Word scheduling is useless 3


#1

Just dropping by to ask if there is any progress on this…


#2

I still find Skritter to be better than the alternatives, but I agree that scheduling improvements are needed. I’m afraid that if the release cycle is going to be 6+ months then any improvement in the next release might just be offset by new quirks that will take another 6+ months to resolve.

Meanwhile, I have been trying to game the system to get better results:

  • I only use the web version (even from my phone)
  • If it gets “stuck” (which usually only happens when I’m near 0 reviews, which lately I have not been) then I just use the advance arrow to skip the unwanted reviews.
  • I’m using the “easy” button more often to try to de-prioritize things that are showing up too often. And having read the API docs I now realize you need “easy” on all characters in a word to really get the effect.
  • I made a list of words clustered by repeated characters and components: Chinese word list ordered by frequency and clustered by similarity . The idea is to trick Skritter into showing me new characters more often at first, because its lowest interval is not usually low enough for me.
  • I have been experimenting with using the API to extract my leeches so I can apply alternate learning methods. One idea I haven’t tried it re-uploading them to a custom list that I can focus study with Skritter to limit the number of ways the scheduler can screw me up.

For now it’s working well enough that I haven’t just rewritten Skritter…


#3

What? Skritter has an API? I’ll definitely take a look at it. At this point it seems to me that writing my own alternative to Skritter is a faster option than waiting until it’s fixed.


#4

@nirangwo Yes there is an API which can be access here: https://www.skritter.com/api/v0/docs

We have tried multiple times to tame the system that is the legacy scheduling system (SRS), but have not yet been able to do so in a way that works how all users intend. Which is both frustrating for users and us. We have been working on some more aggressive changes to the mobile apps that aim to solve this issue in a most testable way.


#5

It’s obvious that the actual fix will take a very long time. For the time being, it would be enough to add a very simple study mode where items are chosen randomly (with a uniform distribution) from a set of sections selected by the user at the start of each session.


#6

This topic was automatically closed 30 days after the last reply. New replies are no longer allowed.