It seems like there’s enough interest in this to make it happen, so here we go: TagFest 2024 now officially exists!
Ground rules:
The event will run from Oct 1 to Oct 31.
Participants are highly encouraged to read the thread cited by @alyshkalia. (See the first item on the list of related threads below.)
Competition elements will be informal and unofficial. Stats will be provided at a future date.
General guidelines:
Tags are by their nature free-form, uncontrolled and messy. Don’t worry about it. It’s fine. (Really!)
The most valuable thing about a tag is that it points out something distinctive. When considering a new tag, look through the list of existing tags to see if any of them are close enough, and check the games that are tagged to see how other people are interpreting the term.
The second most valuable thing about a tag is that it is applicable to more than one game – even games that you don’t know about or which haven’t yet been written. Use your judgment when deciding whether a tag is likely to be applicable in the long-term.
The third most valuable thing about a tag is that it represents a type of perspective. Your perspective will differ from that of others. When tagging, stick to aspects that are at least potentially demonstrable, i.e. that you could write an essay citing game text or context to support.
Both subset and superset tags can be valuable. For example, if a game features a crow, you could add “crow” or “bird” or both! A tag “corvid” is less likely to be generally useful, but that’s because fewer people know what it means. Go ahead – it doesn’t really hurt, and a future ornithologist might appreciate it (and start extending it to other games).
The system supports multiple users applying the same tag. You may wish to “repeat” a tag that you already see on a game. Doing so might make it easier to judge a tag’s validity via automation, and definitely helps to preserve tags should a given user account be deleted.
I’m laying out those guidelines not as some kind of “master of tags” but as someone who has spent a lot of time trying to convert the tag data into something useful for recommendations. To me, those are generally good guidelines for making useful tags, but see point 1: Tags are “uncontrolled” by design.
Give yourself credit and help make this into a spectator sport with these polls:
So far during the event, I have tagged…
at least one game
a few games
several games
many games
a lot of games
a whole lot of games
ALL THE GAMES!
0voters
So far during the event, I have added…
no new tags
at least one new tag
a few new tags
several new tags
many tags
a lot of tags (which is probably enough)
0voters
Having dutifully checked the list of all tags for synonyms (and checked associated games for usage), I have found that…
none of my tags are covered by existing tags
a few of my tags are covered by existing tags
some of my tags are covered by existing tags
about half of my tags are covered by existing tags
most of my tags are covered by existing tags
nearly all of my tags are covered by existing tags
As a side tangent, without meaning to grump over enthusiasm, this is one of my frustrations with tagging on IFDB.
It’s great that you can use any tag and create your own, but the interface doesn’t provide any help with tag-matching.
On the forum here if you type a tag it provides completion suggestions so if a similar tag exists, you can choose it instead of creating a new one. In the background, I also have admin tools where I can mass-merge similar tags to clean them up. IFDB on the user-end doesn’t check tags against existing ones so you get situations like this:
While those might mean slightly different things, they do complicate searching for a tag. The only real solution is to make sure every choice game has all variations of the “choice” tags which gets cluttered.
That said, the tags are great as descriptors within each individual game, and I’m all for this tagging project. Maybe people don’t search by tag as much as I suspect and just having them visually within game listings is sufficient.
I absolutely agree that the lack of tag suggestions and the inability to merge tags makes IFDB’s tagging feature less useful than it could be. Based on the first thread that was linked above, I think there’s a fair number of people who feel that way, but it’s never reached the point of someone going “here’s my proposed code to make this happen” and so nothing’s come of it.
Perhaps at minimum the ‘tag cloud’ screen could also offer a regular text list (maybe a downloadable text file?) of all the tags and the number of times they’re used instead of indicating popularity by font-size?
I understand IFDB is legacy code and it’s probably a bummer going in there to make significant changes.
A list by numbered use and popularity might offer the chance for a similar re-tagging clean up project where people can go through and make sure all the choice games are tagged “choice based” instead of “choice-based” or vice-versa. At least for the common categories where similar tags are only slightly different by spelling. I get that sometimes it’s useful to have singular wacky tags.
Yeah, I think some significant tag cleanup is needed to help make tags more useful. I’m going to put in a Github suggestion (or maybe several?) related to tags. Will include something about tag consolidation, tag suggestion/autocomplete, and an alternate listing of all tags currently in use.
Also, some tags are redundant of metadata that already exists in dedicated fields (e.g., comp/event a game was entered in). Are there benefits to this duplication of data? I can see “IFComp game” being useful as a tag, as it collects all IFComp games, but is “IFComp 2024” needed when there’s already an IFDB page that lists all IFComp 2024 games?
I find the ECTOComp 2022, etc… tags quite useful since the competition pages don’t list games the same way. I guess if we had a search filter (with a clickable link on the game’s page) we could do away with those tags.
On the other hand the published in 2007, etc… tags I find quite useless and would be in favor of removing.
I’m not certain how much database editing can be done directly by the IFDB team. Like, are there automated tools where you can go into the database directly and change all instances of “sci-fi” to “science fiction” in a single action? Or does a feature need to be added to the code to allow the mods to do that?
Searching and replacing one specific tag probably wouldn’t be too bad, but doing it to a lot of tags on a regular basis would be difficult and risk some kind of errors. To do it consistently new functionality would be required.
Right now, the only one consistently making updates to IFDB’s source code is Dan; if someone else knows enough to work on it, it would help. I’m not really good enough to do pull requests, and Docker is such a memory hog I don’t want it running on my computer.
On a different note, something I’m planning to do for the TagFest is, rather than (solely) approaching it game-by-game, thinking about a descriptor that applies to multiple games I’ve played and then adding that tag to all those games.
And back to the subject of general IFDB tag handling, I found these three existing issues on GitHub:
So I will make a new one solely about the all-tags lists. I do hope the discussion here can maybe prompt these 3-plus-year-old issues being addressed! I know nothing about writing code for such things, so sadly I can’t be of any help in that way…
Edit: I don’t know how to remove those giant GitHub preview boxes
Regarding tag management, I can only repeat: “It’s fine. (Really!)” And also: “Tags are by their nature free-form, uncontrolled and messy.”
Rather than argue about which tags should and shouldn’t be there, I would encourage anyone interested in standardization make suggestions in this thread (or, better, in a related subject area-specific thread to which I can link above) for new tags and their intended meaning(s). For example, some “mood” categorizations have already been suggested.
If you want two tags to have the same meaning, lay out your argument. You may be surprised to hear opposing opinions. (To use bg’s specific example of bird vs. birds: Suppose I say “Bird are an important feature of that game.” See what I mean?)
Thinking about a tag’s utility in advance (as per the guidelines and alyshkalia’s post above) is a very good idea. Sharing your scheme is also a good idea. Every person who likes your proposed scheme can adopt and use it; consensus is in the doing. The process is very democratic but also anarchic. Even without central management, it will still improve the situation.
You don’t need anyone’s permission to create or apply a tag. Just go ahead and do it! (…but please check the spelling first.)
Seeing so much discussion about the tagging system motivates me to look into the pull request thing and see if there’s anything I can do to improve it. A lot of the quibbles people have with tags are the kind of thing you need to go into the code to fix, and I might be able to figure that out if I had a few days… Maybe. No guarantees, but I can try to see if I can accomplish anything by the end of October. I imagine implementing even one of the suggestions mentioned above could make stuff go smoother.
Bonus task: If anyone has the chance to look at the genre category of games (i.e. the actual category, not tags) while tagging, it might be helpful to do some standardization there. If listing multiple genres for a game, I would suggest that they be separated by commas or slashes, with the most relevant (in your view) being first on the list.
Here is a list of valid-seeming genres in use (ordered high to low frequency) as of the start of September:
At first I thought you were suggesting tags that list multiple genres in the same tag, but now I think you must be talking about the genre field, not the tags. Is that right?
