I do feel like I have seen this tried before several times. So studying what happened with those initiatives might be informative. I would suggest focusing on tropes that certain people "auto-buy" rather than generic ways these books could be described but that are not strongly tied to purchasing choices. e.g. people auto-buy specific tropes like mpreg or centaur hero more than general common types like elemental magic or "chosen one" hero.
I wonder if maybe that could just be a filter that could be applied? I feel like review sites/e-commerce sites already do an okay job of making those kind of connections. it seemed to me this wasn't necessarily intended to be part of an e-commerce site selling books so much as its own thing, an independent resource.
Certainly systems like this have been done many times before, and failed. While this is a very, very tiny data-gathering page she has put up, I agree that the limitations are frustrating. I suspect that if anything like this will be ultimately successful, the tags need to be crowdsourced. That eliminates Liosse de Velishaf's concerns about the subjectivity of tag meanings. Language is inherently subjective--as is the interpretation of fiction--but if the crowd has come to a consensus about what something means, then that's what it means.
I think I mentioned on the page that I thought at least some sort of crowd-sourcing feature would be necessary. I posited the definitions because she seemed set on curated tags, and she said in response to my comment that she had considered tooltips/parenthetical elaborations.
I do agree that crowd consensus has the potential to be more effective than curated tags.
Or it might all disintegrate like porridge in dishwater.
Anyway, this project is up my alley, and I thought I'd boost her signal. Will do the same for any similar projects I come across, because something like this needs to happen, and sooner or later someone will get it right.
Given my own experience proposing a similar idea a couple years ago, I suspect there is a good chance it will become porridge in dishwater. I'd think that there'd need to be a jack-of-all-trades involved in such a possibly niche idea in order to maintain steady progress--someone who could be the developer, the planner, and maybe even do a basic database of a few-thousand titles manually to get it up and running and ready for testing.
I think TVTropes does a good job of categorizing by tropes, although it lacks the rec function built on top of such a database.
I might look into re-opening my old version of this depending on how successful or not she is. I can do the development myself--unlike her, I believe--but she has a major advantage in terms of visibility, since my online presence has been fairly weak of late. And I'm doing other projects currently. But I'd definitely like to see some strong attempts at something like this.
Interesting idea. I've left a comment. The problem with ideas like this is that they need to get a certain amount of critical mass before they take off. And it would be very easy for Goodreads or Amazon to do their own version if it looked like being a success.
That's a major danger to any individual developer, and there's the added problem that because of their user/customer base, Amazon/Goodreads could make a much-reduced/crappier version and come out ahead anyway.
And then there's the problem of spamming. How would you stop authors from flooding the system with sock-puppetry?
A good idea, but not easy to implement. Not easy at all.
That's a major issue. It might require either a user account system, which would have a lot of other benefits anyway, or perhaps a delayed-addition feature, such that there'd be a way to do some sort of moderation and delete spam before it affects the rankings.
Well, a big problem, and one that is already pretty significant on Amazon and Goodreads is that when you let people tag things themselves the tags become meaningless. There's been some discussions before about genre that show this. People will make up new labels, that may or may not be a different label for something that already exists, and people will misapply labels because they either do not know what they mean, or disagree with how they are used for whatever reason.
I think you could include in the ranking algorithms a way to ignore outliers like that. You could also maybe have tags that the database hasn't seen before be looked over by someone before they're permanently entered into the system. But it's still a major hassle that would have to be dealt with.
The idea itself is a good one, but I have a hard time seeing how it will ultimately become a useful tool for readers looking for books.
I think it could become useful, but I agree there's a lot of stuff that needs to be worked out.
And what's to stop someone like Amazon or Goodreads from employing tags to limit access to certain works?
I think this particular system is not intended to be part of a major online book retailer, but that's also an issue that would have to be looked out for.