jr conlin's ink stained banana

Several years ago, i had an NEC MobilePro, and let me point out, at the time, it was wicked cool. It sported a full sized keyboard i could touch type on a 640×480 color touch screen and enough software to just barely be usable. The battery lasted a mighty 5 hours, it had a pretty crappy OS on it and it was a hassle to exchange files back and forth.

i later swapped it for a Palm based computer which had a smaller screen, MUCH longer battery life, no network connectivity and it was even harder to swap files back and forth. (Note to Palm: You know sometimes folks might just want to send a file back and forth and not sync everything, their brothers and two sheep.)

i learned a very important lesson from my experiences with that thing. If you want a computer, just get a freaking computer.

The reason i'm reminded of that is because i found out that besides the Sony UX50 lust toy (which i still prefer my Tungsten C to), there's now the FlipStart from Vulcan which is a not-quite-pda that runs XP (no surprise, Paul Allen is one of the principles here), on a five inch 1024×600 screen.

Sure, it's cute, and fits in your shirt pocket, but i'd note that the battery lasts around 3 hours, the optional port replicator (so you can plug in a grown up display, keyboard and other bits) is.. optional. And while i've not seen any prices so far, i'm willing to bet it will be squarely in the "not cheap" area.

i fully respect the value of a PDA as portable brain, particularly one that you can do things like try and write a few pages of story on when you're stuck in a boring meeting, but seriously, if you want a laptop, Dude, get a Dell. Having a geek toy is nice and all, but you'll be amazed how fast the gleam comes off of those things, particularly when you suddenly realize you've only got 10 minutes of battery left in a 60 minute meeting, and your $200 custom power adapter is at home.

Suddenly, that less alpha geek sitting across from you with the lap warmer he paid $300 more for and the $40 spare wall wart she got from Fry's doesn't seem like such a sucker after all, does she?

Damn, does it mean i'm getting older when i look at super sexy geek toys and think about how crappy they really are?

Nah, just means i get to laugh harder at the suckers who get them.

A bag of artificial butter flavored microwave popcorn.

Seven kinds of cancer in one neat bag. Mmmmm.

And let me wash it down with a HFCS laden Sierra Mist sugar water.

i can almost hear someone in Humboldt County crying.

Design is easy. It's the details that are hard.

One of the constants that i've always noticed is the fact that most engineers love to solve problems. As i've frequently noted, they're also really lazy and prefer solving said problems using the tools they've already developed, or the knowledge they already have. Database guys love databases and are personally convinced that databases are the only way to manage any set of data, and if you don't store your three line config file as a fully normalized database table, you're an idiot. Likewise Perl guys have unnatural love regular expressions, C++ guys know that there's an STL class that can be extended to handle that, and.. well, you get the idea.

The biggest problem, though is the fact that the vast majority of these designs tend not to think past the initial idea. Inevitably, anytime i've seen these sorts of designs there's a cloud or box where there's something scrawled like "Data Normalization" or "Process Communication". These boxes are usually pretty small compared with the elements of UI design or the layout of the tables, or the various class types. Unfortunately, those same boxes are what takes 80-90% of the time for any given project.

Why? Mostly because no one ever gives them the proper amount of thought.

There's a fear that most engineers have of "Appleing". This is named after Apple's semi-notorious method of doing such careful and methodical design and creation of perfection, that they inevitably miss their own market due to other (Redmond based) companies racing in with something less than perfect. Many engineers are afraid to spend too much time on design because it's dull and because they don't want to focus on too trivial details.

There's a really simple solution to that. Do design until there are no more questions or boxes. In other words, once you know all the inputs and outputs of a module and generally how the inputs become outputs, you're done, move on. Until you've done that, you cannot give any estimates about how much time or how complex a system you're really building. Nor should you even mention to someone else that you're "nearing completion" since you have no idea what "completion" entails.

Still, i'm well aware of the fact that i'm again tilling at windmills in this regard. What's worse is that i'm just as guilty as anyone else at forgetting to do that level of design.

But it's fun to be a hypocrite.