Talk:Spaced repetition

Software classification
The classification of the available software into Free Open Source Software and Commercial looks a bit dodgy. There is closed-source non-commercial software, as well as open-source commercial software. 88.97.173.222 21:58, 12 November 2007 (UTC)

No original research
Graph showing spaced-repetition sequences as generated by the language-learning program Gradint. The horizontal axis represents time; each ellipse represents a separate sequence. The sequences shaded pink do the newest words and have the most repetitions. The shorter sequences do background revision of older words. Sequences are arranged so that a lesson can switch between them in an irregular pattern as shown by the red line. The "springs" (wavy lines) represent degrees of freedom in the sequences, which have been stretched or squashed to make that arrangement possible. —Preceding unsigned comment added by Silas S. Brown (talk • contribs) 10:04, 18 January 2008 (UTC)

The recent graph and description of spaced repetition software looks like an original representation of the problem, specific to a particular and unique computer software program, which addition is in conflict with Wikipedia's "No Original Research" or "Notability" or both. aruffo (talk) 20:16, 12 January 2008 (UTC)

I don't understand how "no original research" can apply, since this approach was not first published on Wikipedia (as cited on my website I had a peer-reviewed paper in the Proceedings of the 10th International Conference on Human-Computer Interaction about gradint), and No original research has a section on "Original images" which I understand to be saying that diagrams are OK to illustrate an existing idea. As for Notability, it says "Notability guidelines give guidance on whether a topic is notable enough to be included in Wikipedia as a separate article, but do not specifically regulate the content of articles". I accept that Gradint is not notable enough to warrant a separate article but I don't understand why one of its diagrams can't be used to illustrate spaced repetition, at least until such a time as somebody draws a clearer diagram by hand (which I can't do). Silas S. Brown (talk) 09:07, 17 January 2008 (UTC)

—

The top part of the article should be explaining spaced repetition, not presenting the user with a complicated graph that "helpfully" mentions the software used to generate it. But said mention is not the main issue - the problem is that the graph does not make the concept of spaced repetition clearer to the user! The author of the software that generated the graph may feel that it is an adequate portrayal of spaced repetition, but I do not think it is enlightening at all. The graph on http://en.wikipedia.org/wiki/Forgetting_curve would be a much clearer choice. —Preceding unsigned comment added by 58.3.183.73 (talk • contribs)

—

I was hoping that somebody would replace my graph with a better one; I only wanted to "start the ball rolling" with the best I can do (which isn't as good as what others can do I know; I have CVI so my visual skills are not great, but I was hoping for an improvement not just a deletion). Mentioning the software was besides the point (and by the way it's GPL and I don't get anything for it), sorry if that caused any problems. (I could have said "one particular piece of software" instead of "Gradint" but I thought Wikipedia prefers specific references.)

OK so who wants to put a better graph in? I'm not so sure about the graph at Forgetting curve because it does not illustrate spaced repetition; it only illustrates a forgetting curve. And I'm a bit confused by the fact that its horizontal axis is measured in days, whereas I always thought spaced repetition was dealing with timescales of minutes, as in Pimsleur where everything is done over a 30-minute lesson. The promotional graph for Pimsleur at http://www.pimsleur.com.au/images/PimsleurintervalRecall.gif illustrates spaced repetition quite nicely but (1) it's copyright (2) it's rather small and blurred (I only know what it says because I've seen the real document in a Pimsleur box) and (3) it does not show the fact that other words can be learned during the gaps. My graph was an attempt to address these 3 points, but perhaps it introduced problems of its own. Silas S. Brown (talk) 09:06, 18 January 2008 (UTC)

If you have access to JSTOR (provided by some university libraries), see also the graph on page 3 of this paper: http://links.jstor.org/sici?sici=0026-7902(196702)51%3A2%3C73%3AAMS%3E2.0.CO%3B2-3 (A Memory Schedule, Paul Pimsleur, The Modern Language Journal, Vol. 51, No. 2 (Feb., 1967), pp. 73-75, doi:10.2307/321812) Silas S. Brown (talk) 09:42, 18 January 2008 (UTC)

—

Appreciate you were trying to help, but I'm pretty sure the diagram doesn't adequately demonstrate spaced repetition. Also in Firefox it rendered at the size of my browser (1900 odd pixels big), which meant that it took up half the screen.

I do not believe that the difference between days and minutes in the diagram is of consequence. Whether the repetition is on an (ultra) short term memory basis or on a longer term basis, the concept is the same. You'll also find that most of the flashcard applications out there operate on a long term basis, rather than trying to train for short term memory.

As for demonstrating multiple cards at once - the article is on spaced repetition, not spaced repetition systems. Addressing the issue in the context of a single item of data makes sense to me.

58.3.183.73 (talk) 12:32, 18 January 2008 (UTC)

-

Maybe a better graph would be something along the lines of:

item   time -> #1  o o  o     o          o              o                                    o                       #2   o o  o     x o  o   o     x o  o    o      o             o                       o  #3       o o  o    o         o                 x o    o         o             o  #4                              o x o o   o      o            o                 o where o = successful repetition, x = failed  —Preceding unsigned comment added by 129.97.51.132 (talk) 04:10, 21 March 2008 (UTC)

Is spaced repetition the same as graduated interval recall?
Wikipedia's entry for Graduated interval recall was a re-direct to spaced repetition, and the spaced repetition page started off by suggesting a list of alternative names including "graduated intervals", which suggests that "spaced repetition" and "graduated-interval recall" are synonyms. Is spaced repetition really the same as graduated interval recall (as described by Pimsleur 1967 see reference above), or is graduated interval recall merely a particular type of spaced repetition? Some points in the above discussion about the graph suggest that the two terms are not the same. We need to be clearer about this. Silas S. Brown (talk) 10:29, 18 January 2008 (UTC)

Dispute over the software list
Revision 16:13, 30 November 2007 and earlier of the Spaced Repetition article contained a "Computer software using spaced repetition" section.

This section had been spamed and contained many links to commercial software and websites with no clear link to spaced repetition (most being flash card based but with no indication as to whether scheduling was done using spaced repetition or was anything more than purely random).

However it did contain links to many interesting open source projects, e.g. Mnemosyne, and to commercial products such as Supermemo whose author Piotr Wozniak is well known in the field and whose site, while commercial, provides much interesting and detailed academic discussion.

When I first looked at this article I found this list very useful in finding out who was doing what in this field. I'm sure it was also useful for those who simply wanted to find a good software application to help them with learning using spaced repetion.

Instead of pruning out the advertising links this section was completely removed in revision 16:41, 10 December 2007 by 130.113.219.231.

This removal was reverted by the User:ClueBot on the same day, then the next day the removal was redone by User:Aruffo.

Since then user User:Aruffo has reverted the subsequent readition of this list by 58.3.183.73 in revision 7:11, 19 December 2007

Following the lists removal from the article it was added by 85.180.233.22 to the article's discussion page.

I discovered this sequence of events on coming back to the article after not having looked at it since it had contained the list.

As I'd found the list useful but saw that it was clear User:Aruffo was determined to keep it out of the article I added a pointer at the bottom of the article (in revision 13:36, 26 February 2008) to the lists existence on the discussion page

My revision, in addition to the pointer to the list, included sentences that were more appropriate to the discussion page.

Instead of just removing those sentences User:Aruffo reverted my entire revision.

Following this I added a very neutral single line pointer to the lists existence in revision 22:01, 26 February 2008.

User:Aruffo reverted this as well with the comment "rv - if it belonged in the article, it wouldn't be on the discussion page". This seems somewhat circular logic as they are the person who forced the situation where the list could only exist on the discussion page.

Despite knowing it was petty (and aware of the 3 reverts rule) I reverted his revert, and he has subsequently reverted that revert.

As things stand the article no longer contains the list or any pointer to it. I feel this list is useful and unlikely to be found by casual readers unless it is in the article or there is at least a pointer to it.

While general Wikipedia policy may be against list of links I think this list is useful and I think its simple removal rather than attempting to provide equivalently useful information in a more Wikipedia acceptable form, or at least providing a link to a similar list elsewhere if its believed to be simply replicating something else (I don't think this is the case), is not constructive.

Note: IP 165.222.186.230 and 80.219.197.123 are both me - the author of this note. 165.222.186.231 (talk) 11:06, 29 February 2008 (UTC)


 * You are correct to state that the links, commercial or non, are against Wiki policy. aruffo (talk) 06:03, 1 March 2008 (UTC)


 * Just to add a citation for the relevant policy docs:

External_links and an entry on "what wikipedia is not" Particularly this bit: "Some external links are welcome (see What should be linked), but Wikipedia's purpose is not to include a comprehensive list of external links related to each topic. No page should be linked from a Wikipedia article unless its inclusion is justifiable. // The subject of this guideline is external links that are not citations of article sources. If the website or page to which you want to link includes information that is not yet a part of the article, consider using it as a source for the article, and citing it. Guidelines for sourcing, which includes external links used as citations, are discussed at Reliable sources and Citing sources." If you think "Supermemo whose author Piotr Wozniak is well known in the field and whose site, while commercial, provides much interesting and detailed academic discussion" then could this be worked into the article and cited? (Disclaimer: I have not read all Wikipedia policy documents) Silas S. Brown (talk) 21:14, 15 March 2008 (UTC)

Reliable source
Yug (talk)  18:33, 12 March 2012 (UTC)

Another source
This Wired article has a lot of good information that should be incorporated into the article. Also this article should link to Forgetting curve prominently in the body of the text. Also: at least mention Hermann Ebbinghaus's work, as well as Wozniak's. JKeck (talk) 04:04, 20 May 2012 (UTC)

Reference 7 broken
reference 7 seems to be broken — Preceding unsigned comment added by 79.182.185.181 (talk) 13:54, 1 March 2013 (UTC)
 * You were right. The link didn't seem to be a reliable source, plus there was another source, so I simply removed it. Thank you for telling us!  Lova Falk     talk   20:24, 1 March 2013 (UTC)

How much improvement over non-spaced repetition?
If there's a literature review that roughly quantifies how much improvement spaced repetition usually brings, please add that information to the article. — Preceding unsigned comment added by Rolf h nelson (talk • contribs) 03:15, 19 June 2013 (UTC)

Flash cards "unwieldy" creates an emotional bias
The Research and Applications Section sounds like flash cards are a deprecated technique. But they are still in widespread use. So the wording should be changed (for WP:NPOV to reflect that computer systems are an alternative to flash cards but that they do not provide an automatic advantage for training sets below a certain size. Draketo (talk) 07:27, 14 April 2014 (UTC)

"Programs…tailor to learner performance" conveys the incorrrect notion that flash cards do not do this
Since there are numerous ways to tailor flash cards to learner performance, this goal does not differenciate computer programs from flash cards. I see the incorrect notion it conveys as WP:NPOV problem. Draketo (talk) 07:26, 14 April 2014 (UTC)

Merger

 * Support: These articles are the same topic. --Wikiman2718 (talk) 04:36, 6 July 2019 (UTC)
 * Are you able to put the work in to integrate the information from that page here? These pages indeed should be merged. -Ramzuiv (talk) 03:30, 18 November 2019 (UTC)
 * Will try my hand at merging the information. MasterChief1986 (talk) 20:59, 14 February 2020 (UTC)
 * Support makeswell (talk) 21:07, 23 September 2021 (UTC)

What content still needs additional citations?
Aside from some brief parts of the history section the article seems to be well-cited. Cheers. Symphony Regalia (talk) 02:39, 10 May 2020 (UTC)

Merger discussion, March 2021
I propose to merge Distributed practice into Spaced repetition. These are the same subject, as the first sentence of Distributed practice says: Distributed practice (also known as spaced repetition or spaced practice). The current situation is a content fork. Biogeographist (talk) 00:42, 18 March 2021 (UTC)

I think there is a subtle difference between the two with the current wording/definition, namely the second line of Spaced repetition: "Newly introduced and more difficult flashcards are shown more frequently, while older and less difficult flashcards are shown less frequently in order to exploit the psychological spacing effect.".

This seems to define spaced repetition as repetition *where the length of spacing varies depending on evidence (i.e. whether the person remembers it well)*. The distributed practice page, on the other hand, seems to me to be a broader definition that encompasses *any* length of spacing, not necessarily one that depends on feedback/evidence from the learner.

An example for clarity:

Spaced repetition: If you see a flashcard and didn't remember it, you see it again in 1 day. If you remembered it, then you see it again in 7 days. The second time you see it, if you remember it, you'll see it again 30 days. If you don't, you'll see it again in 3 days.

Distributed practice: If you see a flashcard, you see it again in 3 days. The second time you see it, you'll see it again in 5 days. Whether or not you remembered it is irrelevant.

So I would consider "Distributed practice" to be a superset of "Spaced repetition". Since both pages have a fair amount of content, perhaps it would be more appropriate to edit the lead paragraph of the Distributed practice page for more clarity.

Edasaki (talk) 07:02, 2 November 2021 (UTC)


 * From my research into learning techniques in sport and academic education spaced repetition as a term is quite infrequent when compared to distributed practice. The spacing between tests, practice, or efforts are most often referred to as spacing and then often reference distributed practice literature. Interleaving is another term that is thrown into this conversation. From my perspective distributed practice is the more widely used term, and spaced repetition is used by a select group of researchers. Another factors to consider is the comparison of practice modalities. Mass practice and deliberate practice are other terms used in the literature which can be associated with distributed practice easier by name. Just some thoughts to consider. DannyHatcher (talk) 19:18, 4 December 2021 (UTC)


 * Adding to this conversation should be the Spacing effect article which seems to also repeat the same information. I am happy to work on a merge of these articles at the beginning of the new year if there are no objections. DannyHatcher (talk) 19:53, 6 December 2021 (UTC)


 * Based on the above, I would also support a merge of the articles. I agree that "spaced repetition" is a specific subset of "distributed practice" more generally. Personally, I had only ever come across the term "spaced repetition" before. Bibeyjj (talk) 08:58, 2 January 2022 (UTC)

Spaced repetion vs uniform repetition
The discussion of spaced repetition vs uniform only talks about retention. There is a big gap though in that if retention is on par or even lower than uniform repetition, spaced repetition would allow learning higher volumes of knowledge during the same period. This would also likely induce a better psychological response to the learning. — Preceding unsigned comment added by MichelleValentinova (talk • contribs) 10:15, 19 June 2022 (UTC)

Add the new open-source FSRS algorithm from Anki 23.10 and MaiMemo
Anki release 23.10 added FSRS, an new open-source spaced repetition algorithm, at the moment as an opt-in feature. There have been two papers about FSRS, A Stochastic Shortest Path Algorithm for Optimizing Spaced Repetition Scheduling and Optimizing Spaced Repetition Schedule by Capturing the Dynamics of Memory (paywalled). Also it has already been used in the Chinese MaiMemo learning app. Therefore, I think it is already at a point where it can be included in the list of spaced repetition algorithms, the significance is in my opinion no lower than e.g. the SuperMemo neural network algorithm. There is a lot of documentation even besides the papers on how the algorithm works, e.g. this short description of the FSRS v4 algorithm and this pedagogical introduction into spaced repetition algorithms from the beginnings to FSRS. In these posts there is a lot that could be used for this Wikipedia page, even beyond FSRS. Though I'm not sure if a write-up on a github wiki counts as a good source or we should cite the papers directly. Elimik31 (talk) 15:09, 14 November 2023 (UTC)


 * I now added FSRS myself in this version and hope others might improve on that. I'm still not sure whether I should add some sentences how it works. It's based on the DSR memory model and employs a stochastic shortest path algorithm for finding optimal parameters (all explained in the references), but that might be a bit much to explain in a single bulletpoint. Elimik31 (talk) 20:18, 14 November 2023 (UTC)