Another long-time user here. I packed my old iPad in the hospital bag when my first was born, thinking (ridiculously, as I look back) I might have some downtime to study. Now he’s in first grade and learning characters himself, and his brother isn’t far behind. Let me put in another STRONG request to include the auto-add functionality in the new app. Or, at the very least, not to disable access to the legacy app, which, while not exactly pretty anymore, has worked well for me for many years.
The particular aspect of Skritter that I’ve found most effective is the spaced repetition at the very early intervals, right when a new item is making the transition from short term to long term memory. I see a new item, give a bit of thought to it (not much), then mark it wrong. The next card then comes up, which is a true “review”; i.e., I have about a 90% chance of getting it right, but it still requires some effort to recall. The effort of recalling a few “review” items has the effect of pushing the new item just to the fringe of my short term memory. After about 30 seconds or so of this, the new word shows up again, and there’s an almost physical sensation where my brain was about to forget it, but something screams “WAIT, no, don’t hit delete, we need that!” The old algorithm hits this sweet spot beautifully; and it happens 4 or 5 times at progressively longer intervals until by the end of my commute, the new word is usually sufficiently fixed in my memory that the next day, it can serve in the “review” role in between the early, super short intervals of the next new words.
I think I basically understand the proposed work-around (marking as “learned” even though it is not actually “learned”), though I must say that isn’t intuitive to me. But that’s beside the point, because even with the work-around, the very short early intervals don’t seem to happen because “review” doesn’t update itself dynamically (i.e., you have to clear every review that was due at the beginning of the session before the app considers the dueness of things that you got wrong in that session.) A work-around I used for a bit while trying to give the new app a chance involved exiting the review session and starting a new one every 30 seconds or so to force it to recalculate. That’s extremely tedious, made more so by the fact that the time-spent counter/clock seems to have been removed from the interface. (Separately, I disagree with this decision, since my goals are based on time spent, having dispaired of regularly clearing everything that as due.)
I didn’t play with “review continuously,” as I didn’t see the setting before I read this thread. It does sound like that behaves in a similar way to restarting the session every 30 seconds (though is it able to recalculate without a network connection, like the old app did?), and coupled with a way to add and “mark as learned” in bulk, it might work reasonably tolerably.
Trying it now, I second the view that “review continuously” feels endless without a progress bar or due count. It would seem sensible to bring back the clock for “review continuously,” since there’s already an empty space at the top of the screen where the due count would appear in the default mode.
I think I now understand why these design decisions were made: “We see a clear distinction between the process of learning something for the first time, and reviewing it to help it stick in long-term memory.” If I’m understanding correctly, you mean there was a deliberate and conscious decision to remove spaced repetition from “learning,” and do SRS only for things only for things that have been learned by other means (“In-app videos, high-quality example sentences (with audio), grammar points, guided stroke order support.”) I think this decision is fundamentally wrongheaded. For me, removing spaced repetition from the process of learning completely eviscerates the utility of Skritter. Please reconsider.
A totally unrelated minor nitpick: in the old system, it used to be possible when doing writing practice to mark a word as wrong without marking any of its component characters wrong. There doesn’t seem to be a way to do that anymore, and I would miss that feature. But I can see how that one might be considered overly complicated.