Getting awfully repetitive

Suddenly, Skritter Chinese on Android seems to be stuck only asking me about very recently added words. I’m getting a lot of the words I added today over and over and not words I added yesterday, which I’d really like to be tested on since I don’t know them all that well yet. Occasionally older words will appear but not as often as I’d expect.

This seems to be a weird quirk of the SRS algorithm. The timings say that I’m caught up on those words from yesterday but apparently in absence of something new to ask about, I just keep seeing the most recent things added. I would expect that having gone through those a bunch, the app would start picking older words even though I might not be scheduled to see them until later.

Adding a new word didn’t seem to shake things up and get other words tested.

I post this because it feels like something changed from the behavior I’m used to seeing where it wouldn’t be so fixated on new words.

Also… when are we going to see a new Android beta version? I last asked about this in December and the reply was ‘a few weeks’.

One of the major focuses is on study loops and the SRS, we’re not quite happy where we’re at yet though so we’re not ready to push the update quite yet.

It’s been awhile as it’s a fairly major update. You can absolutely expect it soon. Be on the lookout!

I have the exact opposite problem. I keep adding new words but none of them get reviews. I just keep getting the same words I added the first day or two over and over. I’m using version 2.3.5 on android.

1 Like

@kokaku @stenaf Our 2.3.9 website update will include some code that targets study loop issues and it should be out today or tomorrow (doing last round of testing now). You might want to give it a go once things are live to see if the queue is scheduling things better for you. Even if you don’t usually use the web, it’ll be a good indicator of the updates we’re working on.

I’m really hoping that the new mobile beta will also address the issues you’re both bring up. We’re handling the queue on the client for most of the interactions now, and the studying I have been doing on my personal account and lots of test accounts is working better than any previous versions of Skitter beta (and perhaps legacy and iOS, too).

We’re just putting some extra elements back into the study screen again before we release this thing. We have a lot of projects going on right now, but @josh and I are doing everything we can to get this app out before the end of May.

it seems like the update happened because I’m seeing a lot more words to review in the Android app than I have in recent days (unless that’s just coincidental timing with a wave of reviews in the app)

finally things were feeling just right - a good pace to review old words and add new ones - and now it feels too much - i suspect every single user has a slightly different need for pacing - it would be amazing if you could somehow expose the parameters of the spacing algorithm at finer detail within the app so that every user could fine tune them specifically to their own way of studying (something finer grained than the high level retention rate setting)

1 Like

also… I’m curious about a pattern I’m seeing - it seems like when words are introduced, all the parts (drawing, meaning, tone, pinyin) are being tested in closer succession, including simplified and traditional variants

i’m wondering if this is intentional and has been seen to increase retention

Sorry for making you wait on a reply!

@SkritterMichael has been making some changes to the code for how things are displayed in the queue when item due values are similar. We’re trying some new things out with 2.3.9 updates, and have some additional scheduling adjustments for 2.3.10 that we’re testing internally at the moment.

How are you feeling about the changes thus far? Obviously we don’t wanna get too extreme with study queue and scheduling changes, but we do think that there is room from improvement and feedback is a great way to keep tuning and improving.

Let’s be honest: the Skritter interface is fantastic, but the more you use the program, the less you trust its algorithms for the study queue. It might be improved in Android Beta 3.X because I’m now suddenly seeing things I haven’t seen for months, but 3.03 had a terrible habit of fixating on one or two words and repeating them over and over. So far, 3.04 doesn’t seem as bad.

My biggest beef with Skritter is that, unlike Anki, it gives you no clue when an item will show up again. All the program says is “trust me.” And then the program has one little bug after another, so you end up not trusting it at all. The program is very frustration for its inability to deliver on its promise. So close and yet so far.


We hear ya.

2.3.8, 2.3.9, 2.3.10 updates are all trying to target some of that on the web. Mobile beta has taken huge steps to address this as well. And we’ll continue to improve it.

I’d love to see a due value below the grading buttons. First priority, however, is making sure as many users as possible can study on website and apps without issue. A tall order at times with accounts being so vastly different in nature, and with so many devices on the market. Once we have the app working and stable we can focus on improving the UX/UI so you not only trust the system, but also understand exactly what it’s doing.

1 Like