Collection of Software Bugs(www5.in.tum.de)
www5.in.tum.de
Collection of Software Bugs
https://www5.in.tum.de/~huckle/bugse.html
10 comments
Can anyone explain how the bridge collapse in Miami was a software bug?
I'm going to go ahead and commit to saying it wasn't. I believe they are misinterpreting the findings of the National Transportation Safety Board. It sounds like, among other errors, they used inappropriate load parameters in the modeling software they were using. That's not the software's fault.
The reality is that despite the fact that we have decent software tools for structural design and analysis, they depend heavily on intelligent initialization of inputs from a human. While naively, that may sound like a software failing to non-civil/structural engineers, it would, IMO, be a nearly unsolvable problem to create software that can do it all for you.
For example, in this case, I believe one of the several failings that led to the collapse was related to inadequate structural analysis of construction staging. So the loading on the failed nodal connection was much greater during one phase of partial construction, and most of the analysis occured on the final, fully constructed design. (Though even in that state there were design failings.)
You can have a perfect model, but if you fail to actually use it correctly to assess for all the scenarios it won't help you.
The reality is that despite the fact that we have decent software tools for structural design and analysis, they depend heavily on intelligent initialization of inputs from a human. While naively, that may sound like a software failing to non-civil/structural engineers, it would, IMO, be a nearly unsolvable problem to create software that can do it all for you.
For example, in this case, I believe one of the several failings that led to the collapse was related to inadequate structural analysis of construction staging. So the loading on the failed nodal connection was much greater during one phase of partial construction, and most of the analysis occured on the final, fully constructed design. (Though even in that state there were design failings.)
You can have a perfect model, but if you fail to actually use it correctly to assess for all the scenarios it won't help you.
Is there some traceability on what input parameters were used in testing computer models?
Is there a hypothetical world where the modeling software failed to show this scenario for the correct input parameters?
Either way, I agree, based on the links presented in the OP, there is no evidence presented of it being a software bug.
Is there a hypothetical world where the modeling software failed to show this scenario for the correct input parameters?
Either way, I agree, based on the links presented in the OP, there is no evidence presented of it being a software bug.
It could have been if some software bug resulted in wrong calculations, but nothing points in that direction. AvE did a nice, but definitely not authoritative video on the subject.
His theory is that some structural part was missing because the bridge was in construction, and because of an obstacle on the ground the supports were not in the correct position.
Also a not-software-bug in this list is "Germanwings 9525 Locked Door". There is nothing software-related about a mad pilot locking himself in the cockpit and crashing the plane.
His theory is that some structural part was missing because the bridge was in construction, and because of an obstacle on the ground the supports were not in the correct position.
Also a not-software-bug in this list is "Germanwings 9525 Locked Door". There is nothing software-related about a mad pilot locking himself in the cockpit and crashing the plane.
I have randomly selected 3 links and 2 were dead. Using some king of webarchive might be a good idea for sites like this one.
Definitely incorporating this into my software engineering course. Thanks for this!
I'd sure hate to be the guy who lost $460M lol
"Knight Capital Group realized a $460 million loss in 45-minutes."
Eh, that kind of money is more or less pocket change to a Capital Group.
For some of the biggest capital groups maybe, but that loss actually destroyed Knight Capital Group: https://en.m.wikipedia.org/wiki/Knight_Capital_Group#2012_st...
Which means they were something of a penny-ante company by the standards of the street.