Dan Luu

The context here is unlike bridges, it's impossible to understand a whole codebase, and you can't rediscover legacy code without the institutional knowledge. Is that really true though? Not the code, the bridges. Are bridges super different from codebases in this manner?https://twitter.com/stevekrouse/status/1083909608262041600 …

I don't know any bridge builders well, and I don't know anything about bridge building myself. I poked around a bit and it seems a super long and expensive process? Here's a 120 page plan, not the bridge itself, just the submitted construction plan: http://dotapp7.dot.state.mn.us/eDIGS_guest/DMResultSet/DisplayDoc?docnumber=885466&identifier=885466 …

This is an example plan, and it's another 60 pages of tiny fine-print diagrams. Is all the knowledge of the bridge documented here? I dunno! I don't know how bridge building works. But my guess is... probably not? https://www.dot.state.mn.us/bridge/pdf/abc/br-62646-plan.pdf …

Thinking a lot of Dekker's work in safety in hospital systems, how they're massive complicated systems that nobody really understands, where the reality has drifted really far from the design of how things should work. I think most systems are like this, and code isn't special

Arguably there's probably a few (A FEW) ways that coding is more disciplined than other forms of engineering, like we have VCS. As an amateur historian lots of history is lost, and that makes analysis really hard, but git repos are something we only really see in coding

I really need to find more people who do both software engineering and some other form of engineering, to see what they think the differences are

(Gonna fanboy Leveson here, who does both software and areospace. Her take is that one of the biggest differences is no mapping between computational systems and physical phenomena, which unbounds complexity)