Brad DeLong's Semi-Daily Journal: A Weblog

Much of what is now called the "Longhorn technology" was first promised for delivery in 1998, "right after Windows95 is done and we can get back to working on NT" in 1994 or so. (I think the name was Cairo but I confess to having forgotten as that game stopped being fun quite a while ago). With the exception of Digital Restrictions Management which is fairly new.

Microsoft has never brought out a significant technology advance on its own. Networking it got from IBM and 3Com. NT it got from Digital (shorthand). Windows 95 user interface of course from Apple and such technology advances as it contained were copies of improvements 3rd parties had made to MS-DOS.

W2K, which really isn't bad, was a rewrite of NT to look and act like W95 with a dash of BSD (Unix) networking code, Citrix remote access, and X-Windows user interface added.

Longhorn my arse. The only real innovation I expect from Microsoft in the next 5 years is the cleverness of their lobbyists in trying to write a law that will make the GPL illegal without destroying copyright.

Cranky

Brad,

You might also note that the apple version of WinFS, spotlight, is due to be shipped Q2 2005.

http://www.apple.com/macosx/tiger/spotlighttech.html

They call it an "integrated search" engine, but what it really is is a fully indexed file system. I don't know if allows transactions yet. But it does the most important part for desktop applications.

"For Apple to keep...its prices low."

When did that happen? (And I speak as Mac user.)

This has nothing to do with my beloved Texas Longhorns.

Microsoft's fundamental problem is the topology of its organization & IP/licensing scheme force it into committing all its resources into a single, monolithic campaign that takes years to realize, with no safety net for failure and little ability to correct the course along the way. Basically they have to bet the farm over & over, looking farther & farther ahead each time. It's a losing strategy in the long run.

On the other hand, Linux, BSD etc. can go in a hundred directions at once and capitalize on the incremental improvements of each project that succeeds while losing little from those that fail. Occasionally they win really big from a project that makes an evolutionary leap over the current methods. In the long run, it's sure to win.

What's the difference between the two strategies? Anybody who's been listening to me knows the answer to that - Self Organization is the key. Put transparency, stigmergy, reputation & arbitrary connectivity together & watch the magic happen. Self organization will become the dominant social, political & economic model for the 21st Century, because it just plain works better.

Tim

Some of this is a semiotic issue. While Apple's OS X was a "new OS", Jaguar and Panther were incremental updates with incremental new features, given spiffy names and shrinkwrapped in cardboard boxes for retail sale.

Since WinXP shipped we've seen the emergence of .NET runtimes, some big changes to the start menus and Windows Update, the court-mandated features that let users change their default web browser and emial programs centrally, advanced DRM added to Media Player, updates to IE, Outlook Express and Windows Movie Maker, a major rev of DirectX, multiheaded display support for DVD playback and a gajillion security patches and small UI and API changes.

If MS were Apple or a Unix or Linux vendor, this would be enough to justify a name change and a new product designation. From MS, it becomes an opportunity to frustrate customers because the user experience on a generic XP system and an SP2 system is significant but both are stuobbornly simply called "Windows XP" with no consumer distinction. One system will have no UI for selecting a default mail program and the other will. One will have a software firewall running by default. The other won't. One can play DRM-locked Windows Media files out of the box, the other can't. Such-and-such old pre-XP game will work on one machine but not the other, and so on.

This is good and bad. Major new UI changes like some of those pitched for Longhorn are disruptive in a corporate or institutional environment and IT departments would go insane if those things landed every year or two. On the flip side, by insisting on bundling dozens of major and minor new technologies and alterations into one monolithic release and holding up feature X because there's a big delay in progress on feature Y, MS ends up sitting on a bunch of things that an Apple or Sun or Red Hat would trickle out in yearly or biannual upgrades. To-may-toe, to-mah-toe.

Cranky Observer gets it exactly right. Brad, you should really have a longer memory for this stuff.

The major technology promises of Longhorn were never intended to be implemented. Sure, MS threw some R&D money at them (it's practically free, as far as their balance sheet is concerned), but the point was never, ever to actually develop a real product on a schedule that could be shipped to customers.

"Longhorn" was to the 2002-2004 era precisely what "Cairo" was to the 1994-1997 era: a set of technological "talking points" that would keep MIS managers convinced that switching away from Microsoft now (to OS/2 in 1994; to Linux or OSX now) would leave them locked out of the technologies that all of their peers would be using in 2-3 years.

Microsoft has been playing this game for a very long time now, and they do it better than practically anybody else in the business world. What's depressing is how easy it is for them.

(One small correction to Cranky Observer: Win2000's main feature wasn't the UI upgrade to make it more 95-ish. The big new thing in Win2k was the new security/authentication/directory infrastructure, known collectively as "ActiveDirectory", which was of course cobbled together out of a number of Open Source technologies, primarily LDAP and MIT's Kerberos.)

Dr. Memory:

While your description of MS strategy is correct, they didn't invent that strategy (see Longhorn, above, for MS on innovation). It was used successfully by IBM for 20+ years before them.

IBM got fat, dumb, and happy and let MS take the OS market away from them. The one thing in the world Gates, Balmer, and Co. are determined on is to do what it takes to keep it.

The MS combo believe (FWIW, I think correctly) that a major factor in IBM's paralysis was an excess of caution caused by the long-running ant-trust action against them. Thery're determined not to repeat that caution, either, and I think they've avoided doing so.

Isn't the Windows problem that Microsoft is trying to be too all-encompassing? Integrating IE into the OS hasn't brought much benefit to the consumer, but such efforts distract from the work of building a really good basic OS.

X1 (http://www.x1.com) is a great file search program, btw, and it's available today. Even in 2011, I doubt Microsoft will have come up with anything that would make we want to switch..

s.m. kopelman writes:

> Some of this is a semiotic issue. While Apple's OS X was a
> "new OS", Jaguar and Panther were incremental updates with
> incremental new features, given spiffy names and
> shrinkwrapped in cardboard boxes for retail sale.

Not exactly true. Mac OS X 10.0.anything was really a beta, and 10.1 was the first truly release-worthy version. The differences between 10.1 and 10.2 or 10.2 and 10.3 are fairly important, but the way to note this progress is to compare 10.1 (the first real version) to 10.3 (what we have 4 years later). *That* difference is truly massive. Everything works better, everything is faster, and there are new things (like Expose) that really change your experience fairly completely. In the mean time, they developed a new brower that's more standards-compliant than Windows, the iLife suite (stoopid name, great stuff), Keynote, and the best PDF reader on the planet. Hey, I'll even go out on a limb and say the TextEdit is a better Word than Word for what most people really do with a word processor.

These days I have to use XP, Mac OS X, and the latest greatest Linux stuff, and nothing out there is as strong all-around as Mac OS X 10.3, whether the steps to get there were big or small.

"The MS combo believe (FWIW, I think correctly) that a major factor in IBM's paralysis was an excess of caution caused by the long-running ant-trust action against them. Thery're determined not to repeat that caution, either, and I think they've avoided doing so."

Gates & co. may believe this, but what really happened at IBM was a lack of vision on the part of Jim Cannavino, the PC project manager. He let Gates & co in because it looked like a quick and dirty solution for a product that he didn't think was gonna be as big as it eventually got to be. If he'd paid attention to what he was assigned to do, the world would never have heard of Bill Gates.

The fallout from this at Armonk was huge, and sidetracked several major careers, including Cannavino's. He was a strong candidate for IBM CEO at one point.

It wasn't until Gerstner arrived that IBM regained it's footing.

Thanks to Cranky Observer and Dr. Memory for clarifying my own memory. I couldn't quite believe MS still hasn't got its Fancy File System developed after ten years trying.

The danger here is that we think this is anything to do with MS in particular---that they are somehow more evil than other companies. Jonathon Goldberg is right that they didb't invent FUD and they won't be the last ones to use it. It is an inevitable consequence of holding a potentially limited-term monopoly -- you do what you can to postpone the day someone else kicks it out from under you.

IBM's decision to outsource its operating system to Gates in particular was influenced by Gates' mother being on a corporate board with IBM's chairman at the time. That puts the lie to the "Gates, self-made man" myth.

However, IBM thought Gates would be a proper nepot and take the gift gracefully; instead he took their business away.

To take the topic of Microsoft's innovation and OS influences back further in time, when consulting a reference recently it was remarkable just how much MSDOS pulled directly from CP/M (data structures and functions), as well as Unix. The Microsoft Press reference (Advanced MSDOS, published in 1986, which was around the release of MSDOS 3.2; a museum piece), repeatedly explained the resemblances as intentional, designed to ease the porting of applications between operating systems.

> will there be any reason for people to
> switch to Longhorn? Or will it just be
> another resource hog?

Only Microsoft could make a pig out of a cow.

Tim Keller wrote: "Microsoft's fundamental problem is the topology of its organization & IP/licensing scheme force it into committing all its resources into a single, monolithic campaign...."

You're at least partially correct, Tim, and part of these announced changes addresses this issue.

Doctor Memory wrote: "The major technology promises of Longhorn were never intended to be implemented."

Given that I've been working with and on these technologies for the past two years, I would have to say that you are most likely not correct.

"Apple...and its prices low." I hope you aren't talking about hardware, or maybe you haven't done a dollars/horsepower calculation lately?

Or maybe you haven't, supersaurus, considering that Apple and the Univ of Virginia put together one of the world's most powerful supercomputing clusters at a much lower cost that could have been done using x86 technology.

I'd say Apple's G5 line is pretty damn competitive at it's price range.

For Fancy File System you need mram or something like. Harddisks are just to slow.

The cluster that was sure to give the wrong answer?

PaulB: My apologies, I was posting in the middle of a busy day at work, and sacrificed semantic accuracy for brevity.

Yes, of course, WinFS, et al, are actual technologies, with actual code being written. Like I said: R&D is practically free for MS; there's no reason for them not to throw money at dozens of projects to see if any of them pan out.

But there is, as far as I can tell, exactly zero connection between the amount of talking-time devoted to them by Microsoft's PR arms, and the actual likelihood of them shipping, in any timeframe, in a product that actual users can buy. And indeed, the more ambitious and nebulous the project (*cough* Cairo *cough*), the more likely it seems to end up being a chimera.

As others have noted, Microsoft is far from alone in this behavior: heck, it's damn near universal. What's unique to Microsoft is merely that they seem to be able to get away with it more often than not. As an IT manager myself, I find my professional peers' blind spot for Microsoft somewhat baffling.

PigInZen: The G5 was quite competitive when it shipped last year. At the moment, with IBM's 90nm production processes keeping the 970FX stalled at 2.5ghz (and with terrible yields even then), it's still...kinda competitive with Intel's current lineup, but not even on the same page as AMD.

I don't doubt that IBM will sort out their fab issues soon, and the G5 platform has other advantages aside from pure speed, but unless IBM pulls a serious rabbit out of their hat, Apple isn't going to be competitive on the high-end again until early 2006.

It could be worse. They could still be waiting for Motorola/Freescale to save them.

"If Longhorn slipped much past 2006, we'd be closing in on a decade between new Windows releases. . . Nor would it make many Microsoft customers happy."

Why would I be dismayed that my newest computer's obsolesence has been pushed back a few years?

Yay! Microsoft gets its turn at the Copland experience!

The open source community has put a huge dent in MightySoft's market. Open office, even the windows version is free. The many flavors of 'nix have been chipping away at the OS market for quite awhile now.