IBM Invests $4 Billion to Build

Matt Jensen
Wed, 22 Aug 2001 18:59:42 -0400 (EDT)

On Wed, 22 Aug 2001, Adam L. Beberg wrote:

> You really think a company big
> enough to need 50,000 CPUs doesnt already HAVE them in house, oh please.

In an earlier life, I created Evolver, the first product to, among other
things, use a genetic algorithm for optimization in spreadsheets[1]. This
was before the Web broke big.  We used to talk and talk about building a
networked P2P utility-like "service", but it was a bit too early for that.

Typical problems business customers would solve would be distribution,
planning, process re-engineering, etc.  Real meat-and-potatoes,
non-linear, hard-to-solve-globally, save-real-money business problems.

The applicable thing here about business optimization problems is that the
complexity of a problem for a $5 million company is often comparable to
that of a $5 billion company.  And believe me, those $5 million companies
cannot afford their own dedicated network of 50,000 crunch clients.  The
IBM service could (hypothetically) allow small and medium-sized companies
to upload a problem and get a result in two minutes, rather than two days.  
It's economical, especially if it's a problem you only run once in a

-Matt Jensen

[1] We sold Evolver to Palisade, which now sells it.
       ( )

    I no longer get any cut for pimping like this.