There's a longish, well-reasoned article in today's NY Times entitled,
"Internet Fuels Revival of Centralized 'Big Iron' Computing". The article is
at:

     http://www.nytimes.com/library/tech/99/05/biztech/articles/19net.html

"Well-reasoned" in this context means that the ideas that the article
portrays are in fundamental agreement with the conclusions that I've come to,
too :-). When we last discussed this topic about a year ago, I entitled the
subject line something like, "The death of the PC," and because of that, the
discussion went off on a tangent, arguing strenuously that PCs weren't dead,
nor were they anywhere near so.

I remember being too busy to reply at length at the time, but I totally
agree. PC's aren't going anywhere, but their role is changing dramatically.
Indeed, what's going on now is more of a paradigm shift than any of the
preceeding, pre-announced "paradigm shifts." What's changing about the PC is
that it's going to become a very bright and pretty terminal -- that has the
capability to run local wordprocessing and spreadsheet programs if desired --
but that's about it.

The era of large, near-mainframe-like PCs mounted on everyone's desk is
probably already over. Maintaining 500, or 5,000, or 50,000 of these devices
is an extraordinary expense. Over the last 25 years we've gone through a
number of "paradigms": peer-to-peer networking, cooperative computing,
distributed computing, complex client/server, thin client/server  -- but
where we're headed is right back where we started 35 years ago: a central
host surrounded by relatively simple, relatively cheap terminals. It is the
only model that is simple enough and efficient enough and easy enough to
maintain, particularly so on a large scale.

What I meant to emphasize in my earlier posting a year ago was that the PC,
as it has been traditionally used for the last 10 years, is quite clearly
dying. And at the same time, it will proliferate everywhere, especially now
that it has become cheap. And in the process, applications are going to move
off of the PC and back on the central hosts.

The NY Times article comes to the same conclusion:

"But Gates also warned of the rising threat to some Microsoft products
likely to come from online services. "A company such as America Online
is in competition for all our information-management software, because
they can do it through their servers," he observed.

"In Silicon Valley, dozens of start-ups have been created as Internet
services to centrally handle personal information. The new companies
mostly focus on e-mail, calendars and back-up file storage to insure
information is not lost when an individual's PC crashes.

"Many of these applications should be moved onto the Internet because it
is more reliable, available everywhere and cheaper," said Eric Brewer, a
University of California at Berkeley computer scientist who is a
co-founder of Inktomi Corp., a Web software company."

Several people have asked about the Financials/Payroll package that I
announced a year ago, when Cort more or less abandoned the HP3000. We're
still working on it, but after just a bit of thought a year ago, it became
obvious that once we fully develop it, we shouldn't sell it. Rather, the
obvious way to market it now is not to an installed base of HP3000 users,
which is a fairly constrained population, but rather to the world at large,
hosting the applications package here locally on our HP3000s and letting
anyone, anywhere run the software from their PCs, using a new and more
attractive version of QCTerm that is also currently in development.

This is exactly the same conclusion that Microsoft has come to independently
regarding their next generation of products. You're not going to purchase
them. Rather, you're going to pay for them as you use them, over the internet.

The second piece of the NY Times article that struck a particularly resonant
bell was the section on the death, or at least substantial diminishment, of
the DP department, a subject I also profoundly agree with, but have been a
little timid to mention. They write:

For all the attention understandably focused on the meteoric rise of online
retailers and auctioneers, like Amazon and Ebay, the biggest economic
impact of the Internet in the next few years is expected to be inside
old-line companies like Goodyear -- boosting productivity by
electronically automating back-office transactions.

Though Goodyear has an in-house datacenter, it chose to let outside
experts provide the computing power for its Web site -- a role known as
hosting -- and run its electronic commerce network. Many companies
are making the same choice. That is why the Web hosting business of
both established companies like IBM and AT&T, and newcomers like
Exodus Communications Inc. and Verio Inc., is projected to grow from
$696 million last year to $10.7 billion in 2002, according to International
Data Corp., a research firm.

Internet companies like America Online, to be sure, will have their own
server farms. But others increasingly view computing as a utility, a service
to be purchased like electricity. Indeed, technology historians note that
when factories began using electricity in the late 19th century, each had
its own power plant. Later, regional utilities were created and sold
electric service to the factories.

At Goodyear, Hargreaves seemed to apply the same logic to his
company's decision to have its Web site for dealers managed by IBM at
its server farm in Schaumburg. "We're in the tire business," he explained.
"Why run the digital power plant ourselves?"

Again, where we're headed is right back where we started, with
centrally-located time-sharing services. The only difference is that 30 years
ago, telephone charges made computing over any distance extraordinarily
expensive. In that, the internet changes everything, and outsourcing
information management is likely to become an exceptionally common process,
particularly so for businesses with 10 to 1000 employees. There are 10's of
thousands of these kinds of companies in the US alone -- and for the most
part -- they don't want to run their "digital power plants" either.

The world's changing. And there's going to develop over the next five years
an enormous opportunity for the establishment of centralized clusters of
highly reliable commercial database engines.

Wirt Atmar