I was in an awkward position a few weeks ago when I tried to introduce a colleague of mine to Academia Stack Exchange. This colleague has a neuro/muscular disorder that makes typing quite difficult. Occasionally she will hit the wrong key (typically the largest keys like enter and the spacebar), but most of the time she is just slower to type than a fully able-bodied person like myself and, statistically speaking, you the reader.
Whilst introducing her to the site, she prematurely posted a comment on the chat after accidentally hitting enter. When she tried to edit the comment, she typed about 40 words, hit enter, and was told the comment was lost to the abyss because she was too slow. The ability to edit the comment timed-out before she could finish, and the comment she spent 1 minute writing was gone for good. She was then unable to delete her half-written message, also due to the time-out.
I understand the rational behind the time-outs, however I must say that their timeframes are particularly short. Clearly the site recognises the need for quick-edits. Clearly those quick-edits need to be done.. quickly. But the definition of "quickly" is very different for someone like myself, and for someone like my colleague. Perhaps the cool-down period should be extended from what it currently is, or perhaps even better, once someone starts the process of editing their message (which you can only do within a reasonable timeframe), the editing itself can last indefinitely so long as the textinput is being updated once every 30 seconds.
I'm aware that such rules could be "exploited" programatically, however I see the risk-reward ratio being significantly in favour of reward.