If you really want to compare software to bridges, imagine that humans had written the same simple program millions of times over thousands of years. We'd be pretty good at it by now. (Even that analogy, though, doesn't level the playing field. The physical world is not programmable.)
Why do people assume that it is acceptable to make mistakes in the software world, but not in the "physical" world?
We know the answer to this. It is possible to make software that has very low defect rates -- among other things, you have teams of programmers intensively review every line of code -- but these practices have drastic consequences: projects become massively more expensive, development slows to a crawl, and innovation is greatly restricted. There are only a few fields where those tradeoffs are worth it. Elsewhere, they aren't close to being economic. The net benefit of software to society would be crippled if we built it this way. Of course we never would, because any software company trying to would be out-competed into oblivion.
As for Dropbox, when I see programmers jump all over other programmers for making a mistake, even a big mistake (or series of mistakes compounded), I think schadenfreude. People who engage in such gleeful condemnation are making an implicit claim to their own perfection. I'd think twice about doing that.