More IFDB data

On my previous post, Ron commented speculating about what has changed about IF in the last few years, and that spurred me to check a couple of other hypotheses.

(1) Slice of life games have gotten more popular in the post-2000 period.

Not so much, it turns out.

I’d really love to do a full breakdown on genre percentages and see how horror, SF, fantasy, and other elements have fluctuated over the years, but that would require a Lot of Counting.

(2) Games have gotten more novice-friendly since about 2004, with more tutorials, help menus, built-in hints, maps, improved parser messages, smarter can’t-go replies, navigation by GO TO ROOM, and so on. Certainly it feels like we’ve talked a lot more in the past five years or so about how to make games accessible and reduce stuckness, and there have been a host of libraries, extensions, and goodies designed to make this easier.

But is it making a significant difference?

This one is harder to address, because there aren’t tags for most of these features. (I did go in and tag games that I know of that have included feelie maps, built-in maps inside the game, room-name navigation like GO TO THE KITCHEN, or tutorial modes, but my memory is faulty and incomplete. If other people want to go in and add their own tags, that would be very welcome.)

The one tag that is readily searchable is “built-in hints”, and it doesn’t give the results I might have expected:

In fact, it’s a bit dippy around 1998. But wait! As we saw last time, 1998 was the year of the puzzleless-game spike. Do we expect built-in hints in puzzleless games? Probably not so much. A “Puzzleless *or* built-in hints” chart looks like this, a kind of anti-stuckness graph:

Still, even with the adjustment, it looks as though game “friendliness” or accessibility has developed in just the opposite way to what I would have expected.

This could be another sign that my expectations are just wrong, all wrong, but I think there might be some other issues.

These graphs don’t account for game length: games used to be longer on average, and therefore there was more need for built-in hints in order to get through the morass. And there are a lot of options that IFDB tags don’t currently cover, as mentioned, or don’t cover enough to be significant. For instance, several of my games don’t have built-in hints but did feature invisiclues-style accompaniments outside the game, and there are other people who have done feelie hints as well in preference to something inside the game. So I feel like we’d need a lot more tagging of various kinds to give a clear idea of accessibility features as they’ve evolved over the past few years.

Another thing this graph doesn’t deal with is the difference between major releases and trifles or toys. Speed IF isn’t likely to come with any player niceties — because there’s no time, and no one is expecting any polish. But a year with a lot of Speed IF competitions in it tends to bulk out (in a numerical sense) with dozens of tiny games that are never intended for presentation to a wider audience. Likewise, IntroComp games might reasonably be lacking the features of a full and complete release. But it’s difficult to factor out those sorts of things without a different and more sophisticated counting method.

Or, you know, I could still just be totally wrong. That’s possible too.

17 thoughts on “More IFDB data”

      1. There’s a small bug in IFDB you could exploit: if you search for “published:2000 -tag:asdf” (without quotes) you get a list of all games published in 2000 that have at least one tag. (Apparently games without tags are never included in the search when searching by tags, even when the search is negative, so if you instruct it to exclude a tag that doesn’t exist you get a list of all games that have at least one tag.)

        Looks like 3144 games are tagged out of the 3725 in the database, which is quite a good coverage. This isn’t to say they’re all correctly or comprehensively tagged, though.

  1. Ouch. We haven’t been doing much lately. Perhaps because we have a rotating cast of brand-new authors in recent years, whereas in the 90s tended toward established authors (or at least fans of the old games from which they emulate)? There’s been talk of why authors don’t write a second game, but AFAIK that kind of talk didn’t much happen back in the day.

  2. Or maybe IFDB isn’t fully-enough tagged.

    Where is S. John Ross when we need him!

    (For the unwitting: Ross ran a “drive” to get more ratings & reviews into the IFDB when it was still fairly shiny-new.)

    1. I just ran across this poll of his:

      “As of the founding of this poll, the IFDB has only seven games with the ‘amnesia’ tag. I don’t buy that for an instant.”

      It’s up to twenty-one now, but I still think that’s probably low. (Which may also suggest that under-tagging is a serious problem for trying to draw any serious results here.)

      1. Yeah, I especially think that may be an issue with things like hints — people tend to tag games because they have what are considered to be memorable or exceptional traits, so if hints have become very commonplace, maybe they’re being undertagged for that reason.

        But, again, I could be wrong. It feels like the plots for “single room” and “puzzleless” have enough entries and enough of a distinctive shape that they probably mean something even if there are a few outliers that got left out of the tagging.

        On the other hand, when I ran numbers on the tags “linear” and “conversation”, the results were so tiny and chaotic that there didn’t seem to me much of use to say about them.

  3. Hi-ho :)

    I still do occasional pokes and prods to try to get folks a-tagging (even when my polls aren’t obviously labeled as tagging drives, I’m still quietly using them that way, running around like a phantom tagger in the night).

    But that’s a big ol’ database full of games, and I think “under-tagged” does no justice to the vast and underly underness of the taggery. I try to do a little each week … and I fail. But if we all went in and failed a little together, it’d be great for cool stuff like Emily’s trying to do and just cool in general for finding more little gems to try out.

  4. Research like this seems to assume that people are using a common set of tags. Maybe it’s just that I haven’t tagged anything since the early days, but my memory is of tagging to suit. Certainly on delicious & librarything my tags are idiosyncratic. Are the modern IFDB regulars coming to a consensus ontology?

    1. Consensus remains elusive, alas. For example, there is a “single room” tag … but there is also a “one room” tag AND a “one-room” tag.

      For another example, you can find each of the tags “sf,” “sci-fi,” “science fiction,” “science-fiction” and “sciencefiction” in use across the database so far.

      In many cases, games are redundantly tagged in a way which helps combat the problem (I’ve done a bit of that myself over the past year, though I’d prefer some way to tag a tag to be absorbed by another tag), but it’s all a bit haphazard (at this moment, for example, Alabaster is tagged with two of the three “single room” equivalents).

      1. This is true, though, for what it’s worth:

        a) My “maze” stats include the “guided maze” tag as well (which is why those numbers in recent years aren’t absolutely zero;

        b) For “single room” I did have a look at some of the things tagged with “one room” and add the single room tag to these as well before counting. I seem to have missed out on “one-room” with a hyphen, though. (Alas.) I’ve just gone back and fixed that, but there were a grand total of two rooms labeled “one-room” and not “single room.” So it’s not a very large anomaly.

  5. Yeah, it’s definitely the case that the variant tags are often an itsy-bitsy portion of the whole, and in this context I don’t think a tag here or there is anything to worry about. Apologies if I sounded negative (not at all the intent) … I was just agreeing with Tom H’s general observation about the do-it-yourself nature of tagging. It’s one of a handful of ways in which the IFDB tags are still (and perhaps inevitably) “not living up to their full potential,” as my English teachers always said …

    Also, it’s one of those issues for which any fix is bound to be controversial (I think we’d all agree that “dungeon crawl” and “dungeon-crawl” are meant to be the same thing and could be safely merged, but it’s easy to imagine lots of borderline cases where things are kinda-sorta synonymous except when they’re not …)

    In the spirit of the IFDB, the best fix just might be some kind of gentle automated suggestions that pop up when tagging, rather than any kind of forceful editorial hand. But I dunno; I have only a vague, user-end impression of the IFDB’s guiding philosophies.

  6. Another problem with IFDB tags is that everybody can add them, but as far as I’m aware, nobody can remove them except the person who added it (and the site administrator(s), I guess). It doesn’t make me want to add too many tags, because if someone (myself included) adds an incorrect tag by mistake, it’ll stay there and will pollute the search results. On a wiki, at least, you can fix mistakes, or argue about them and hopefully reach a consensus on the Talk Pages, but there’s no such system on the IFDB…

    It’d also be good if we could add to each tag a text describing exactly what it’s supposed to mean, because it’s certainly not always obvious, at least for me.

  7. The problem here is that this doesn’t tell you anything about whether audiences have responded to the changes. For example, more user friendly features only matters if players found those things more user friendly. Did these increase enjoyment of the games? Did more people start playing once you had these features in place? There may be no way to know this, I realize. But the inclusion of features doesn’t matter much if you can’t measure the impact of those features.

    This is a lot of good quantitative stuff but unless I’m missing something, you have no recource to anything that indicates a qualitative metric.

    1. It’s true that we don’t have a good way to measure how people are responding to the changes.

      However, there is *some* data about what stumps players and gets them to stop playing, which has fed into our ideas about what user friendly features we should include to start with. (Not a lot. But some.) See Aaron Reed’s articles on this — in fact there was one just recently,

  8. Regarding tags: ideally, the default would be “unknown”. One could then tag “maze” or “no maze”, If a set of suggested tags with this behaviour were added, they could slowly replace many of the more popular existing tags, while giving greater clarity.

    Regarding counting: if every game were weighted according to its size (minus the size of an empty game in that system), speed-IF wouldn’t do that much harm. Supposing that code size correlates rather closely with amount of exposure when playing (play time, amount of text encountered, ..) it seems fair to give larger games a bigger influence in the stats.
    The huge number of authoring systems (including ad-hoc code, as in the classic Adventure) makes this difficult, but for Z-code and TADS it might be worthwhile.


Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s