Tuesday, September 4, 2001

Well, gee, that was pleasant. I just pulled up a Shockwave page, and was prompted to update my Shockwave plugin. I figured, “Sure, I’ll update.” The installer then proceeded to close every browser window I had open before continuing with the install. It didn’t even warn me. How rude of the designers; they couldn’t even add a single line of code:

Alert *alert = new Alert( "To install the new version of Shockwave, all web browser windows will now be closed.", "OK" );

That’d be better than what I got, though of course, it wouldn’t be as polite as this:

Alert *alert = new Alert( "To install the new version of Shockwave, all web browser windows must be closed. If you click "Close and Continue", all browser windows will close and Shockwave installation will continue. If you click Cancel, browser windows will remain open but Shockwave will not complete installation.", "Close and Continue", "Cancel" );
if( alert->Button() == 1 )
  ...


OK, on to other things. I’ve been meaning to talk about Showstopper, a book by G. Pascal Zachary about the race within Microsoft to create Windows NT. (Here’s a presentation that covers a lot of the basics of the NT development effort, particularly compared to developing Windows 2000.) After reading about the strained families, the 16-hour-a-day schedules, the innumerable heated arguments, and all the other stresses that nearly broke a large team of highly intelligent developers, I was amazed to read this sentence in the epilogue of Showstopper!:

“Why did NT meet most of its goals and arrive, if not on schedule, at least in time to affect the competition? How did Cutler’s team avoid the loss of purpose and initiative that often burdens large teams?” — Showstopper!, page 280

To my widened eyes, this made NT sound like a success. Any project that came close to cracking that many people and that slipped its schedule so many times (the original estimate was 1.5 years; it took 4.5) is no success by my standards.

In no other industry — even within engineering — is chronic long hours and behind-the-curve schedules considered not just normal, but acceptable. Can you imagine any car manufacturer delivering the 2005 models some time in April of 2005?

We, as consumers and developers, can stop this cycle. If a publisher comes out with a schedule, expect them to stick to it. If you as a publisher need to set a public deadline, be generous.

The problem is, I’ve been reading that managers set aggressive schedules because of their belief that developers will slack off at the beginning of a generous schedule. Maybe they will. However, what’s more important: delivering well-tested, high-confidence products on time, or making sure that your developers are working feverishly?

Enough of that. Time for a new poll. Here are the results of the last one:

What else. Well, what did I do today? After work, I headed over to a friend’s house to play Nobilis. One of the players didn’t show up, so we chatted about DragonCon and played a game of Settlers of Catan (play it in a Java applet!), a fun little strategy board game where you try to build settlements and roads before your opponents. I wonder if my parents would enjoy it.

I didn’t get home until 11:30, so I just got in and crashed into bed. My oldest niece is here for a week, but I didn’t get a chance to see her.