Discussion:
Why don't anki or mnemosyne use newer SM versions?
Add Reply
n***@gmail.com
2018-04-19 04:48:39 UTC
Reply
Permalink
Raw Message
Hi everyone! I'm new here.

I've been using spaced repetition software for quite a while with the "repetitions" app, and later on switched to the new tinycards app from the creators of duolingo. However, the webapp limitation to have only 150 cards per collection is really frustrating me and I've been evaluating anki or mnemosyne as a new flashcard app.

I've noticed, though, that both anki and mnemosyne seem to use something called SM 2 algorithm to calculate the frequency of repetition. However, I've also seen that SuperMemo (creators of the sm algorithm?) works with a much newer version.

Upon doing some more research I came across this site that explains what has been improved with each iteration. https://www.supermemo.com/help/smalg.htm#Historic_note:_earlier_releases_of_the_algorithm

It all sounds good and the latest versions seem to improve the older versions in very convenient ways.

Now I'm also intrigued about using supermemo, but I still do like some featurese that mnemosyne has, like different types of cards so that I can add pronunciation to the words that I learn in other languages.

What's the reason behind sticking to version 2 of SM? I can't help but to feel like we're missing optimizations. I want to pick a new flashcard software ASAP but this makes it quite hard.

Cheers!
--
You received this message because you are subscribed to the Google Groups "mnemosyne-proj-users" group.
To unsubscribe from this group and stop receiving emails from it, send an email to mnemosyne-proj-users+***@googlegroups.com.
To post to this group, send email to mnemosyne-proj-***@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/mnemosyne-proj-users/f1e4733f-2522-4a17-81c6-4a388a2d5141%40googlegroups.com.
For more options, visit https://groups.google.com/d/optout.
Peter Bienstman
2018-04-19 07:03:57 UTC
Reply
Permalink
Raw Message
Hi,

Personally, I'm not convinced that later versions offer all that much improvement over older versions. Keep in mind that predicting intervals is a heuristic anyway, and not an exact science. Additionally, Supermemo is commercial software, so having a new scheduler each release is a nice commercial selling point :-)

Finally, Mnemosyne is built in a very modular fashion, so if somebody wants to write a plugin with a different scheduler, that's entirely possible.

Cheers,

Peter

-----Original Message-----
From: mnemosyne-proj-***@googlegroups.com <mnemosyne-proj-***@googlegroups.com> On Behalf Of ***@gmail.com
Sent: 19 April 2018 06:49
To: mnemosyne-proj-users <mnemosyne-proj-***@googlegroups.com>
Subject: [mnemosyne-proj-users] Why don't anki or mnemosyne use newer SM versions?

Hi everyone! I'm new here.

I've been using spaced repetition software for quite a while with the "repetitions" app, and later on switched to the new tinycards app from the creators of duolingo. However, the webapp limitation to have only 150 cards per collection is really frustrating me and I've been evaluating anki or mnemosyne as a new flashcard app.

I've noticed, though, that both anki and mnemosyne seem to use something called SM 2 algorithm to calculate the frequency of repetition. However, I've also seen that SuperMemo (creators of the sm algorithm?) works with a much newer version.

Upon doing some more research I came across this site that explains what has been improved with each iteration. https://www.supermemo.com/help/smalg.htm#Historic_note:_earlier_releases_of_the_algorithm

It all sounds good and the latest versions seem to improve the older versions in very convenient ways.

Now I'm also intrigued about using supermemo, but I still do like some featurese that mnemosyne has, like different types of cards so that I can add pronunciation to the words that I learn in other languages.

What's the reason behind sticking to version 2 of SM? I can't help but to feel like we're missing optimizations. I want to pick a new flashcard software ASAP but this makes it quite hard.

Cheers!
--
You received this message because you are subscribed to the Google Groups "mnemosyne-proj-users" group.
To unsubscribe from this group and stop receiving emails from it, send an email to mnemosyne-proj-users+***@googlegroups.com.
To post to this group, send email to mnemosyne-proj-***@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/mnemosyne-proj-users/f1e4733f-2522-4a17-81c6-4a388a2d5141%40googlegroups.com.
For more options, visit https://groups.google.com/d/optout.
--
You received this message because you are subscribed to the Google Groups "mnemosyne-proj-users" group.
To unsubscribe from this group and stop receiving emails from it, send an email to mnemosyne-proj-users+***@googlegroups.com.
To post to this group, send email to mnemosyne-proj-***@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/mnemosyne-proj-users/6f69e8f887a14df7b92da4147f6d57c6%40xmail102.UGent.be.
For more options, visit https://groups.google.com/d/optout.
n***@gmail.com
2018-04-19 14:32:05 UTC
Reply
Permalink
Raw Message
Post by Peter Bienstman
Hi,
Personally, I'm not convinced that later versions offer all that much improvement over older versions. Keep in mind that predicting intervals is a heuristic anyway, and not an exact science. Additionally, Supermemo is commercial software, so having a new scheduler each release is a nice commercial selling point :-)
Finally, Mnemosyne is built in a very modular fashion, so if somebody wants to write a plugin with a different scheduler, that's entirely possible.
Cheers,
Peter
-----Original Message-----
Sent: 19 April 2018 06:49
Subject: [mnemosyne-proj-users] Why don't anki or mnemosyne use newer SM versions?
Hi everyone! I'm new here.
I've been using spaced repetition software for quite a while with the "repetitions" app, and later on switched to the new tinycards app from the creators of duolingo. However, the webapp limitation to have only 150 cards per collection is really frustrating me and I've been evaluating anki or mnemosyne as a new flashcard app.
I've noticed, though, that both anki and mnemosyne seem to use something called SM 2 algorithm to calculate the frequency of repetition. However, I've also seen that SuperMemo (creators of the sm algorithm?) works with a much newer version.
Upon doing some more research I came across this site that explains what has been improved with each iteration. https://www.supermemo.com/help/smalg.htm#Historic_note:_earlier_releases_of_the_algorithm
It all sounds good and the latest versions seem to improve the older versions in very convenient ways.
Now I'm also intrigued about using supermemo, but I still do like some featurese that mnemosyne has, like different types of cards so that I can add pronunciation to the words that I learn in other languages.
What's the reason behind sticking to version 2 of SM? I can't help but to feel like we're missing optimizations. I want to pick a new flashcard software ASAP but this makes it quite hard.
Cheers!
--
You received this message because you are subscribed to the Google Groups "mnemosyne-proj-users" group.
To view this discussion on the web visit https://groups.google.com/d/msgid/mnemosyne-proj-users/f1e4733f-2522-4a17-81c6-4a388a2d5141%40googlegroups.com.
For more options, visit https://groups.google.com/d/optout.
Thanks for your response! and thanks for your work with the software, as well :)

Although I also thought of the commercial reasons, and I personally believe that sm2 is enough to learn stuff by itself, what I've seen that "worries" me a little is the difference in workload. Particularly about this part in the algorithms's comparison article:

"Feb 22, 2018: We have finally added a simulation of Algorithm SM-2 to SuperMemo for Windows and have come up with the average least squares metric of 53.5685% (for Algorithm SM-2). For comparison, Algorithm SM-17 results in 37.1202% (a million repetitions dataset). This may not sound impressive, however, for shorter intervals, the load of repetitions might easily be 2-10x greater assuming no delays (i.e. executing repetitions as prescribed). Back in 1989, we could see that even Algorithm SM-5 would reduce repetition loads twice as fast as SM-2.

For comparison, Algorithm SM-17 rarely goes 5pp above Algorithm SM-15 in R-Metric measure, and even small gains have a significant impact on workload (esp. at shorter intervals)."

link http://supermemopedia.com/index.php?title=SuperMemo_or_Anki

What do you think?
--
You received this message because you are subscribed to the Google Groups "mnemosyne-proj-users" group.
To unsubscribe from this group and stop receiving emails from it, send an email to mnemosyne-proj-users+***@googlegroups.com.
To post to this group, send email to mnemosyne-proj-***@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/mnemosyne-proj-users/f5514894-0888-4508-9e9a-b1eda26e0d57%40googlegroups.com.
For more options, visit https://groups.google.com/d/optout.
Peter Bienstman
2018-04-19 14:51:05 UTC
Reply
Permalink
Raw Message
Hi,

Workload is only one factor, there is also accuracy of recall :-)

I would suggest picking the tool you feel most comfortable with, based on algorithm, features and ease of use.

Kind regards,

Peter

-----Original Message-----
From: mnemosyne-proj-***@googlegroups.com <mnemosyne-proj-***@googlegroups.com> On Behalf Of ***@gmail.com
Sent: 19 April 2018 16:32
To: mnemosyne-proj-users <mnemosyne-proj-***@googlegroups.com>
Subject: Re: [mnemosyne-proj-users] Why don't anki or mnemosyne use newer SM versions?
Post by Peter Bienstman
Hi,
Personally, I'm not convinced that later versions offer all that much improvement over older versions. Keep in mind that predicting intervals is a heuristic anyway, and not an exact science. Additionally, Supermemo is commercial software, so having a new scheduler each release is a nice commercial selling point :-)
Finally, Mnemosyne is built in a very modular fashion, so if somebody wants to write a plugin with a different scheduler, that's entirely possible.
Cheers,
Peter
-----Original Message-----
Sent: 19 April 2018 06:49
Subject: [mnemosyne-proj-users] Why don't anki or mnemosyne use newer SM versions?
Hi everyone! I'm new here.
I've been using spaced repetition software for quite a while with the "repetitions" app, and later on switched to the new tinycards app from the creators of duolingo. However, the webapp limitation to have only 150 cards per collection is really frustrating me and I've been evaluating anki or mnemosyne as a new flashcard app.
I've noticed, though, that both anki and mnemosyne seem to use something called SM 2 algorithm to calculate the frequency of repetition. However, I've also seen that SuperMemo (creators of the sm algorithm?) works with a much newer version.
Upon doing some more research I came across this site that explains what has been improved with each iteration. https://www.supermemo.com/help/smalg.htm#Historic_note:_earlier_releases_of_the_algorithm
It all sounds good and the latest versions seem to improve the older versions in very convenient ways.
Now I'm also intrigued about using supermemo, but I still do like some featurese that mnemosyne has, like different types of cards so that I can add pronunciation to the words that I learn in other languages.
What's the reason behind sticking to version 2 of SM? I can't help but to feel like we're missing optimizations. I want to pick a new flashcard software ASAP but this makes it quite hard.
Cheers!
--
You received this message because you are subscribed to the Google Groups "mnemosyne-proj-users" group.
To view this discussion on the web visit https://groups.google.com/d/msgid/mnemosyne-proj-users/f1e4733f-2522-4a17-81c6-4a388a2d5141%40googlegroups.com.
For more options, visit https://groups.google.com/d/optout.
Thanks for your response! and thanks for your work with the software, as well :)

Although I also thought of the commercial reasons, and I personally believe that sm2 is enough to learn stuff by itself, what I've seen that "worries" me a little is the difference in workload. Particularly about this part in the algorithms's comparison article:

"Feb 22, 2018: We have finally added a simulation of Algorithm SM-2 to SuperMemo for Windows and have come up with the average least squares metric of 53.5685% (for Algorithm SM-2). For comparison, Algorithm SM-17 results in 37.1202% (a million repetitions dataset). This may not sound impressive, however, for shorter intervals, the load of repetitions might easily be 2-10x greater assuming no delays (i.e. executing repetitions as prescribed). Back in 1989, we could see that even Algorithm SM-5 would reduce repetition loads twice as fast as SM-2.

For comparison, Algorithm SM-17 rarely goes 5pp above Algorithm SM-15 in R-Metric measure, and even small gains have a significant impact on workload (esp. at shorter intervals)."

link http://supermemopedia.com/index.php?title=SuperMemo_or_Anki

What do you think?
--
You received this message because you are subscribed to the Google Groups "mnemosyne-proj-users" group.
To unsubscribe from this group and stop receiving emails from it, send an email to mnemosyne-proj-users+***@googlegroups.com.
To post to this group, send email to mnemosyne-proj-***@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/mnemosyne-proj-users/f5514894-0888-4508-9e9a-b1eda26e0d57%40googlegroups.com.
For more options, visit https://groups.google.com/d/optout.
--
You received this message because you are subscribed to the Google Groups "mnemosyne-proj-users" group.
To unsubscribe from this group and stop receiving emails from it, send an email to mnemosyne-proj-users+***@googlegroups.com.
To post to this group, send email to mnemosyne-proj-***@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/mnemosyne-proj-users/9d2ed397327842f7bb9113e450aec848%40xmail102.UGent.be.
For more options, visit https://groups.google.com/d/optout.
n***@gmail.com
2018-08-12 06:33:10 UTC
Reply
Permalink
Raw Message
Post by n***@gmail.com
Hi everyone! I'm new here.
I've been using spaced repetition software for quite a while with the "repetitions" app, and later on switched to the new tinycards app from the creators of duolingo. However, the webapp limitation to have only 150 cards per collection is really frustrating me and I've been evaluating anki or mnemosyne as a new flashcard app.
I've noticed, though, that both anki and mnemosyne seem to use something called SM 2 algorithm to calculate the frequency of repetition. However, I've also seen that SuperMemo (creators of the sm algorithm?) works with a much newer version.
Upon doing some more research I came across this site that explains what has been improved with each iteration. https://www.supermemo.com/help/smalg.htm#Historic_note:_earlier_releases_of_the_algorithm
It all sounds good and the latest versions seem to improve the older versions in very convenient ways.
Now I'm also intrigued about using supermemo, but I still do like some featurese that mnemosyne has, like different types of cards so that I can add pronunciation to the words that I learn in other languages.
What's the reason behind sticking to version 2 of SM? I can't help but to feel like we're missing optimizations. I want to pick a new flashcard software ASAP but this makes it quite hard.
Cheers!
Hello again, Mr. Peter!

Following your suggestion, I went on to check available options. I ended up liking anki because of some automation add ons that I found very useful. However, because of the questions that I had regarding supermemo algorithms I set to test myself SM-2 and a newer version (specifically SM-8) to try and see if the bold claims from supermemo were true, since I didn't find any previous third party comparisons and I wanted to know.

I conducted a two month long experiment with a set of japanese words and I reported my findings as I went through it here http://anki.tenderapp.com/discussions/effective-learning/1258-sm-2-vs-sm-15/comments/

I don't know if it's of use or interest to you, but I thought it might be of some interest. This caused another user to conduct a similar experiment that's still ongoing. I'm waiting to see what his findings are.

Cheers
--
You received this message because you are subscribed to the Google Groups "mnemosyne-proj-users" group.
To unsubscribe from this group and stop receiving emails from it, send an email to mnemosyne-proj-users+***@googlegroups.com.
To post to this group, send email to mnemosyne-proj-***@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/mnemosyne-proj-users/ad3d4209-272a-4159-b53b-3361cce08346%40googlegroups.com.
For more options, visit https://groups.google.com/d/optout.
Peter Bienstman
2018-08-13 04:09:43 UTC
Reply
Permalink
Raw Message
Thanks for the info! BTW, which automation addons did you find useful?

Cheers,

Peter


________________________________
From: ***@gmail.com
Sent: Sunday, 12 August 2018 08:39
To: mnemosyne-proj-users
Subject: [mnemosyne-proj-users] Re: Why don't anki or mnemosyne use newer SM versions?
Post by n***@gmail.com
Hi everyone! I'm new here.
I've been using spaced repetition software for quite a while with the "repetitions" app, and later on switched to the new tinycards app from the creators of duolingo. However, the webapp limitation to have only 150 cards per collection is really frustrating me and I've been evaluating anki or mnemosyne as a new flashcard app.
I've noticed, though, that both anki and mnemosyne seem to use something called SM 2 algorithm to calculate the frequency of repetition. However, I've also seen that SuperMemo (creators of the sm algorithm?) works with a much newer version.
Upon doing some more research I came across this site that explains what has been improved with each iteration. https://www.supermemo.com/help/smalg.htm#Historic_note:_earlier_releases_of_the_algorithm
It all sounds good and the latest versions seem to improve the older versions in very convenient ways.
Now I'm also intrigued about using supermemo, but I still do like some featurese that mnemosyne has, like different types of cards so that I can add pronunciation to the words that I learn in other languages.
What's the reason behind sticking to version 2 of SM? I can't help but to feel like we're missing optimizations. I want to pick a new flashcard software ASAP but this makes it quite hard.
Cheers!
Hello again, Mr. Peter!

Following your suggestion, I went on to check available options. I ended up liking anki because of some automation add ons that I found very useful. However, because of the questions that I had regarding supermemo algorithms I set to test myself SM-2 and a newer version (specifically SM-8) to try and see if the bold claims from supermemo were true, since I didn't find any previous third party comparisons and I wanted to know.

I conducted a two month long experiment with a set of japanese words and I reported my findings as I went through it here http://anki.tenderapp.com/discussions/effective-learning/1258-sm-2-vs-sm-15/comments/

I don't know if it's of use or interest to you, but I thought it might be of some interest. This caused another user to conduct a similar experiment that's still ongoing. I'm waiting to see what his findings are.

Cheers

--
You received this message because you are subscribed to the Google Groups "mnemosyne-proj-users" group.
To unsubscribe from this group and stop receiving emails from it, send an email to mnemosyne-proj-users+***@googlegroups.com.
To post to this group, send email to mnemosyne-proj-***@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/mnemosyne-proj-users/ad3d4209-272a-4159-b53b-3361cce08346%40googlegroups.com.
For more options, visit https://groups.google.com/d/optout.
--
You received this message because you are subscribed to the Google Groups "mnemosyne-proj-users" group.
To unsubscribe from this group and stop receiving emails from it, send an email to mnemosyne-proj-users+***@googlegroups.com.
To post to this group, send email to mnemosyne-proj-***@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/mnemosyne-proj-users/4df49e02-7299-47f8-a4d5-2358ec2fab3b%40owa.ugent.be.
For more options, visit https://groups.google.com/d/optout.
n***@gmail.com
2018-08-14 04:45:09 UTC
Reply
Permalink
Raw Message
Post by Peter Bienstman
Thanks for the info! BTW, which automation addons did you find useful?
Cheers,
Peter
Sent: Sunday, 12 August 2018 08:39
To: mnemosyne-proj-users
Subject: [mnemosyne-proj-users] Re: Why don't anki or mnemosyne use newer SM versions?
Post by n***@gmail.com
Hi everyone! I'm new here.
I've been using spaced repetition software for quite a while with the "repetitions" app, and later on switched to the new tinycards app from the creators of duolingo. However, the webapp limitation to have only 150 cards per collection is really frustrating
me and I've been evaluating anki or mnemosyne as a new flashcard app.
Post by n***@gmail.com
I've noticed, though, that both anki and mnemosyne seem to use something called SM 2 algorithm to calculate the frequency of repetition. However, I've also seen that SuperMemo (creators of the sm algorithm?) works with a much newer version.
Upon doing some more research I came across this site that explains what has been improved with each iteration.
https://www.supermemo.com/help/smalg.htm#Historic_note:_earlier_releases_of_the_algorithm
Post by n***@gmail.com
It all sounds good and the latest versions seem to improve the older versions in very convenient ways.
Now I'm also intrigued about using supermemo, but I still do like some featurese that mnemosyne has, like different types of cards so that I can add pronunciation to the words that I learn in other languages.
What's the reason behind sticking to version 2 of SM? I can't help but to feel like we're missing optimizations. I want to pick a new flashcard software ASAP but this makes it quite hard.
Cheers!
Hello again, Mr. Peter!
Following your suggestion, I went on to check available options. I ended up liking anki because of some automation add ons that I found very useful. However, because of the questions that I had regarding supermemo algorithms I set to test myself SM-2 and a
newer version (specifically SM-8) to try and see if the bold claims from supermemo were true, since I didn't find any previous third party comparisons and I wanted to know.
I conducted a two month long experiment with a set of japanese words and I reported my findings as I went through it here
http://anki.tenderapp.com/discussions/effective-learning/1258-sm-2-vs-sm-15/comments/
I don't know if it's of use or interest to you, but I thought it might be of some interest. This caused another user to conduct a similar experiment that's still ongoing. I'm waiting to see what his findings are.
Cheers
--
You received this message because you are subscribed to the Google Groups "mnemosyne-proj-users" group.
To view this discussion on the web visit
https://groups.google.com/d/msgid/mnemosyne-proj-users/ad3d4209-272a-4159-b53b-3361cce08346%40googlegroups.com.
For more options, visit https://groups.google.com/d/optout.
I still didn't set it up definitely because I was doing the experiment with stock, 90% retention settings without add ons, but from the top of my head the night mode, automatic TTS add on and some japanese ones to create sentences automatically as well as to better process kanji are quite useful. There were more in the add on site but I didn't check them yet.
--
You received this message because you are subscribed to the Google Groups "mnemosyne-proj-users" group.
To unsubscribe from this group and stop receiving emails from it, send an email to mnemosyne-proj-users+***@googlegroups.com.
To post to this group, send email to mnemosyne-proj-***@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/mnemosyne-proj-users/9aa91ff1-1d55-41d7-8e93-f388dee5d200%40googlegroups.com.
For more options, visit https://groups.google.com/d/optout.
Loading...