My frustrations with Skritter

Hallo everyone,

first let me say this is a long post to express my emotions. I do not plan to include a TLDR and this is mostly directed at the lovely skritter team and other members of the community who are interested in my experience. Please do not skip parts as for I included them for a reason.

Preamble

First, i would like to express my sincere support for Skritter and what Skritter is trying to achieve. I am a great fan of their mission and consider it to be the best tool for learning Chinese character based languages. I am currently paying including myself for 2 people and have recommended Skritter to all my friends interested in learning Chinese or Japanese as well as many strangers I had discussions about the same. Many of a friends tried Skritter out to much rejoice, but were ultimately not able to afford it.

In detail, when learning a language, the learner has to juggle time, concentration, learning material and motivation. Traditionally learning was book based which lead the learner to have manage all three of these completely manually. However, with the advent of computers many of these things could become automated and ultimately give us the luxury not to be able to learn much more efficiently than our forbearers. So now we do not have to juggle these four as much any more: Learning material is available digitally, requiring no work expenditure to copy, Spatial reputation learning allows us to only spend n*log(n) time for learning concepts, and splitting up learned materials into the smallest possible chunks allows us to use our brains automatisation to speed learning and make it less reliable on concentration; it has become easy to enter ‘the flow’. What is left is motivation, but even it has become addressed: Gamification, daily notification reminders, and automatic adding of new information to make learning independent from motivational swings. All this is great for learning anything in-specific. The free software Anki allows anyone to do just that. However, Anki is insufficient for learning anything specific:

Let’s take Japanese as an example. As said before the trick to automatised learning is splitting information into smallest chunks. There are language-general chunks: Grammar, Sentence-production, Sentence-comprehension, pronunciation, word-listening, listening-comprehension, vocabulary-meaning, vocabulary-usage, word-roots, script. Also there are Japanese specific chunks: Kanji-onyomi-reading, Kanji-kunyomi-reading, Kanji-meaning, Kanji-recognition, Kanji-writing, intonation. Also because I want to make a point about it later: for Chinese there are obviously the Chinese specific chunk: the tones.

Of these there are some where we already have software solutions for the following:

  • Wanikani/Skritter: vocabulary-meaning, Kanji-onyomi-reading, Kanji-kunyomi-reading, Kanji-meaning, Kanji-recognition
  • Bunpro: Grammar, Sentence-production

Now there are a couple we do not have automated solutions for yet. Yes there are of course ways to learn these, but not while having software automatically track your progress. For some of them it is possible to make our own content and use Anki to track the progress for us. However that requires us to make judgements about how we were actually doing, collect or create the learning material, which makes it suboptimal:

  • possible with anki: word-listening, listening-comprehension, vocabulary-usage, word-roots, script
  • not possible: pronunciation, intonation, tone-producing

The problem with free solutions like Anki is however, that the collection of data has to be come individually and often is cumbersome. Many people fall back to taking the first image of a search engine or using TTS to generate to pronunciation. Not very optimal for learning. I for example have voice recorded all of A1 and A2 german vocabulary in addition to 5 example sentences each, but you can only do that for your native languages not for the one you learn. Worse, anki requires you to judge yourself how well you did. That requires a lot of dedication and concentration to accurately judge how well you recalled the word and not cheat yourself. And even if you dont cheat you still have to interrupt your flow and check your own answer, you cannot just concentrate on producing, you also have to mark. This is were Skritter shines:

First it offers native pronunciations for all the common words and hanzi and it also offers example sentences with native pronunciation. So instead of first creating the material, one can focus on learning.

And now the thing that makes skritter outstanding: Handwriting judgement. Judging if your handwriting is correct takes a lot of effort. Stroke order, relative position of strokes to each other, stroke direction. This is one of the most cumbersome tasks when learning Chinese and Japanese, and thus often ignored by learners. However merely using pattern recognition training never gets you to perfect Hanzi/Kanji recognition, for that writing is needed. So even if one were to not care about writing, writing is a must also for recognition. From my experience, even with Skritters imperfections, I can do 5 to 10 times more writing-trainings in a given time span than before i found Skritter. This is amazing.

Further for tones, Skritter also optimises this to be as low effort as possible. Keybindings make it so that i can dash through 60 words in under a minute. That is great.

So to summarize: Skritter has three strong suits: Great quality material, writing recognition, and optimisations that make learning take less time.

My Criticism

I hope what i wrote in the preamble illustrates believable that I am a big fan of Skritter. I am sure many have seen that I have somewhat spammed this forum with Bug reports, and I hope they can forgive me. Whenever I find a bug I try my best to find out as much information as possible about the bug and report all of it together in a report from what I believe to be one holistic bug. I am also aware that I am not always able to give a step by step procedure for reproducing a bug, but that only happens if the bug up to my current research appears to be happening at random times. Also I sometimes happen to think I already report a bug or let it annoy me for too long before I report it. I must apologise for that, this is clearly on me and I should report bugs before they make me emotionally charged. I will do so in the future. Also I have used Skritters alternatives for reporting bugs. The Helpdesk thing broke for me and my previous reports were gone, which was quite disappointing, that’s why I switched to the forum in the first place. Also I prefer not to use email for reporting bugs as I want all my reports to be in one organised place and this is harder to achieve with emails.

Still, I hope to be able to expect bugs to be fixed in due time and that they end up fixed properly and not just by a workaround that sometimes is arguably worse. I have helped report many bugs during the advent of the studying websites Bunpro and Kitsune and these being one-man projects have all fixed my bugs in under a week worst under a month. Here I have filled quite a lot of bugs, and many have been fixed quite quickly. However there have been some bugs that I have reported month ago and that get me every single day that are still up to be fixed:

Links to bug reports

Skritter Firefox/Win10 pen input as mouse input bug
Windows Ink Input does not work
[Urgend] skritter gets stuck
Study Kana settings turns itself on - #4
As well as this one that i then posted as a feature request but it is most definitely a bug.
[Feature Request] preload cards early

I personally do not understand why Skritter only loads vocab in certain block sized and then loads more when i finished learning them. Sometimes making me wait for up to a minute just so can dash though the next bunch in less then a minute.

I do not understand why Skritter only saves progress every so often, so every time a bug breaks Skritter so that the save button no longer works requiring me to redo the vocab that was not saved.

I do not understand why my settings get messed up every now and then when i switch between Japanese and Chinese.

All these would be fixable by ditching the website and using the App. But the app too has problems. The app does not have the option to type the reading and instead requiring self validation which costs time. The strength of Skritter of doing this for me is lost. I understand that phones are not suitable for that anyway. Using a hardware keyboard allows for much more accurate 10 finger writing than i would ever be possible on a touchscreen. I do not blame skritter for that at all, but I do think they should promote using the computer for the app as it is more efficient.

Also the app does not have the smooth learning by automatically adding new vocab during your reviews.

However the app too has bugs. I long while ago I reported that the app becomes desync sometimes when switching from the computer to the app resulting in the current learning session crashing and a long sync being performed. This was addressed by instead making me wait that full sync every time I open the app or finish a learning session. I hope it is understandable that this made using the app more frustrating not less.

Lastly, I really hoped for feature requests being implemented. For Japanese pitch accent is paramount for a native sounding accent and even like in chinese can make differences in meaning (雨 vs 飴, 蛸 vs 凧, 橋が vs 端が vs 箸が, and 読んだ vs 呼んだ). As i talked about above, there are no good solution currently available for learning it using software, but it is an essential part of the language essentially ignored by learning solutions.
(I also wish for the listening feature request, i thought i asked for it a year ago, but i cannot find any evidence now. maybe it was in the helpdesk, but idk so i cannot really ask for it now)

I am a developer myself. And i would be happy to help Skritter get bugless and add new features like these. In case you need any help be it more investigations into the bugs, coding or otherwise. please let me know.

Cheers,
sirati

Thanks for taking the time to write up such a detailed report, and for being a long-term Skritter user. Reading this it is clear that you care about the product and the mission-- and we appreciate the feedback, the bug reports, and the criticisms you have with the platform, and the product you’re paying to use every single day.

The underlying issue that is at the core of everything you’re mentioning right now is the fractured nature of Skritter–a company that has been around since 2008 and operates on a data layer just as old, which currently exists in various forms on the web, iOS, and Android.

Skritter isn’t a one-man team, but we are small relative to how this project has evolved over time, and we’re dealing with lots of moving parts across various platforms and code bases. We’re in a phase of restructuring and unification at the moment and this phase is slow going. This restructuring is taking a significant amount of time, but the end goal of this process is to actually give us the opportunity to move faster and grow as a company.

Some projects we’ve embarked up have taking far longer to achieve and implement than when we originally set out to do them. And on the surface, it probably feels like things are moving slowly, or perhaps that issues are being ignored. Honestly, I’m sorry that you’re feeling that way.

We try our best to determine just how critical bug fixes or features are in terms of their overall impact on all Skritter users, and also in the context of the grand unification. Right now, the reality is that the mobile project and building a new API layer are taking top priority.

To give some context to this statement, you have to realize that the Skritter userbase is essentially 9-1 on mobile vs. the web for daily active studying. For a small team trying to grow, the highest potential is on the App Store and Google Play Store and also bringing that experience to the website as quickly as possible.

The reality is that the Skritter website is a placeholder for the future. It generally works, and it does the job of serving up reviews, but it is also using our v1 API, which can be slow and unforgiving when things fetch, save, and fail.

Given that reality, it is hard to justify spending time rapidly fixing/improving the website if the bugs are not 100% breaking stuff when it will ultimately be replaced by a version of Skritter that over the past year has outperformed its predecessors in every single way.

Just know that we’re working every single day to tackle problems, and working toward making Skritter less error-prone, more powerful, and more cohesive regardless of the way you decide to do your studying.

-Jake

P.S. If things go according to vision, then we’ll certainly be looking to grow the size of the dev team in the future, and we love opening this process up to passionate Skritter users first :slight_smile:

1 Like

We’d like to help but as @SkritterJake has mentioned there are other factors we need to take into consideration. If you’d like to reach out directly to a developer to discuss specific things further you can send me an email at josh@skritter.com. You’ve hit on several issues that might be easier to discuss in a call. I can’t make promises that I’ll be able to address everything but we might be able to hash through some things more clearly and come to some kind of resolution.

1 Like

Hey Josh,
i have sent you the email you requested.

1 Like

Awesome! I’ll keep an eye out for it but don’t see it in my inbox yet.

Ah yes my bad. I forgot to trust my mailservers cert, now its sent