Pros and cons of a single supermemo collection: Difference between revisions
mNo edit summary |
|||
| (5 intermediate revisions by the same user not shown) | |||
| Line 1: | Line 1: | ||
{{IncrementalWarning}} | |||
<div style="float:right">__TOC__</div> | <div style="float:right">__TOC__</div> | ||
| Line 4: | Line 6: | ||
===Interleaving=== | ===Interleaving=== | ||
Video/audio repetitions break the flow | Splitting your learning across collections you sacrifice some degree of [[interleaving]], specially macro-interleaving. I won't ever interleave Chinese or other languages in a dedicated collection while I use my main collection. You still get micro-intervealing. If you use SuperMemo extensively every day with large collections, the downsides are minimized but never nullified. | ||
Having a single collection will open the door to interleaving Video/audio repetitions that break the flow. | |||
===Stats=== | ===Stats=== | ||
Averages may be not representative | Averages may be not representative | ||
You can quickly see the volume of your learning, time invested and such when you select or switch a collection, so it is very easy and quick to keep track and compare these stats. If you only have one collection, t | |||
===Analytics=== | ===Analytics=== | ||
===Duplicates=== | ===Duplicates=== | ||
===Backup=== | ===Backup=== | ||
Using the built-in backup solution is increasingly inefficient as collection count grows. You would need to backup each collection one by one. It makes close to no difference if you are using a [[Supermemo_backup_using_Git|git solution for backup]], as I do strongly recommend to do. | Using the built-in backup solution is increasingly inefficient as collection count grows. You would need to backup each collection one by one. It makes close to no difference if you are using a [[Supermemo_backup_using_Git|git solution for backup]], as I do strongly recommend to do. | ||
===Secondary storage=== | |||
===Time investment=== | ===Time investment=== | ||
You can favorite the most frequently used collections, which make the process of switching collections very quick. Across the day, the time invested into switching collections is just a matter of seconds. | |||
===Registries=== | ===Registries=== | ||
The big perk is that you have all registries in one place. First, no need to switch to other collections ever: all images, sound, templates, references, etc are could be found with a single search. | |||
When I had the English collection split from my main collection, sometimes I was trying to link an image that was indeed not found in that collection. So you have the choice of either duplicate the image in a second collection, or either pass with no image. | |||
===Searches=== | ===Searches=== | ||
Searches are done in assembly language for maximum speed. Even tough, when you reach large volumes you can appreciate the delay. Therefore, having a single collection will foment larger volumes of text and other registries making searches slower. It will make a difference if you are strarting a new collection for learning a language and perform searches very often. One exception would be English language, as it is the vehicular language for Incremental Reading. The same would apply if you are using another language instead of English. | Searches are done in assembly language for maximum speed. Even tough, when you reach large volumes you can appreciate the delay. Therefore, having a single collection will foment larger volumes of text and other registries making searches slower. It will make a difference if you are strarting a new collection for learning a language and perform searches very often. One exception would be English language, as it is the vehicular language for Incremental Reading. The same would apply if you are using another language instead of English. | ||
Latest revision as of 13:06, 2 November 2021
There is no good and right about having one single collection or several for SuperMemo. You may decide when it is best to start a new collection, merge collections, or transfer branches between different collections according to many factors and current needs.
Interleaving
Splitting your learning across collections you sacrifice some degree of interleaving, specially macro-interleaving. I won't ever interleave Chinese or other languages in a dedicated collection while I use my main collection. You still get micro-intervealing. If you use SuperMemo extensively every day with large collections, the downsides are minimized but never nullified.
Having a single collection will open the door to interleaving Video/audio repetitions that break the flow.
Stats
Averages may be not representative
You can quickly see the volume of your learning, time invested and such when you select or switch a collection, so it is very easy and quick to keep track and compare these stats. If you only have one collection, t
Analytics
Duplicates
Backup
Using the built-in backup solution is increasingly inefficient as collection count grows. You would need to backup each collection one by one. It makes close to no difference if you are using a git solution for backup, as I do strongly recommend to do.
Secondary storage
Time investment
You can favorite the most frequently used collections, which make the process of switching collections very quick. Across the day, the time invested into switching collections is just a matter of seconds.
Registries
The big perk is that you have all registries in one place. First, no need to switch to other collections ever: all images, sound, templates, references, etc are could be found with a single search. When I had the English collection split from my main collection, sometimes I was trying to link an image that was indeed not found in that collection. So you have the choice of either duplicate the image in a second collection, or either pass with no image.
Searches
Searches are done in assembly language for maximum speed. Even tough, when you reach large volumes you can appreciate the delay. Therefore, having a single collection will foment larger volumes of text and other registries making searches slower. It will make a difference if you are strarting a new collection for learning a language and perform searches very often. One exception would be English language, as it is the vehicular language for Incremental Reading. The same would apply if you are using another language instead of English.
Collection repairs
With everything related to collection size, collection repairs will take much longer, more or less the same time you would need to run repair to all of your collections. For my year old collection, it takes long enough to get away from the keyboard or move to a difference software while the repair takes place. The lexicon is taken to a lunatic dimensions. I do no recomment do built it ever, specially if you are not using plain text in your collection.
Collection parameters
Same collection parameters for different difficulty material is shared. That can be percieved as either positive, neutral or negative. In the case of a single collection, it is a matter of time to become a negative aspect. Some parameters are calculates from your entire collection, such as the optimal interval, ratention rate, forgetting index.
Final Drill
Having multiple collections make the selective final drill very accesible, as you can turn on/off the finall drill for each collection. For instance I could turn it on for language learning collections. Having a single collection makes the finall drill a everything or nothing choice. There is a word-around by creating subset reviews from the drill queue (video).
Priorities
Requires more precision for prioritization