In the real world, when you are writing a character and forget a stroke, then you have forgotten how to write that character. In Skritter, you click the middle for a hint and it is still correct. You can get 2 hints maybe more and it is still correct. Is there any way clicking the hint button will automatically mark it wrong? (This would help save time from clicking the middle a few times or dragging it to incorrect and wasting time marking new words and review words wrong that should be wrong in your mind)

Also, am I supposed to let these words pass as correct? If I don't spend the extra time deciding and marking the word as wrong it will pass automatically. For example, if a word I have not seen in a while comes up and I forgot a stroke or two, it will pass and the next time I see it may be over a year. Then in the real world, I will not know how to write it without the extra Skritter hint or two.

BTW, this is in regards to the iOS app.

Well I wouldn't say "huge" waste, but I would definitely like to improve my study efficiency.

Also, when double clicking the middle to mark it wrong, you expose the whole character. Sometimes on a complicated 20-stroke character you might have trouble remembering the fifth stroke only and don't want to see the rest because you want to make sure you can recall them from memory.

Thus, a great solution would be to let us have a "strict" setting where clicking for just one hint will automatically mark the word as wrong.

I hope this can be taken seriously as many learners feel the way I do. Now that I think about it, the title of this post should be changed to "Words that should automatically be marked as wrong".

Skritter guys: Please think about it. These words should not pass. What if you are taking a test and forget a stroke. Do you think the teacher should mark you as right since you "almost" got it, or mark you wrong so that you will take special attention to it? Thanks!

Hi Malaili, it seems there are other people as well who feel this way. When you mentioned this in the past, what was the explanation given?

On a side note. I think this has do with the retention rate. If everyone was so strict with themselves, then the 95% advertised rate would drop. Mines is around 80%. I will mark it wrong sometimes if the stroke is in the wrong position.

To offer an opposing viewpoint, I often have to mark characters correct that Skritter marked wrong.

I often use Skritter on the train, and the train sometimes will accelerate or decelerate suddenly, causing whatever stroke I am currently writing to be written incorrectly, even though I do in fact know how to write it.

Hence, given that where a Skritterer happens to be will affect the accuracy of what is intended vs. what actually gets written, making Skritter stricter should be an easily accessed option.

You would still get a couple tries on a stroke before counted wrong. I think I know what you mean. Only when you really do not know the next stroke and click the hint for what the next stroke would be is when you would be marked wrong.

So a strictness setting for grading (based on hint clicks and/or unregistered strokes) on a future version would be the best solution.

I've also mentioned this several times. I have to mark it incorrect so often. I have started to try and get in the habit of double tapping if I don't know, this marks it incorrect but reveals the complete character which is not ideal. I think it might be about retention, if your studying all parts and are strict with grading, I don't believe anyone would get 95%. I think mine is 85% and that's not being nearly as strict as I want.

For me this issue is the most frustrating problem with skritter.

Since the iOS version doesn't have hot keys like the web version does it seems like the best solution would be to add another option to adjust how quickly it marks things incorrect.

I do agree if you swipe 2 times and nothing registers it's something that should probably mark wrong. A hint means you don't know how to write a character flawlessly so at the very least should be marked so-so, but realistically is wrong.

The perfect solution would be just to have a scrolling degree of strictness, so the people who want the hints and chances can have them and the others not. I would assume that is not very difficult to implement, but I don't know perhaps its difficult to do.

Hi Skritter guys, it looks like we have a consensus. Let's put all other projects on the backburner and implement this ASAP for all of us beloved paying customers. :D

1. If there is too much coding involved with a strictness setting, then just make the "hint click" an automatic wrong. I wonder who was the first to want two free sequencial hints on an unknown word without a wrong penalty anyways?
2. Regarding the 95% retention rate mentioned in the Skritter commercial. Let's just relax on that and bring it down to 90%. The new customers will not run away.

To Skritter guys: Thanks for the help on this! So can we get a timeline on when this change can be implemented? If not, an explanation would suffice on why Skritter was designed this way. Then I am sure more people will start chiming in. You cannot ignore us forever…

I don’t think there would be consensus on this (but I suppose we are both guessing) – although if done on an optional basis would be fine of course. I for one spend more time changing answers that are marked wrong to right, and if 90%+ are known in any review, and "mistakes" are going to go both ways, I definitely want the system to err towards marking things right rather than wrong.

For me, the hints reflect that most of us (I think) aim to type and read – writing on skritter is a way to remember the characters (and distinguish between characters), but at the end of the day, handwriting in china is almost always semi-cursive or cursive so knowing how to write thousands of characters by hand doesn’t give so much assistance in true (hand-) reading/writing (as opposed to computer produced characters).

In any event, raw squigs mode is designed for those who wish to write by hand (or, more accurately, print characters by hand) – as in real life you would not get confirmation that any stroke was correct before you had to move on to the next one and finish the character.

Regarding retention rate - there is an option to change that, you can set it at 90% yourself.

Maybe it would be best to have a button on the interface to immediately bring up the grading prompt to mark something wrong or right. In my case, depending on what it is, I have to often change the grading to correct when it was marked wrong, or vice versa.

Tendency to be marked wrong, even if I know:
When I write on the train, if the train suddenly accelerates or decelerates, I often bump the screen, sometimes in rapid succession, which gets interpreted by Skritter as a hint and then an incorrect answer. I also have a tendency to mess up a lot of strokes when I use Skritter one-handed or while walking (such as when switching train lines). In those cases, I have to change the grading to correct.

Tendency to be marked correct, even if I am not:
In a lot of cases I make a minor mistake (the ones that bring up a little popup like "should hook"), Skritter accepts it as correct, but I have to go back and mark it wrong. That could be another place where an option could be provided to have stricter grading.

As background, I have raw squigs enabled and set the retention rate to 97% (unfortunately the the highest it will go), but my retention rate tends to be in the lower to mid 80 percent range.

Ricksh - I think the argument about retention rate is that because Skritter gives hints, the apparent retention rate across all Skritter users is 95%, even though if you gave most users a blank piece of paper to write the characters with, they would not remember the characters completely correctly 95% or more of the time. There is also the fact that if you write a stroke without a hook (when there should be one) or vice versa, it often gets counted as correct when it should be marked wrong.

Hi Ricksh,

I think there is a slight misunderstanding here? Talking only about what words should be automatically marked as wrong…(There are two things that are going on). #1 is when you make either a guess stroke or known stroke in Skritter and it is not registered and #2 is when you do not know or forgot the next stroke and click a hint for Skritter to tell you the next stroke.

Regarding #2, when learning new words and reviewing almost forgotten ones, Skritter often times marks words as correct that should be wrong. It is because Skritter gives you atleast 2 hints before marking you "so-so"", then it will mark you wrong on the 3rd hint you ask for. What we feel is that this is a big waste of time since we OFTEN have to go back and change the grading for these cards which probably more than 30% of all our reviews. I am curious and would like to ask what you do for these cards that you got 1～3 hints for? Do you let them pass as correct/so-so because you feel that you should have a chance or do you go back and change them to wrong because you feel the card is not learnt yet?

So let's say Skritter decides to implement a strctness feature at some point in the future. What would help in your case for "words that should automatically be marked as correct"? which is opposite of what I am talking about.

Edit: Aside to greenteapanda, I agree on your point about hooks and alternative possible solution. After a card is done and Skritter has automatically graded your card, then a quick click on the bottom to alter the correct grading would also work. That is, when you are waiting for the next word to popup. As it is in the iOS app currently, you have to drag it to the desired grading correction. If the buttons are all there (w/o needing to drag) then a simple click can change the grading.

I think anything that gives more options is fine - greenteapanda sets his retention rate at 97%, I set mine at 87%, so that's an example where people want different things ...

Regarding #2, I will let some 10/15 stroke characters pass with a couple of hints, as I am more focussed on recognition of a large number of characters, rather than perfection in hand-writing a smaller number. Plus skritter cannot allow for the fact that I will see characters "in the wild" so is always overly conservative on showing characters. So, as a simple example, a one or two stroke hint for 糖 would be fine as I will likely see it in the wild, will always recognise it, and am unlikely to need to write by hand from memory (e.g. on a form). It's the same as in english I won't sweat it if I can't write (spell) Schwarzenegger exactly right but do want to be able to write (spell) California even though California is equally not an English language word and is a one-use word.

As far as I'm concerned the 95% and learning an item every 54 seconds is just skritter advertising puff (interestingly on the same page as the 95% figure a 90.2% rate is mentioned - see http://www.skritter.cn/learn_more ) . Who can blame them, another website promises to teach 2289 characters in 3 months (100% retention??!! 2289 in 3 months - really, where did I go wrong?), and the book stores are full of learn chinese in 15 minutes a day for 3 month books.

Basically, I see your point of view, and the only bit we differ on is when you say "If there is too much coding involved with a strictness setting, then just make the "hint click" an automatic wrong" - I'd prefer a strictness setting, and no change if that is not possible.

Re. your edit, while I don't mind, I am sometimes surprised that e.g. doing a hook the wrong way is accepted, so I see that point too. And I fully agree that in the ios app shouldn't have to drag out the grading options - in addition to correcting "wrong" gradings, I like to use "so-so" grading too on some that go red/green.

If you don't want to know how to write, you just want to learn the definitions, why don't you just turn those reviews off?

I think the problem with the hooks is that they have to be written very accurately to be recognised or even exaggerated. Therefore it is very difficult for skritter to distinguish. (That is the only area I feel skritter thinks I have written incorrectly even when I do the hooks.

With the hint where the shadow of a stroke is shown, that should definitely be instantly marked incorrect for sure. With regards to giving chances I would prefer to have no chances at all, as I rarely struggle with skritter marking something incorrect that I felt was correct.

I would be more than happy to give this a try with a really strict setting enabled to see if it works or not.

Richsh, I don't know what country your in where the trains bump you around so much, but if its that bad I would imagine its just not the right time to do writing prompts on skritter.

Hi 夏普本,

Others may be able to, but I don't think I could learn a lot of characters for recognition as pictures or whatever, so writing for me is essential to emphasize the small differences between them. Sure, you can learn recognition as pictures for 1000 most common or whatever, plenty of people have, but I think each 100 or 1000 more characters would become increasing more difficult. That's what skritter is about for me.

In an ideal world, I will learn to write every character without hints, but for me that would be a secondary goal for after learning to recognise the 现代汉语通用字表 7000 or 现代汉语常用字表 3500 or wherever my characters etc. mean I can read what I want to, plus type what I want. Plus after focus on traditional reading/writing (or my version of writing!). It's a high goal though - many chinese miss a stroke or two on difficult characters, as they moved on from "printed" character styles at school to various cursive styles after school.

Is the demand for 100% strictness/no hints as some university courses are requiring handwriting or is it useful for your daily life? I thought it was a sign of the times that HSK went computer based (i.e. pinyin plus recognition required rather than handwriting)...

As to strictness settings, as mentioned above, sounds good.

I don't skritter on the train - that's greenteapanda - but I would if I used the train!

Yes, I am the one in this thread who uses Skritter on the train. I live in Japan, so my Skritter accuracy is affected by being pressed on all sides by other passengers during rush hour (so I hold my iPhone above my head for some space) and the fact that Japanese trains are all wheel drive, so they can accelerate and decelerate a lot more quickly than trains in most other places.

If I were sitting down and not at the whims of the (usually involuntary) movement of other passengers, then I could certainly be more accurate with Skritter on the train here, but that is usually not the case.

Anyhow, the expectation of accuracy (and according relevance to daily life) seems a lot stronger in Japan than in China. I wrote cursive characters all the time in China, they all got marked wrong here until I unlearned that tendency. My Chinese classmates make lots of mistakes with Kanji, including the ones that are the same as the Hanzi in Chinese.

In Japan, one of the ways job candidates are weeded out are by forcing them to write resumes by hand (even if you found the job online and submitted information electronically initially). Writing ability is also required for the Japanese University Entrance exam (EJU) amongst others.

Even if I did not need to write, I learn my doing, not just by looking, so character writing practice helps improve my reading ability.

I would like to propose a "grading strictness" setting to Skritter and would like to get everyone's thoughts. Currently, you get about 4 stroke tries (where you enter a stroke from either your memory or as a guess) before being marked as "so-so" on the fifth try and about 2 free hints (when you ask what is the next stroke) before being marked as "so-so" on the third hint. I think this changes depending on how well you know the word or how many total strokes there are (can someone explain this to me?). Thus, I would like to call the current setting "easy" in my proposed "grading strictness" settings as follows. Any comments are appreciated.

Setting #1 (Easy): 4 stroke tries and 2 hints before being marked so-so.
Setting #2 (Normal): 3 stroke tries and 1 hints before being marked so-so.
Setting #3 (Strict): 2 stroke tries before marked so-so and on the first hint, you are marked so-so. If you missed a hook you are automatically marked so-so.
Setting #4 (Very Strict): 1 stroke try before being marked as so-so and on the first hint you ask for, you are marked as wrong. If you missed a hook you are automatically marked as wrong.

Or better yet, the stroke tries, hints, and whether hooks should count as wrong can be individually set and tweaked per user need! This would make us all very happy.

I agree that writing is useful in many ways. My classmates think I am crazy for spending so much time on writing, they think it's a waste of time.
I can see the argument that writing is rarely used. I also rarely hand write English. But it is a huge help in reading as well.

Hugh723 the amount of hints is relative to the amount of strokes. I think also if you write a stroke out of order or positioned in the wrong area of the screen it maybe does not count as a mistake straight away. The majority of characters I get wrong, I do actually know and once I see the position of the first one or two strokes, I know how the rest should look.

I have stopped using so-so after I heard it was marked "correct". For me it's wrong or right.

Well, if the stroke is out of position or order and it is accepted by Skritter, then there would be no change from the current settings. Only when you make a guess stroke and it is wrong. You know, when you make a swipe and nothing happens. Currently, you get about 4 of these swipes before counted ad so-so. On my proposed "very strict" setting, you could make one swipe and miss and it would not affect your grading. If you swipe again and miss, then the word is counted as "so-so".

I do realize Skritter's handwriting recognition has flaws so ideally this number os swipes or "stroke tries" could be configurable along with the amount of hints you get before your grading is affected.

I wouldn't even want that one hint to be honest. One mistake means I couldn't write it in a real life situation and for me that means I don't know it. It really is so rare that skritter doesn't recognise a stroke correctly that I definitely knew.

What I was saying is when you make a swipe on the screen. Somrtimes you need a chance before the stroke is recognized if you get bumped or something. Not sure what they are called but I called them stroke tries or swipes.

But any hints asked for would automatically mark you as wrong on the strictest setting.

I dare say this is a lot of discussion for something you can quickly mark one way or the other when you're finished writing the character.

You don't even have to hold the button and drag to the right one. Just click on the check mark and it will change to wrong... you only have to click & drag if you want "so-so" or "easy".

It's just annoying having to change it so often.

The reason it works the way it does is because we did a lot of tweaking of the number of wrong strokes and hints allowed until we satisfied the largest number of users in our testing when we actively polled people. One problem with trying to get consensus on a forum thread like this is that most users who are satisfied don't post, so you get more apparent agreement for changing things from the dissatisfied users who do post. I'm not saying that the current thresholds are perfect, just that they're hard to improve without doing much more testing than just discussing here.

As mentioned above, a more complete way would be to add an extra preference for how many wrong squigs and hints are allowed, so that you could drag a slider and make it work the way you want. However, we are loathe to add an extra preference, especially when from our point of view, we already made it very fast to toggle correct/incorrect by tapping the button in the lower left after you've finished writing a character.

Extra preferences are very expensive in non-obvious ways. They make it harder for users to find the other preferences they need. They take up space in the UI. They necessitate maintenance and make it harder to add support for new platforms. They make it more difficult to track down problems, because you're multiplying the number of possible configurations. They require explaining.

If a preference isn't going to be useful to a large fraction of users, then we try to keep things simpler.