So if a derail is non-legit, you could enter false data, clearly marking it as false, rather than letting it derail and telling Beeminder it was non-legit.
This has the advantage that it doesn’t mess up your rhythm, so you can go back to the task the next day. If you, like me, get mainly spurred into action by beemergencies, the current system will have you waiting around a couple days before you see the beemergency for that task again.
This may be the best solution until this issue is fixed: Request: hard commit to allow 0 mercy days
But, this makes your graphs inaccurate; you have to look through the data to see which is real and which isn’t.
Though this problem can be solved by regularly going through and deleting all “fake” datapoints, then emailing Beeminder to tell it the derail was not legit. That way you’re only out once total, instead of once for each derailment.
What are people’s thoughts?