It seems like the author is trying to make the assertion that all sorts of poor decisions were made - it'd be helpful to have a comparison to other similar projects, in order to make that determination. Otherwise, these could be completely routine issues for all I know (as a layperson). If you took any software engineering project, and described it by detailing all the mistakes in isolation, it would look like a complete disaster.
The weld is really the key difference here. For welds to be structurally sound, and for them to work in the calculations that have been performed for the structural integrity of the structure, they have to meet a certain standard. These welds didn't. The company said that they themselves would state whether those standards were met. So this would be more akin to a computer chip manufacturer, where every Nth chip is faulty, and then still inserting them into a computer. This isn't about detailing mistakes in isolation. Nor is this about a software program where good enough is sufficient. This is more like the NASA launch controls where it has to be perfect (or however close NASA claims it needs to get to clear a launch program.)
Though I'm not sure that answers your question. Since I'm not showing similar projects.
Yep. The big reason that the convenient approximations (lies) we use in mechanical and civil engineering work is that it is expected that the implementer of our designs (elaborate lies) doesn't use the wrong parts--there are exhaustive specifications on things as boring as bolts and threads and welds specifically to ensure that the behavior they exhibit is within the realm of what the designer expects.
It's nice to see software engineering practices being backported to more conservative and stagnant forms of engineering. 8)
The entire design was done assuming proper welds. When they accepted substandard welds, nobody went back and did the calculations that probably would have shown they needed twice as many supports.
