Removing favourites?!?

Hey all,

PLEASE offer your input over on github.
We REALLY need options, not just things being yeeted into a hole.

(We were asked for feedback btw)

1 Like

Clicking a little star icon is difficult to manage :rofl:

In all seriousness my UX sense says an MRU is going to be an even worse pain in the ass. Consider that pie menus are empirically proven to be a good pattern because the gesture a user does to use one while reading it is the same when not reading it–they develop muscle memory to direct the program. Now consider than an MRU by necessity moves items around on each use which is antithetical to building familiarity.

To make the MRU concept palatable you would want to store a set S of every module that is installed. Instancing a module places it at the front of the set. Then \text{take}(S, n) and use some kind of stable sort when presenting them to the user. This way the most n used modules are shown on the front page, but they do not jitter around the screen so much.

An even better extension would possibly be a day-to-day journal, where any module that gets used on a given day has its score increased and modules not found in the previous day’s journal is decreased (winnowing.) This set Sj would be sorted by score, have n modules taken from the front [\text{take}(Sj, n)], and then a stable set so it also doesn’t jitter around. This only requires at most one day’s journal and at most one entry per module actually used on that day, since winnow scores are cumulative.

The second option I would assume learn the modules a person tends to actually use across their sessions, be resistant to “trying this one module out” but still adapt to genuine changes in preferences over time. I know of nobody who actually does this, but it’s how I would do it (if I were doing this.)

(On the “never going to happen” sidebar: I would be very curious to know what would happen if small scale GloVE or skip gram embeddings were used in a module browser like this. I suspect modules are added in particular groupings more than thought, and such a model would be able to predict “you just added a VCA, perhaps you will also be adding an ADSR soon.”)

3 Likes