[Beowulf] Sidebar: Vista Rant

Douglas Eadline deadline at clustermonkey.net
Wed Jul 18 20:30:00 PDT 2007

> Uh... I must be wayyyyy behind in my entertainment quota.  Will the
> entertainment police stop by and force me to watch Oprah?

There was a television show years ago called "Max Headroom" taking
place in the "not to distant future", where it was illegal to have
an on/off button on a television set (they are always on). Those
having such machines were arrested. Maybe not to far off the mark
the way things are going, though I find myself constantly turning
off lone televisions in my house and I have not been arrested -- yet.

>From my perspective, now that Jack Bauer is on vacation after saving
the world (again) and Tony Soprano is ... (well who knows), and the
Simpson's are in reruns, there is nothing to watch on the
hundreds of content channels that get delivered to my house each
day. Which is why I'm enjoying reading this thread at 9:56 PM
on a Wednesday evening.

While I have some time to comment, I have just finished
reading a book called "The Wisdom of Crowds"
Quite a fascinating premise -- the right kind of groups
have been shown to provide better solutions than expert(s).
The idea is that experts tend to think narrowly, while groups
think globally and the random erroneous stuff cancels out.
One of the authors examples was Linux vs. Windows, but really
what he was talking about was the open source model (the bazaar)
vs. the top down Microsoft model (cathedral).

Of course open source has it's gatekeepers (e.g. Torvalds)
but, innovation can come from anywhere in the crowd. Linux
HPC clusters are a perfect example. What was once considered
"not feasible" (HPC with "free" software and commodity hardware)
is now a big market segment. Intel states that 20% of the server
market is used for HPC, and IDC surveys show that Linux is used on a
majority of these HPC cluster systems (in one survey, Linux
was 55%, Unix was the next largest OS with 29%). The crowd got it
right, the experts did not. Note, this discussion does not mean to
diminish the enormous contributions by many of the pioneers in this
area, but rather the fact that had the "experts" drafted
an 10 year HPC road-map in 1995, Linux clusters would probably have
been a non-issue.

So to wrap up my little RGB style rant (he is leaving for vacation
after all). Microsoft will undoubtedly manage to service a certain
sector of the market in the future, but just as the fluidity of digital
information has allowed them great riches, so to does it create
opportunities for others. I doubt that the next big thing will
ever come from Microsoft or any other monolithic organization,
but rather as it often does from the chaotic, un-predictable,
un-manageable hoards that collectively see the future. To me the
the open source development model seems to have harnessed
this energy better than anything in the past. This trend is what
Microsoft should fear, but then they are the market leading experts.


More information about the Beowulf mailing list