I think you could dream up some one-hour tasks if you really wanted to. The issue is that they'd be so obviously unrelated to overall job performance that the process would be self-evidently silly.

I'd argue that software is similar: projects don't falter because it takes someone five extra minutes to run a gel (or implement a red-black tree); they fall apart because people don't think about whether they ought to be doing that at all.