Moinsen!
thanks for the ideas on the proposed solution, seems like the temporary "other" with exit option towards single categories is generally an acceptable solution.
I do share will_c's fear of people using the offered options in erratic ways (lets face it, it happens everyday). It won't be long untill the first witty german will introduce the instrument "Arschgeige" to proove how incredibly funny one is (that can be translated to ass-violin, an idiom to describe someone who acts in an unfriendly way). Well, we will have to decide wether we want to risk these side effects for a potentially better category system - and I do feel it is worth doing that.
As Wade points out, the new system might as well lead to some existing instruments disappearing because they do not match the needed amount of tracks at first - I'm not sure if I really want to do that, but if one applied the new setup consequently, that would happen, depending on the minimum amount of tracks needed. 50 (as proposed by Wade) seems too high IMO, to ask a single user to create 50 tracks before being rewarded by the category moving on seems a little hard. If some instrument is truly rare to come by online, people would appreciate finding it even if there are very few tracks, so I'd think a minimum number of tracks around 10-20 should do on the instruments, one might apply a higher mandatory number on style categories, tho.
Now, on to the possible sub-classes of "other".
As for the instruments, thanks for the explicit list by Wade, I was missing quite some terms in english there, which you just supplied.
You came out at 7 sub-categories of instruments, which does seem longer than what I feel comfortable with - a general niterface design rule tells us options that exceed the number of five options are perceived as complex by the average user (sorry, sidenote).
A simpler grouping by sub-classes may reduce Wades 7 to:
1. other winds
2. other strings
3. other percussion *
4. other sounds (other others, lol)
Thing is, our current "Percussion" category basicly IS that "other percussion", since we never split the tablas from the shakers there.
One aspect that does seem relevant is to weigh the amount of words needed to explain a categories concept against real benefit of having some complicated category. I just can't offer something needing 6 words to explain the concept in a dropdown menu (these things tend to inflate in translation as well, the french will take 30% more words...), so one-word self explanatory terms are ideal to start off with in this regard.
To split the "winds" into three categories as Wade did is too precise IMO, if one would go for this level of distinction, I feel one would have to split the "bowed" from the "plugged" string instruments as well. Since Keyboards are available as a standard category, I'd think it would make more sense to offer "other tuned percussion", which is a easier to grab concept than the mix of these two.
OK, lets take a step back:
We are introducing something new here, which can be summed up as "category hierarchy".
This is a new concept to wikiloops, whichs categories are all on the very same hierarchy level right now.
To ask users to dig down a hierarchic menu when uploading a track is one thing - a click flow from "other wind" to "Brass" to "(proposed cat) specialHoooorn" would be no issue for someone happy to have the correct instrument stated.
BUT please do not forget the category systems main use is on the search pages of wikiloops, where people tend to select 2-3 instruments to search for a certain lineup.
To handle different hierarchy levels in that usecase is much more difficult and a pain to do - imagine you would have to select "general percussions" and then range down to what you are really after, to then move on to "general strings" and dig down to "guitar" - that just won't work in a nice flow.
Looking at the organicly grown list of currently offered instruments, we have to notice that these cats are treated equally in hierarchy, but we can spot cases that would go on different levels once we apply our new logic.
Examples: Didgeridoo and Fluegelhorn are very specific and would remain on category level 1, while both "Keys" and "Percussion" are rather global cats that would happen in the layer above, lets call it level 2.
Let's not change the running system in a confusing way - so, we will have to live with level 2 categories being listed on the search interface.
What will be an odd user experience at some point is the moment where the system ads a new level 1 category (p.e. by splitting out all Tabla tracks from the Percussion), and a user will find "percussion" and "tabla" in the same selection menu... depending on what one reads first, one will expect that level of distinction to continue, and one will either wonder why in the world tabla is the only thing singled out from percussion, or will be disappointed to find there is no darbuka category... this is indeed a very tricky problem, which already does exist in a mild form in the "acoutic guitar" vs "guitar" categories.
Sorry for the abstract thinking here, but one really needs to try to foresee things...
For the sake of completion, let's think about the newly created "other X" categories - I would not have integrated them in the filtered by instrument search offerings, because these rather loose & temporary cats do not work well for that purpose (who would search for a lineup consisting of Drums, Bass and "other wind instrument", really?), so, I'd place those in hierarchy level 3, which shall not be seen on the search interface.
Houston, we do have a quite difficult problem here, and I'm affraid my collection of thoughts here may exceed what the average bypasser will find easy to digest... hope some folks may still enjoy witnessing the process of how human thought may be converted to a computer maintainable logic - that's what developing is about.
I'll take another approach. Let's collect different usecases to spot the problematic aspects.
Number 1: Ali from Hindustan proposes to have a Schningiwingi instrument category, files that under "other stringed instrument" for the time being and adds 12 tracks this way.
Our system to come recognizes that and opens a Schningiwingi category - beautiful, no problem here.
Number 2: Jürgen from Ida-Oberstein has a Steinway at home, and really doesn't feel comfortable filing his tracks in the quite general "Keys" category. So, noticing our newly added features, he starts to file his tracks as "other stringed instrument" and proposes to open a "concert piano" category. System waits for the minimum amount of tracks and opens that category.
Problem: We now have a Keys category, which includes quite a lot of older concert piano tracks, and a concert piano category featuring only Jürgens tracks. That is definetly a problem which would apply to all rather loose instrument categories, mostly in the Keys and Percussion segment, which are very likely to become sub-splitted, but any sub-split of a previously existing category has that problem to it, think of someone splitting out the alto sax from saxophones, same problem there as well.
Number 3:
Juniper from Azerbeydjan visits wikiloops for the first time and has a look at the search page of wikiloops. She is looking for Darbuka tracks, and starts looking down the list of offered instruments, which is currently delivered in alphabetical order. She bypasses "Congas" and finds the List continues with "Digeridoo" - and the impression is: "they have a sophisticated system that grants Congas a seperate category, but obviously they do not have darbuka here." - wether Juniper reads on to spot "Percussions" further down the list and give that a try is questionable, once we have some specific categories, people will expect this to be the standard.
One solution to prevent this situation could be to actually hide categories which have been sub-splitted - Example: Once the Keys category has been sub splitted into "concert piano", "organ", "synth & pad" and "e-piano", one could hide the global Keys category, but as in the previous case, all older tracks which are not assigned to these newly created subclasses would become unavailable for searching by lineup.... veeery difficult situation.
Number 4:
Phil the bass nerd feels the need to propose a category for "pre 1972 american build fender jazz basses" - (it will happen, trust me) - I guess it's a far out enough example to notice: If we do not build in some kind of "sorry, that's not really what this is meant for" exit option, we'll wind up with way too many user created categories.
If you look at the way other wikis handle these problems (think of: "thread closed, duplicate post of..., format of question does not match guidelines" on Q&A forums as stackoverflow or endless discussions about categories and worthyness on the wikipedia), then we should be aware that there is a lot of time and energy to be spent on explaining folks like Phil the bass nerd why their idea of what would be cool is a bit too much for what we want to have here...
conclusions for today:
1st: The "what to do with older content once an existing category gets split into subsegments" seems the major issue here.
One could ask people to help re-classify their older tracks, but I have doubts all users would be willing and able to do that. One could try to re-categorize older tracks by running text searches before opening a new category (all older tracks tagged with "concert piano" could be easily moved into Jürgens new category from above example), but that would probably only find a rather small percentage of tracks.
2nd: To avoid confusion, the instrument selections offered on the search interface should be re-ordered in a logic way instead of by alphabet. If the List read
"
[i]rythm[/i]
Drums
Congas
Tablas
other Percussion
[i]strings[/i]
Guitar, electric
Guitar, acoustic"
etc etc,
that would make a lot more sense than spreading these items by alphabetical order, where "Acoustic guitar" and "Guitar" are seperated by several other instrument cats.
3rd I guess we need to communicate some sort of line of which level of subsplitting is considered relevant enough to be used in a category system. The "minimum amount of tracks" rule will not solve that by itself, so, manual selection based on a common standard will be needed there.
It will be quite hard to find and argue such a common standard, simply because the involved peoples perceptions may vary quite strongly. While Wade gets goosebumps when thinking of Ragtime being classified as Jazz, it feels the most obvious and simple thing to do for me (I admit to having uploaded bird song labeled as "flute", too).
I would light-heartedly place BoogieWoogie tunes into the blues segment, too, to find a general consensus of "this category does make a lot of sense" will be difficult.
Maybe adding a "at least three users must have uploaded to a category"-rule could be one way to make sure we are giving the right amount of attention to something, such a rule would slow down the process very rudely. Sometimes one single starter to a new category is all we have... difficult.
4th - looking at the amount of work that might be necessary to cover the mentioned problems, I'm a bit undecided wether this project here shall be included in the 9.0 update which I'm working on these days... I'll wait and see if we can agree on a desireable solution for the open ends I layed out today and make up my mind then.
