Impossible to compete without large sums of money.

I Find Karma (
Tue, 17 Jun 97 21:48:03 PDT

[man, Red Rock Eaters was good tonight]

> X-Url:
> X-Mailing-List: <> archive/latest/1626
> ...
> I keep trying to lay off Microsoft, but they won't let me.

Sounds like my line.

> For example, the 6/9/97 issue of Fortune quotes Bill Gates as follows
> (page 136):
> "IBM has decided that this market [groupware for corporate intranets]
> is strategic, meaning that they are willing to take noneconomic returns,
> hoping it will help them imagewise or marginwise in other businesses.
> Microsoft has decided that Exchange is 'strategic'. This means we are
> willing to take noneconomic returns. So you'd call this space hyper-
> competitive between IBM and Microsoft. It is kind of surprising to have
> somebody else [i.e., Netscape] come along and say, 'Oh, yeah, me too'."

Did we need any more proof that Netscrape has fools at the top?

> "Noneconomic returns", of course, is a euphemism: Gates is saying that
> Microsoft is willing to take a loss to ensure that its appalling Exchange

I wouldn't call Exchange appalling. Horrifying, maybe. Terrifying,
possibly. Frighteningly unusable, perhaps. But it's far beyond appalling.

> program is adopted as a de facto standard in the corporate groupware
> market.

Corporations deserve what they get.

> (Whether it's really Exchange that they keep pushing, or some
> other Outlook-based mailer, doesn't much matter. In fact, that just
> restates the problem.)

He got that right. This makes up for the grad student survey papers idea.

> This particular market is distinctive, of course,
> in that Microsoft is competing against a company with a similar capacity
> to take losses.

So what are the chances of them both dying in the process?

> The point, though, is that the software market has
> evolved to the point where it is impossible to compete without huge cash
> reserves to use in undercutting competition.

And that is why Armageddon is near. 1999, you heard it hear first.

> Venture capitalists are now said to evaluate new PC software start-ups
> purely in terms of the likelihood that Microsoft will buy them.

The end is near, repent!

> In the old days when the most famous commodities in the marketplace were
> oil or steel -- and even when they were memory chips made by Japanese
> companies -- this practice of deliberately taking losses for competitive
> purposes was regarded as the very definition of an anti-competitive
> practice.

That would be my definition, yes.

> In this case, the purpose of a loss-taking strategy is clear
> enough: once a de facto standard becomes entrenched in the marketplace,
> its owner can then leverage it to extract monopoly rents, and to gain an
> advantage in selling other kinds of software that benefit from integration
> with the standard.

Is there no way out?? Eudora, save us!!!

> Now, my own analyses of Microsoft's rapidly expanding dominion have almost
> entirely avoided this language of "anti-competitive practices", which has
> been central to most other analyses. Partly this is because the issues
> run deeper than that: because software has such different properties from
> oil and steel, and even from memory chips, it is quite possible in theory
> to establish a monopoly without coming close to the classical definitions
> of anti-competitive practices.

Good point.

> But more to the present point, even when those definitions seem to
> apply, the whole theory of antitrust is so tied to the properties of
> traditional commodities that it does not yield many legally actionable
> consequences -- as the US government learned a few years back to its
> embarrassment (see the article on this subject by Francois Bar et al
> that I cited in TNO 3(2)). You can't reduce market shares in the PC
> operating system market, for example, just by breaking Microsoft into
> several pieces, the way that Standard Oil was broken up a hundred
> years ago.

All you can do is pray that Microsoft software has been found to cause
brain cancer in innocent children. One good class-action lawsuit,
that's all it will take...

> On the general subject of opposition to Microsoft, the Boycott Microsoft
> site, which I haven't investigated well enough to either endorse or not
> endorse, is

Why boycott? It's not like it will do you any good.

> Just one more swipe at Microsoft.

Spoken like an addict.

> You've probably seen their ads lately promoting the idea of the
> personal computer.

Is there any way to AVOID seeing these?

> Everyone assumes that they are trying to ward off the threat of
> network computers -- at least the ones that do not run Windows. It's
> really interesting that a company is spending huge sums of money
> simply to promote an abstract concept about the relationship between
> human beings and computers. The real shame is that it's a lousy
> concept.

Hallelujah, preach on brother Weber!!!

> The whole idea of the personal computer was fouled up from the start,
> because people do not work in isolation from one another. People are
> social animals; they think together and they share things.

And sometimes they think they share things together.

> Personal computers, as Rob Kling pointed out long ago, are used
> by people who are located within a whole web of relationships, not least
> the relationship to their coworkers and system maintainers. The whole
> problem with viruses originated with the first personal computers' lack
> of a coherent model of the social relationship between the user and the
> software on her machine.

Well, that and some genius with the morality of an 8-year-old thought it
would be fun to stick a monkey wrench into the machinery.

> The bogus idea of the personal computer is responsible for the singularly
> unilluminating flap between the ActiveX and Java models of security for
> software downloaded from the Internet. ActiveX carries insurmountable
> security problems because it offers the user no defenses at all against
> bad behavior by downloaded software, besides someone's vague promise that
> such things wouldn't happen.

Oh yeah, that's a security model.

> Java is famous for addressing this problem
> by enclosing downloaded software in a "sandbox" that restricts its ability
> to grab hold of the machine's innards.

So ActiveX permits access to everything and Java permits access to
nothing. Nice security policy: a single bit, all or nothing.

> As the Microsoft people gleefully
> point out, however, this restricts the functionality of Java applets to
> relatively trivial applications.

Hence the name "crapplets".

> Microsoft owns the operating system, and
> they will happily explain the advantages of close integration between the
> operating system and the applications programs.

Yeah, now programs from remote galaxies of the Webiverse can crash my
computer, too. Microsoft has increased by several orders of magnitude
the number of ways my system can fail, die, choke, reveal sensitive
data, and clobber things I don't want clobbered! Woo hoo!

> It follows that the Java
> model will have to replicate many of the functions of an operating system,
> or that it will have to create a more sophisticated platform-independent
> interface between the applets and the operating system. This is pretty
> much of a contradiction in terms, of course.

Of course.

> It's important to understand how Microsoft benefits from the primitive
> technology of its operating system.

Windows is primitive technology? So simple a primitive could use it?
Or so complex it turns you into a primitive if you try to use it?

> I hope that I will not show my age
> if I suggest that this new generation of Internet-centric computer people
> might benefit from having a look at an ancient operating system called

Gosh, what's this guy, like 80?

> You've heard of MULTICS -- a timesharing operating system for
> mainframes.

Yeah, yeah, we've heard the party line from many a person that MULTICS
was the be all and end all of decent operating system design. So what?
The user interface blew...

> We all grew up being indoctrinated into the anti-mainframe
> religion,

Actually, I like mainframes. They make sense to me.

> of course, but now we've subjected the whole world to the evils
> of the opposite extreme -- a model of computing which lacks the most basic
> conception of how to permit multiple processes to operate simultaneously
> inside a computer without trashing each other.

Even if I'm a single user on an NT Sever 4.0 box.

> Yes, computer security people know all about this stuff. But they're
> just powerless to help us, because we've all become locked in to
> antisocial technologies that simply don't work.

Rohit, look! Web of Trust is important!

> Social creatures need protocols to safely negotiate gradually
> increasing levels of interaction and intimacy with one another.
> That's what normal social interactions are all about. And yet, if I may
> risk invoking a stereotype here, it is as if the whole personal computer
> model was designed by people who never had any experience of such things.

:) Bill Gates. Paul Allen. Steve Ballmer. Andy Grove. Gordon Moore.
Yup, pretty much sums it up.

On a different note...

> Recommended: Chris Freeman, The economics of technical change, Cambridge
> Journal of Economics 18, 1994, pages 463-514. A good survey article on
> studies in the (very broadly and heterogeneously defined) Schumpeterian
> tradition of economics. (You may recall that Schumpeter is the patron
> saint of entrepreneurs.)

Yeah, Schumpeter!

> Someone sent me a list of e-mail addresses of accused spammers. I was
> naturally tempted to broadcast it to the entire list, but the pitfalls
> of that plan should be pretty obvious. It's probably a good thing that
> vigilantism needs to be organized in a more decentralized way than that.

Maybe Seth wants to send him an email?

> I've noticed many subtle positive consequences of the Web. One of them is
> that people interviewing for jobs are much better informed than they used
> to be. When I interviewed for my current job at UCSD in 1991, I was not
> able to obtain even a simple list of the faculty in the department before
> I arrived for the interview. The last few people who have interviewed
> for jobs in my department, by contrast, have shown up in my office with an
> encyclopedic knowledge of my background, which they got from my Web pages.
> This is good, even if it's a little creepy to have someone actually know
> everything on my Web pages.

Amen to that. Suddenly people I talk with know a whole lot more about
me than I know about them.

> A few months ago, some RRE subscribers were kind enough to offer comments
> on the latest round of my students' studies of current or prospective
> Internet user communities. Some of their finished projects are available
> online through the class pages:

Interesting little projects.

> The big news in personal computers these days is laptops for under $1000.
> I think that's great. In fact, I have a laptop that I bought for $900.
> I even wrote 200 pages of my book on it. I bought it in 1989. It's a
> Toshiba T-1000, and it has the best keyboard I've ever used. It has no
> modem, just a serial port, but I could have gotten a modem for $150 if
> I had wanted to read my e-mail even more than I already do. It only has
> 512K of memory, but that has never been a problem because the program I
> use 99% of the time is Emacs, a powerful text editor that occupies such
> a small part of 512K that I have room left over in memory for about 100
> pages of text as well.

I wonder which Emacs this is. The emacs we use here is a nasty

> What's my point? My point is that laptops could
> have cost less than $1000 all along if our software wasn't so grotesquely
> bloated and nonmodular.

Oh! Another hit! He shoots, he scores!

> It will be objected that my T-1000 doesn't have
> a hard drive. But the only reason why personal computers need hard drives
> is that no high-capacity floppy-drive format has become established yet,
> and the main reasons for that are institutional and market dynamics, not
> technical limits.

Another hit! He's on a roll!

Which brings us back to...

> We are constantly told that technology is endlessly revolutionized for the
> better, but that's not really true. Some aspects of the technology do get
> steadily better -- hardware whose external behavior is standardized and
> software whose functionality had not formerly existed. But other aspects
> of the technology improve at glacial rates or not at all -- operating
> systems, for example, or software that has been standardized in such a
> way that its external behavior cannot change without breaking everthing
> else. Multics was a better operating system in many ways than what most
> people use today. My point is not that anyone is bad or stupid, but that
> we should ditch the big generalizations about the progress of technology.

My point is that there are a lot of bad, stupid people out there.

> Instead we should analyze things case-by-case, keeping watch for cases of
> technological lock-in that retard progress or create perverse incentives
> that make things worse.

We already know this exists; what the heck can we do about it?

Rohit, watch closely and you'll find a really cool argument for *TP...

> As a prime example of technological lock-in, I nominate the basic Internet
> electronic mail protocol, SMTP. SMTP is well over 100 in dog years and
> close to prehistoric in Internet time. Its shortcomings underlie many
> of the Internet's worst problems, for example the ease of forging headers
> and the consequent difficulty of meaningful screening of spam. It cannot
> change significantly without breaking a thousand applications. By far the
> most significant change to Internet e-mail is MIME. Nathaniel Borenstein
> and the other people who pushed MIME through are heroes, not principally
> for their undoubted technical genius but for their diplomatic skills --
> building consensus around a simple mechanism that permits complex data
> structures to be included in e-mail messages. But MIME is the exception
> that proves the rule: it is built on *top* of SMTP; its codes are included
> in SMTP messages, and the many e-mail programs that do not recognize them,
> instead of blowing up, simply display them in their raw form to the user.

Yeah, what a sweet Sunday surprise that is.

But hmmm... might there be another way to send typed objects by
stripping out the underlying dump transport layer and supplanting it
with something clean, cool, and efficient? The shadow knows...

> This is why I'm happy about the various next generation Internet projects.
> Instead of trying to extend the mess we have now, they're starting over
> fresh. Once the academics develop and exercise the new protocols in all
> of their shiny broadband glory, maybe we can imagine transitioning all of
> the worthwhile activities from the crummy old Internet to the new network.
> I don't even know if that's anyone's plan, but it would be a good plan.

Rohit, it's your plan!! In big lights!!!!

> At the same time, if that *is* the plan then everyone should be concerned
> with the philosophy that's embedded in these next generation architectures.

Maybe Weber'd like to join dist-obj?

> Are the Internet principles of decentralization still in force? Will the
> systems lend themselves to low-overhead operation, or will their economics
> or architectural presuppositions (or both) tend to impose the overhead and
> hassle of detailed accounting for every packet that goes by? Etc.

Argh!!! Would that we had that writeup of munchkins handy..........


No society can surely be flourishing and happy when part of the members
are poor and miserable.
-- Adam Smith