« Deja Vu All Over Again | Main | th0u $h@lt n0t $p@|v| »

Mail Call

From: sjon svenson
Subject: daynote
Date: Thu, 29 Jul 2004 23:53:55 -0700 (PDT)

I am looking from the other side of the fence and can add another good reason for many failing projects. Open specifications.

That's where the client (-state in your case-) gives specification and the programming starts. And then the client adds extra requirements and keeps changing the specs until the deadline is exceeded.

Of course usually more than one thing is wrong. Like choosing the right tool for the initial specs but by the time you reach the deadline the specs have changed so much that the tool is no longer appropriate for the job.

--
Kind regards,
Sjon Svenson


Date: Thu, 29 Jul 2004 21:20:59 -0700
From: JHR
Subject: Centralized Data Systems

Dan -

They DO fail - and with a great deal of regularity. There seems to be some reverse "economy of scale" effect. Californica has had ongoing problems with its attempts to computerize its Department of Motor Vehicles over the past 20+ years. At a glance, it would seem DMV operations would be tailor-made for automation. Didn't work out in practice, due to incompetence, politics, etc. - the usual things that beset Gummint operations anywhere. Turned out to be a vast money and time sink, with no end in sight. The Californica DMV site is still a dicey thing, not to be trusted atall.

OTOH - this vulnerability is not all bad. It tends to mitigate against centralized control and access to Citizen information - a VERY good thing! Tends to emasculate FBI attempts to implement yet another spying-on-citizens attempt (Carnivore, etc.).

Cheer up. There may be some benefit in even the worst circumstances!

Regards,

JHR

Have a Great Weekend Everyone - Aloha!