Comment by billforsternz | Hacker News Reader

Comment by billforsternz | Hacker News Reader

This is one of the reasons engineering is a difficult field, not just edge of the envelope engineering like this but routine product development. Sometimes it amazes me that anything ever gets shipped.

Think about airliners. After billions of dollars of development, you have to pass critical safety tests (eg evacuation time tests). What happens if you fail the test in such a way that you can't fix it without more-or-less starting from scratch? How can anyone sign off on the development schedule? Yet they do and somehow or other eventually products tend to get shipped.

Back in the day when I was a working engineer (embedded systems, telecom systems) I decided there was basically an unspoken conspiracy of silence. The pointy heads would demand a schedule. The engineering staff would provide it. There would be a generous allocation for all kinds of testing. But sometimes the results of testing would require new board layouts or other things that were basically catastrophic for the schedule (I'm thinking things like electromagnetic compatibility testing - stuff that wasn't in my wheelhouse thank god). The schedules would never factor in serious go-arounds like this yet it could and did happen and was one reason schedules were rarely met.

To me the unspoken conspiracy was that we know we can't really control the development trajectory, but we'll sign off on it anyway. Because what else are you going to do? Apologising later is more practical than seeking permission for an open-ended schedule.

To me, a "schedule" implies intent, not a promise. (Just like a "budget".) There always exist factors out of your control that can blow them, and everyone hopefully is cognizant of that fact.

Thinking that a schedule or budget is ironclad is the same thing as thinking you can predict the future.

Good point. Sometimes (often?) though the schedule is taken very seriously as a prediction of the future. The business case for the product is made because it will be ready for the holiday season, or such and such important trade show or whatever. Key people schedule vacations, transfers etc. and are unavailable to help finish the product once reality emerges from optimism.

The intent is important because it becomes the lid on the project. Without it, it is far harder to stay on track and make meaningful progress in finite time. In a good company everyone involved knows the risks that may make the schedule and budget slip and acts accordingly. Done right, the pressure does not rise unduly shortly before a milestone. It should rise earlier, but also to a much lower level. And a good manager will know when enough is enough and move the deadline before it becomes too stressful.

Akin's Laws of Spacecraft Design number 41:

  41. There's never enough time to do it right, but somehow, there's always enough time to do it over.

http://spacecraft.ssl.umd.edu/akins_laws.html

Well, with respect to your airline scenario - most companies would design along with those industry specifications in mind from the outset.

Even without doing so, considerable reverse engineering can still be achieved. The case of the De Havilland Comet comes to mind - they had to basically redesign the entire fuselage construction from the ground up once they realised the single layer pressure cell was a major design flaw.

I think the De Havilland Comet is actually a poster child for exactly how badly things can turn out rather than a good counter-example. The original mistake was apparently quite small - square windows instead of round. By the time they finally fixed this and shipped a solid and reliable product it was too late. Hundreds of people had died, Boeing and Douglas had entered the market with better planes and ate their lunch. The company died as well. You could make a case with only a little exaggeration that this mistake killed the British aerospace industry and setup the US for decades of dominance advanced technology industries.

Margin.

Reply on news.ycombinator.com