Baby, walk the walk

responded quickly to my little nudgeback earlier. And she’s sticking by her guns: But, dang it, this time I am not wrong.

Good on you, Dorothea. I like people who stand their ground. I rarely do so it’s very refreshing for me to see it in others. However, I have to respectfully disagree with your assessment that software can only be created by people who’ve done the work – walked the walk, so to speak. In some ways…, well, okay in some twisted ways, that’s the same as saying I can’t really understand what happens in the massage parlor business just from the explicit requirements given me by the Madame.

Good, well trained programmer/analysts are expert in interviewing clients and putting their words into requirements that other developers can follow. After you work in an industry for a while you pick up the talk, as I have with manufacturing; but you never need to actually do the work to get this understanding. It’s a question of meeting with the folks who are expert and knowing how to maximize their time. The key really is respect.

As for commercial software, at Sierra Geophysics, a petroleum industry software company, we would meet with engineers from the parent company, Halliburton, on a regular basis to check the requirements, and also to test out prototypes as well as review the manuals for completeness. We also had domain experts on site, which isn’t unusual for specialized software companies such as ours; but most of the requirements came from working directly with the clients.

When I worked at Skyfish, I also met with folks who ran airports and who manufactured planes to capture requirements from them. And our software was good because it was the only thing of value when the company crashed and burned.

If I had to walk the walk in each of these businesses in order to capture and document the requirements, as well as build the applications, I would be the first commercial pilot who knew how to drill oil wells, while working on an assembly line. (This in addition to being able to give a darn fine massage.)

Now, as Ralph mentioned in my comments, not all programmers can work with clients, or record requirements, and that’s cool, because not every programmer needs to. As long as everyone follows the spec, the application works.

The issue about having to be a domain expert in order to develop software just isn’t so. Not only that, it can be harmful to a company. A case in point: at one of the insurance company’s I worked with as a consultant, a group of insurance people who came over to the dark side and became programmers tried to convince management that it was impossible to document their application – too complex said they. You’d have to be a domain expert to understand.

Now, it is true that some of the calculations in the insurance industry put NASA to shame. However, the manager wasn’t buying it and called me in. His concern, rightfully, is that this group was about to split off into a separate company, forcing the insurance company to have to ‘contract’ for the software. The manager was a bit peevish about this because the insurance company had paid to develop the software in the first place.

I met with the manager as well as the lead developer and listened to the latter’s spiel about how the calculations were too complex to document, only a domain expert could understand them. He emphasized this by waving around an actuarial manual that could have killed a small dog if it landed on it. How to Intimidate Programmers, 101 – wave around a really big manual.

When he was done, I told him that if he could document the calculations within a programming language well enough so the computer could understand him, he could document them enough in English so I could understand them, because if I wasn’t a domain expert, then neither was the computer.

Ralph also weighs in on this subject.


Comments redux

Recovered from the Wayback Machine.

I’ve decided to re-enable comments for the weblog. Too quiet. At least the comment spams are a variation on the other email spam I get, and warblog baiting could become a favorite game of mine.

As for comments and their impact on writing and what we write – I found that web log statistics and page accesses provide more of a blunt assessment of writing than comments ever did.


Too much noise, too much chatter

A weblog I’ve been following forever is Dan Lyke’s Flutterby. Dave Winer pointed out an item at the Weblog User’s online discussion forum where Dan mentions that he’s been having problems with comments lately:

I always thought I wanted a daily readership of 10,000 or so, but recently I and my co-contributors made a few comments about the war, one of which ended up as #2 on a popular Google search, and we’ve attracted some real schmucks to the comment areas, and I’ve gotten more than usual incoherent paranoid and frankly just dumb emails.

Almost every weblog I went to this weekend that had comments enabled had either comment spam, a comment war, or, as Dan calls them a comment from some “real schmuck” included. Junk, junk, junk. Nice junk, bizarre junk, or nasty junk, but junk.

Thanks to the efforts of the warbloggers and those who have pushed weblogging as the Next Great Thing, our intimate circles have been crashed. The digital termites have invaded. The days of thoughtful discussions on-topic within weblogs are over. Long live the flame, the spam, and the hit and run Google searcher who can’t resist a comment form.

Dan wistfully discusses the possibility of sneaking off somewhere and starting a new weblog among a small intimate group of friends. If you do Dan, make sure you turn off Google. And, invite me.

Personally, I’m beginning to think the thing to do is stop weblogging until the war is over, and the warbloggers have moved on. But I’m afraid it’s too late. Weblogging’s personal and intimate side has been lost.