Thanks for letting us know! I suspect the Easy/Normal/Hard buttons are being handled differently like @morbrorper mentioned - we’ll check it out!
Thanks. That is great.
I have another suggestion if I may.
If I get it right, your are currently picking the fuzz from a set list of 3 options: -5%, 0, +5%.
Probabilistically, that will split a stack of identically scheduled review into 3 equal-ish stacks.
If you are programming in Python, importing the randint() function from the random module allows a much better dispersion. The randint() function returns a random integer from the specified range, inclusive on both ends.
fuzz = ceil(0.05 * interval)
interval += random.randint(-fuzz, fuzz)
Therefore, if the interval is 100, you’ll get an integer range of -5 to +5, so 11 possible outcomes, and a much better dispersion of the initial stack into 11 stacks instead of 3.
A better cluster buster.
Thanks for all the input! A fix is up for Easy/Normal/Hard control next reviews. Please let us know if you’re still not seeing any fuzzy next review dates. ![]()
If you’re willing to share, how frequently do you look at the next review dates? Why / what do you do with that information? I tend to not pay attention to them too much, especially as they get into larger intervals.
Agreed! I was thinking the same thing - the Obsidian algorithm random_choice({ -fuzz, 0, +fuzz }) suggested picking one of the 3. It wasn’t clear why, but figured they must have a good reason and thought we could try something fuzzier in a next iteration. That next iteration is now up
The fuzz should now be applied in a range from -5% to +5% like you suggested (so more possible outcomes / dispersion).
Maybe not so often, after all. It’s only now that we’re trying to evaluate the fuzzing that the exact date is interesting. There are other things that are more important. ![]()
All seems to work great now! Congrats and many thanks.
Like @morbrorper, I’d prefer to see the actual future review date, which means, by the way, that when I face the Easy/Normal/Hard choice, I would see the 3 possible (color-coded?) date outcomes.
