of poor design)
- Creeping featuritis i.e. incorporating all of the features your competitors have
- Planned obsolescence (making the current stuff obsolete)

One thing that hasn't been mentioned is securing Microsoft's monopoly. This is done by innovating incorporating features that reduce compatibility with competitive products and, in some cases, actively undermine competitive products.

[Cranky Observer]

> I really wish Brad would stop writing on MS as if he has
> a clue. When someone talks about MS as if it's mostly
> Word and Excel, you can tune them out. I'm not a fan of
> MS, but they have their fingers in many, many pies and
> these markets are vast

Would you be so kind as to name a few that have been successful, not to say profitable? SQL Server isn't a bad clone of Sybase, but it appears to have maxed out on penetration. Which from my perspective in an enterprise IT environment is a good metaphor for Microsoft's every venture into the enterprise space outside of the Windows/Office desktop.

Cranky

[lise]

What is entirely wonderful for us is that Brad thinks so questioningly so incisively in so many areas. As Brad thinks at us, we learn to think incisively "clue by clue by clue." Think on, Dear Brad.

[Joe Loserman]

Keith,

I read Brad's post to be a question asking for some expert
advice on Microsoft.

And what's wrong with non-technical experts making criticisms of Microsoft? Everyone is stuck using
computers to some degree in their life and having
opinions on them certainly doesn't strike me as being
wreckless as the crowd over at National Review.

-Joe Loserman

[Maynard Handley]

Keith,
Two people have commented on justifications for his doing so.
As the saying goes, put up or shut up --- give us examples of where MS has ventured and been commercially successful that are not XP/Server2003 or Office,
Yeah, there are plenty of examples where MS hopes and dreams that one day it will be important, from XBox to CE to databases to internet search. But my understanding (ie the common wisdom) is that all of these are, right now, dreams, and that, given its track record, there's no reason to believe MS will be especially successful in these ventures.

If the CW is wrong, please tell us all why --- many of us are non-dogmatic folks who are interested in learning new things, and happy to change our minds when confronted with arguments, but we require arguments stronger than "is too".

[anne]

Think how much of a research budget 9 billion dollars is. Why the budget should be such a size is an interesting question. From the thread, I suspect product development as such is "research." I have often wondered about such budgets for pharmaceutical companies.

[bi]

Rick Schaut, a Microsoft developer, has a blog over at blogs.msdn.com, where he tries to explain some of the things that Microsoft does. And the link below links to my take on the age-old "M$ vs. OSS" debate.

[anciano]

It's human nature to overdo things, to expand things which you believe have helped in the past, i.e. create Internet Explorer, invade Iraq, not worry about gas prices. This always produces painful corrections- anyone using IE today is bombarded with constant popups and misleading reports that your computer is malfunctioning and needs the help of a free program which you can easily download. Ms WORD crashes constantly if you work on large files with graphics.
Nobody should use IE, and most of us should avoid WORD, except as the final wrapper for a product that we send to others. Excel is OK but I don't see much improvement. Bush/Cheney/Rove will self-destruct because they are drunk with power and unrestrained by a corrupt Congress- problem is that they will pull down quite a few people with them. General Motors is speeding toward bankruptcy because of poor management and people who thought, "all we have to do is keep making the same shlock and the public will buy it".
It’s always hard to recognize when the rules of the game change.

[Richard Cownie]

This also brings up a wider issue which I've wondered about:
why corporations feel such an urge to grow beyond the things
they're good at. I understand the incentives for managers
to try to maintain growth even when you've saturated
existing markets; but it usually ends up being a big mistake.
In the seventies and eighties we had huge industrial
conglomerates (and in Japan and S.Korea, that's still the
predominant model), but now that model seems to be discredited
in the metal-bashing industries. With the software
market is maturing, surely it's time to move past the idea
of the software conglomerate ?

[Iain Babeu]

They are also investing in better math, which is really the heart of computer science.

The best example is the Active Directory (really just database of stuff on the network and those iutems attributes) and the maximum number of nodes and the time to converge -- the amount of time needed for all Active Directory servers to show the same data. Basically, under Win 2K AD the automagic tool MS provided to maximize performance and minimize convergence time had a limit of a couple hundred nodes before the algorithm produced results so ineffecient as to make the system unusable. Under Windows 2003 the algorithm was significantly updated to provide a maximum of thousands of nodes which are tuned without blowing enormous number of man hours tuning the system.

This is not the only computer science endeavor MS has invested in. The original algorithm for that bastard Clippy was a very elegant solution to determine when someone needed help, it was never implemented so we all know "Hey! It looks like you're writing a letter!"

Some other tools invested in are designed to increase interoperability. For instance, they have developed 3 sets of "generic" command structures that allow an admin to only learn one set of commands that can then talk to multiple vendors Storage Area Network products and perform all the functions they need for SAN hardware, software, and volume management. While much of that money will be spent building business relationships with those storage vendors, I'm sure it is counted as R&D.

Some of the money is being spent legitimately, although that is no gaurantee it is being spent as well as it could be.

[Cranky Observer]

> The best example is the Active Directory (really just
> database of stuff on the network and those iutems
> attributes) and the maximum number of nodes and the
> time to converge -- the amount of time needed for all
> Active Directory servers to show the same data.
> Basically, under Win 2K AD the automagic tool MS
> provided to maximize performance and minimize
> convergence time had a limit of a couple hundred nodes
> before the algorithm produced results so ineffecient
> as to make the system unusable. Under Windows 2003 the
> algorithm was significantly updated to provide a
> maximum of thousands of nodes which are tuned without
> blowing enormous number of man hours tuning the
> system.

Impressive - until you realize that with the Windows 2003 Active Directory redesign Microsoft has achieved about 1/3 of the performance that Novell's NDS demonstrated in _1997_.

That is what drives IT professionals with deep experience and memory nuts: this unstoppable meme that Microsoft has somehow been achieving Great Things(tm) over the last 10 years when in fact they have mostly reimplmented the work of others - poorly.

Cranky

[anne]

http://www.nytimes.com/2005/03/18/technology/18soft.html

Europeans Doubt Microsoft Is Obeying Antitrust Order
By PAUL MELLER

BRUSSELS - The European Commission has "strong doubts" that Microsoft is obeying an order issued a year ago in a landmark antitrust ruling against the company, a commission spokesman, Jonathan Todd, said Thursday.

It plans to discuss its concerns with Microsoft, and could impose fines up to 5 percent of daily worldwide sales until it decides that the company is complying.

In the case, the largest antitrust action ever brought in Europe, the commission found that Microsoft broke European Union law by using a "near monopoly" in its Windows operating system to squeeze out rivals in other kinds of software....

[Rob Salkowitz]

I consult with Microsoft on future product planning (mostly for Office, but also for other products), and in my experience, they take a rather long and broad view in their R&D. As was pointed out above, they do a lot of basic research on future technologies. They are concerned about business models (e.g., open source), but also about ways that issues like demographics, globalization, trends in politics and economics, and the shape of communities will impact demand for their products 10-15 years out.

It's indicative of Microsoft's poor marketing efforts around the latest version of Office that Prof. DeLong can make baseless generalizations about Word and Excel, as if that were the totality of Microsoft innovation around information worker productivity. Actually, Microsoft, along with IBM and Oracle, are delivering some very interesting solutions for communication and collaboration, mobility, information management, application integration and other areas that affect organizational rather than personal productivity. These are having a transformative effect in many industries, especially healthcare and government, which did not benefit fully from the previous generation of information work tools.

[pc]

Re: pharmaceutical vs. software

In a world where you are doing wet lab experiments,
cost of facilities and equipment can chew up large amounts of money very quickly. In software development, the main cost is people. At $200K / person, $9B is 45,000 professionals. It is very difficult to imagine from the discussion so far what such an army of IT professionals could be doing that would not lead to a torrent of new products (which we are not seeing). Asserting that it is all research / long-term development can't account for this kind of expenditure.

[Cranky Observer]

Rob,
Here's a hint: whenever you use the word "around" to describe how a product works, as in

> as if that were the totality of Microsoft innovation
> around information worker productivity.

you are not working on anything that any real organization outside the Accenture/McKinsey orbit is ever going to use.

Cranky

[peBird]

Instead of the absolute value of R&D, look at the ratio of R&D to SG&A over the past four years. They want to turn into a media company, they've got the SG&A to prove it.

[Wm Cowan]

I am surprised not to have seen comments about anti-trust in this discussion.

A prominent defence of monopoly in technology anti-trust law-suits has been "the basic, long-term, research that wouldn't get done if we couldn't keep all the revenue from our existing products". MS Research, as opposed to development, started to grow aggressively only when anti-trust started to heat up.

Is it worth a billion a year to maintain an OS monopoly? You bet it is.

Bill

[RickG]

Most of what MSOFT delivers in the form of software has negligible "manufacturing costs". Virtually all of the costs is arguably R&D. Yes, this is a very lucrative Tax area. The large accounting firms have tax specialists whose entire job is running the documentation projects for the deductions.

Of the Office apps, I haven't seen anyone discuss the impact of Excel, and its use in an awful lot of accounting and finance applications. Companies are now howling about the Sarbanes Oxley (SOX) expense of documenting the internal controls of hideously complex spreadsheet implementations. Often times companies find they need to redo applications when they accidentally riffed the "wizard" who created one of these. No one else understands how it all works.

The comment about there being nothing new about active directory is so true. Mainframers can look at it and recognize it as a descendant of the catalog, and instantly realize it will run like a dog since it isn't designed for performance.

As to security, a lot of the patches ought to be called "integrity" patches. Code written to allow break-ins without nary so much as an attempt at authentication, isn't about security. IBM had an Integrity statement for its MVS operating system, which guaranteed code could not gain privileged state unless you configured it specifically to be allowed to do so (APF Authorization). More than a few companies that are "stuck" with MSOFT apps are starting to run them in Network restricted ports (via ACL) environments. This is a hellishly complex endeavor, but it is a legacy of letting the suits take over IT and implementing this supposedly cheaper operating environment.

[MrM]

One other thing (this is an economist's blog after all :-) ) - it may also have to do with capital market signaling

Imagine you are a stock analyst and you compare R&D expenses across high-tech majors (both in absolute $ and as percent of sales). What will you do if see MSFT drop way below the average? You probably will adjust your long-term growth expectations, which would in turn kill the stock.

Now, if MSFT's R&D expense is on par with others and given all the discussion above, what would you, a stock analyst, do? You can punish MSFT even more for wasting the shareholder's money. Or you could think of that as a VC model: most of investments flop, but a handful make huge returns. (Like it or not, but IE, after all, did kill Netscape.) You might then give MSFT benefit of the doubt.

Finally, now you are Gates/Ballmer and you can't help but worry about those stock analysts. What do you do with your R&D money? There is no way you will concede defeat and turn the spigot down.

[MrLipid]

(Like it or not, but IE, after all, did kill Netscape.)

---

If I remember correctly, MS licensed the core of IE from Spyglass and then gave away the browser, generating no income and thus requiring no payment to Spyglass.

Which is, I guess, some sort of innovation...

[Keith M Ellis]

"As the saying goes, put up or shut up --- give us examples of where MS has ventured and been commercially successful that are not XP/Server2003 or Office."

I never made the claim you're asking me to prove. (And how do you define "commercially successful"? Also, within their server line is a host of applications that were seperate markets where people have migrated to MS's solution. Are they best of breed? Usually not. Did they leverage their OS to get people to migrate? Of course they did. But neither of those things means that they didn't have to spend huge amounts of money to create competitive applications that were at least comparable to those that already existed.) My objection was to Brad characterizing MS's output of consisting of Word, Excel, and making their OS crash less often. If you think that is all MS is, then, yeah, you'd wonder where that 9B is going. But what Rob said. Spend some time at MS's site and look at their genormous number of different product lines. As I said, they have their fingers in many, many pies. Are their products very good? Not usually, in my opinion. Are their products original? Also usually not. But R&D in this business doesn't necessarily mean "completely original work". (And Cranky, SQL Server isn't a clone of Sybase, it *is* [or was] Sybase. Same source code.)

Brad's question is a good one; but it's a good question in the context of *any* big IT company. MrM is closest to the correct answer: the industry demands big gambles be made and you hope that you strike it rich enough to stay ahead. What is Oracle's R&D budget? What has the consumer seen from it? For that matter, what has the consumer seen from Intel's mind-boggling R&D budget? Well, processors are really fast. But the performance increases relative to user perception are *not that large*. Why spend so much money on this? These are good questions, and they are good questions for an economist to ask and attempt to answer because there's good reasons to think that IT in general is a very inefficient industry. They're not particular to Microsoft.

[anne]

PC

Thank you. We need to pay more attention to the accounting involved both for Microsoft and for pharmaceutical research. The drug companies double the research cost for drug development by imputing the money they could have earned had they invested rather than actually developed a product. So, 100 million dollars in investment on research is really 50 million. There is much to consider.

[anne]

http://www.nytimes.com/2004/09/14/health/policy/14conv.html?ei=1&en=a8bdb044e3a15c06&ex=1112360225&pagewanted=all&position=

A Doctor Puts the Drug Industry Under a Microscope
By CLAUDIA DREIFUS

[Here then are Marcia Angell's thoughts on drug research.]

[anne]

When we consider a research budget that would support more than 40,000 professionals, we are likely considering all efforts directed at holding market share.

[Patrick]

You guys are hilarious. "SqlServer is a failure," "MS has failed to penetrate the server market," "they buy researchers so they don't work on successful products." As long as you don't read the papers.

If only MS didn't exist, then the truly noble projects, Linux, that 10-20 year old technology, Java, the free software owned by Sun, Apple, the real David, they would bring us to the libertarian free market paradise.

I assume you also believe in personal accounts and the apriori evils of the federal government.

[pj]

Hey, Wordperfect didn't "go down" in the mid 90s. It still exists, it is still better than Word, and I still use it every day.

[Jim S]

Inevitably the discussion about Microsoft that involves a certain breed of techies devolves into a religious war. Cranky, a and some others will criticize anything produced by Microsoft. I gripe at their products a lot. Of course I gripe at some other products just as much if not more. I just don't try to pile all of the blame for problems in the "computing world" on Microsoft or equate them with evil.

[MrLipid]

And I use Nota Bene 7.0b, soon to be 8, built upon the tremendously powerful foundation of XyWrite.

Unlike Word, Nota Bene considers FOOL to be both a noun and a verb. Word only considers it a verb.

And there are no synonyms for IDIOT in Word.

What sort of word processor takes it upon itself to decide which words are appropriate and which aren't?

[ecoast]

MS, because of its very nature of being in PC business, doesn't have too many opportunities to excel in R&D.
It is stuck in Windows and Office for ever.
Small and Medium businesses are its forte and how much innovation can it do in those sectors and those sectors are not clamoring for newest and latest technologies anyways. It tries to port windows and office to other platforms like cellphones and PDAs. That's its vision, because it sees the whole world in that narrow frame of mind.

1. IBM rebounded by emphasizing services. MS can't play in that space. IBM also bought into Java with its
Websphere platform and it si paying dividends with big
corporate accounts. IBM R&D is heavily into new chip design.

2. Oracle expanded by moving away (or extending) its database with Oracle applications which is entirely different space (occupied by SAP, peoplesoft, JDE).
Now Oracle Applications can stand on their own in the
enterprise resouce planning space and by buying Peoplesoft it doubled its market penetration.
Oracle R&D wenbt into Oracle Applications (manufacturing, financials and different verticals).

3. Intel goes linearly with its Pentium based chips,
but also extending into networking chips and other
markets. (PC-based microprocessors are about 1 to 2% of the whole embedded market.)

4. HP, after acquiring Compaq, became just a PC company.
No innovation opportunity there.

So where are the next killer opportunities in IT?
Distributed special purpose embedded processors, combining different processors in creative ways to solve problems in factory automation, transportation, healthcare, real-time enterprise applications, artificial intelligence, sensor dust networks, nanotech applications, bioinformatics and so on.

Now how can MS compete in any of these areas?
Also, most of MS applications (besides Office) are third party applications. No intellectual property there for MS.

[tcs]

I'm surprised that we're sixty some comments down and there has not been one mention of SharePoint.

There is much to complain about with MS, but incorporating the base version of SharePoint (not Portal) into Windows server is not one of them.

[bi]

Patrick s3z (summarized), "WAH WAH WAH U L33NUX L1B3RT4RY4N H1PP33ZZZ!!1111111" Funny that someone who complains about others who "don't read the papers" is ignorant of Richard Stallman and Linus Torvalds, both of whom are as far from libertarianism as one can get.

Well, I freely admit that Microsoft may not exactly be the spawn of Satan. But it's a fact that Clippit sucks, it's a fact that Word's interface for entering formulae sucks, it's a fact that the spelling and grammar checking also sucks, and it's a fact that Microsoft has cast aspersions of GPL by calling it an "IPR impairing license" ( http://www.advogato.org/article/453.html ).

[Keith M Ellis]

As I've said repeatedly, I have no real desire to defend Microsoft. But I am unbelievably tired of what seems to me to be posturing MS bashing. I think what grates the most is that 99% of the swipes at MS are cliches, truisms, or false. Meanwhile, there's a huge number of *informed* criticisms of Microsoft that can (and should) be made. Maybe this subject is relatively novel for Brad and most of his readers. As an IT person, the OS wars are as stale and silly to me as geeks arguing over the science of Star Trek. XP is 2000 which is NT. How many people that regularly bash MS's OS stability are aware that NT 3.1 was famously stable and which had as its design chief the designer of one of the most well-regarded and stable OSs of all time? And complaints about Word's grammar checking?

Back to the specific issue at hand: people have short memories (or haven't been around that long). As someone who worked in a software shop near the end of the height of the UNIX workstation and server era, there was near universal scorn for MS's attempts to enter that space. But look what's happened. And it didn't come cheap for Microsoft.

Microsoft is *mostly* about consumer operating systems and Office in terms of revenue. But over the years I've watched them relentlessly move into other markets. There's numerous categories of low to mid level back-office software that Microsoft now competes in, and they are suprisingly successful at some high-end stuff, as well. And look at the XBox. And the gaming software they develop. They are also a peripheral hardware company. The usual for them is only middling success in all these other markets—but even given their near-monopoly status in the consumer OS market and the leverage that gives them, it's still unbelievably expensive for them to keep entering these other markets, play catch-up, and achieve even moderate success in them.

[bi]

I don't see anyone on this thread bashing WinXX's stability -- looks like there are already enough other (valid) things to whine about. If anyone wants to whine about M$ bashing, he can at least refrain from setting up straw men.

[MrLipid]

And complaints about Word's grammar checking?

---

Word is allegedly a specialized tool designed to provide support for those who write by checking grammar, offering synonyms, etc. That it does these and other tasks either inadequately or not at all makes it a poor tool.

The larger question is why, with MS awash in cash, it remains such a poor tool.

The answer, as with IE, is that once MS owns a market, it pretty much stops making much of an effort until threatened.

[enfant terrible]

Maybe they conduct a lot of research into methods of destroying competition.

[hey]

where does R&D go? remember that it includes development, all you whiners above about how product development isn't "research"... of course not, it's development, or did you forget what the letters stood for!

also, office has changed. look at the task panes, with inclusion of web services and interesting internet ties. is it apparent to everyone? no. does it matter? yeah as it allows you to do web research within your office document... you can get factiva business profiles, revenues figures, etc, within word (if you have a factiva sub) similar for other web services. This is EXPENSIVE, as it involves a lot of code despite a minimal interface on the front end.

Why does everything at MS cost more and do less than specific product X (re Novell's products here, but applicable to most every complaint). Installed base. MS has to make a much larger universe of test cases work. Lots of people have complaints about how WIN NT sucked (or 2000 sucked) because it wouldn't work with accessory X. 2000 was a server/worsktation OS not intended to be used with infinite consumer hardware, but enough people did that it got a bad rep. Active directory has to deal with insane test cases to ensure that people don't think its crap because it can't interface with win 3.1, 98, 95, nt2, BSD 4, Oracle 6, etc.. all on the same network.. i mean whats the big deal? it should just work!

when you're a much smaller vendor, or much more specialised, with a minimal installed base, you can avoid this blame. Novell doesn't work with X? different companies, can't be expected, plus its a new product. different expectations, different results. plus windows allows for multiple comm protocols... netware, not so much.

but the entire concept is false. MS is like a drug company but moreso. If it is perceved to not be investing, then mass defection is likely, and the upgrade cycle chokes off quickly. there are complaints now about no apparent new features or reasons to upgrade. if MS stopped or massively cutback R&D, it is lilely that the cash would dry up. no one buys EOL products, especially EOL software.

[Jonathan Goldberg]

Iaia Babeu comments:

Some of the money is being spent legitimately, although that is no gaurantee it is being spent as well as it could be.

Figuring out how well R&D money *could* be spent is an "angels fear to tread" area.

Some years ago a friend of mine who worked for P&G told me about a new VP who had taken over as R&D chief. Noting that small companies tend to get about twice the return on R&D as big ones (in the US, that is), his first missive to his new fief was that he'd fire half of them if he could only figure out which half.

A real morale builder, this guy...

[name]

They sure are working on the 64 bit LongHorn platform, in which they will try to turn even the concept of pdf to a M$ propietary format. Also, they're trying to ruin Google by shamelessly copying their concepts.

It looks like pushing fwd great ideas is not worth it anymore, since M$ may come any moment and ruin your project. They've done it before, as you said, with WordPerfect, Lotus, Netscape, RealPlayer and many others.

These people suppose a huge threat to progress and sw user rights.

[danejaneiro]

M$FT EXPENSES 100 percent of their R&D. It doesn't even enter into their capitalization. This contrasts with, say, Oracle, which annually capitalizes about 200 percent of its R&D. That's pretty remarkable.

[James Robinson]

MS spends enough money on R&D that it's hard to provide a capsule summary.

They do legit basic research. I read about a 'softie who's working on IMAX-style curved screens, sized for personal use, and the algorithms necessary to map the current flat displays to the curved surface without distortions. Basic research funding picked up quickly during the antitrust trial, when they were trying to define themselves as innovative.

They also do "tailight-chasing," talent-hoarding and patent-hoarding. (It's not accidental that C# looks a whole lot like Java, although they did have the decency to make a few improvements.)

They also spend a great deal of time and money coming up with "next-generation" Windows technologies and interfaces, which are subsequently abandoned in the face of legacy compatibility.

They also donate millions of dollars worth of workstations and licenses to universities, in an attempt to unseat various UNIXen and catch developers young, and write it off as "R&D."

I don't doubt that I'm missing a number of additional categories. But that's pretty close to comprehensive.