I might be interested in photography but not in car photography. Being able to hide the “cars” topic would result to me not seeing discussions cross-posted in the “cars” and “photography” topics. Of course this could get complex or confusing to people. :)
But it's going to be a huge, huge use to people fine-tuning what it is they want to experience on the platform.
That will probably require actually being able to find tune "how much" you're interested in a given topic/subject/group. If it were up to me, every one of the tags would have the ability to be given a ++, +, -, or -- weighted rating (or, of course, left neutral) which would affect how close to the top of your "interesting" feed things floated. Couple that with being able to apply the same weighted rating to users, and you suddenly, magically, have a way to indicate that you are interested in conversations involving certain people, less interested in conversations involving certain people, and have indicated to the system a perfectly fine way to sort any list of content to be of most interest to a given user at any given time.
Heck, we can get even fancier than that. If we extrapolate some metadata to be weighted values that are user adjustable, you can have things like "active since I was last on" float to the top, "a conversation I was involved in" float to the top, or "posts which haven't seen activity in over a week" float to the top.
And if we take one further step and allow the user, may be behind an "advanced settings" control, to give weighted ratings to dyads of things, you can say "I want to read conversations in which Lex talks about role-playing games ("@lex"+"tag:role-playing games" +) but I am far less interested in him when he talks about politics ("@lex"+"tag:politics" --)."
Or, to express the original issue here, "I am interested in photography, but I'm really not interested in cars."
Which would result in more photography floating to the top and less cars in general, but if you're interested in cars outside of the context of photography, you might want something more like this:
Assuming that the photography tag is already boosted elsewhere.
And if we really, really wanted to get crazy, absolutely nuts. We could notice that this little abstract language that I conjured up out of nothing isn't limited to just referring to indexed tags or pregenerated metadata, but if there's enough computing power kicking around – it could just do a straight body text comparison, in case there's something that you really don't want to see.
I leave it to the more technically minded of the audience to think about what happens if we let a text interest tag be a regex.
Note that none of these things actually tell the system to hide that content for you. Instead, they accumulate to give a value to a given conversation causing it to appear higher or lower on the firehose stream of your feed. For a real, quality experience, they need to be coupled with an awareness by the system of what content that you have already seen/some kind of "marked read" mechanism so that you really aren't getting just an endless firehose but a stream of things ordered such that you are likely to want to engage with things higher up, and once you do or pass on them, they fall off the visual radar unless you specifically go looking for things you've already read.
Now, once you have a system for which we can make a reasonable guess about how interested you are in a given conversation, we can look at it and allow you to set a bottom threshold of "how uninterested I have to be before the system doesn't even show this to me at all." Some people, like myself, might make that bound really, really low (into the negative space) because we have the choice of looking at things which we might not be interested in because of participants or topics, but there might still be interested if we finish looking at stuff we are interested in. Some people might set particular individuals extremely, extremely low and never want to be involved in a conversation in which they are, setting the threshold in a medium sort of position.
This is the sort of thing that gives maximum flexibility to the maximum number of people, and would be a great thing to see on a platform like Cake where there is a ton of good content and you want to read as much of it that interests you as possible, but it is very scant about being able to report back "I think this is interesting to you because," and allowing the user to help sculpt their experience.
You know, just a few thoughts here and there.
@yaypie , thoughts from the engineering end? :)