2009
The Favrd Situation
Stars and followers are a primitive form of reputation currency.
[Since stars have no scarcity,] the more you spend, the less [they are] worth to others. [If] you want your opinion to matter, you don’t want to look like you award too many stars or follow too many people. It takes some experience to realize this, though, and not to be impressed by the 200 stars on a lunchtoot from the Jonas Brothers, or by a social media expert’s four-digit Followers count when it’s dwarfed by his five-digit Following count.
[…]
Simply put, I think what happened to Favrd was that a new crop of users appeared who didn’t know how to value the currency, and thus they inadvertently devalued it.
[…]
I don’t know what to do about this. I like Favstar well enough, but it really replaces only those aspects of Favrd that were satisfying to validation junkies. I don’t see how a strong community can form around Favstar unless Tim makes it work more like Favrd did; a good start would be showing more than one page of recent tweets and adding a webcock filter. But I’m not sure a surrogate Favrd is the solution either.
I’ve chopped up @dwineman’s original post to contextualize the following, but you really should read the whole thing.
Twitter’s starring function is perfect, as-is: a simple, binary switch indicating “meh” or “yes”. Anything more would be too much.
There used to be a scarcity of funny, interesting Twitterers. One way to “subscribe” to their tweets was to follow them. Twitter’s built-in following function is the primary Twitter filter, showing you just the users you trust in advance to say things you want to hear. It’s a traditional sort of “contact based” aggregation function, and it’s a natural default.
Favrd (and Favotter before it) provided another kind of filter: starred content only. Instead of “trusted in advance”, the criterion was “approved of by others” (plus a few other rules). The effect was magical: friendships, laughter and mini-fame followed.
Favstar.fm both outdoes and falls short of Favrd. Outdoes because it is a more comprehensive, robust Twitter client, and not merely a passive list. Falls short because it enables star devaluation in the worst way. This is not Tim’s fault. It’s the fault of the star-happy mob whose informal group contract states “I’ll star all of you if you all star me.”
I have one idea for Favstar. Whether it would help depends on human behavior, though.
For each displayed tweet, Favstar currently shows first the number of stars received (“X FAVS”), then avatars of the first 16 users who starred it, and finally a button marked “X OTHERS” that shows the remaining avatars when clicked.
What if Favstar instituted a new metric: “Favstars”.
The difference between a star/fave and a “Favstar” is that you only get 10 Favstars each day (or maybe 20 or 30 or 40; whatever). After that, your favs are still recorded by the site, but your “votes” stop counting until midnight GMT, at which point your supply is replenished.
Other behaviours:
1. If you star a tweet from someone you don’t follow, it counts, even if you’ve already used up your Favstars for the day.
2. Instead of showing the total number of favs, followed by the first 16 avatars for each tweet, the site shows the number of Favstars (“X FAVSTARS”), then the thumbnail avatars of the first 16 “Favstarring” users for that tweet, and then finally a “SHOW ALL FAVSTARS” button. This displays (just like today) the avatars of all users whose “votes” counted as part of their quota.
Clicking “SHOW ALL FAVSTARS also transforms the button itself into today’s existing “X OTHERS” button. Clicking that button displays all the rest of the stars, as recorded by Favstar, but does not tally them up for you. Of course, you’re welcome to count them if you really want to.
3. All other site filters (100-star page, Popular page, etc.) are driven by Favstars, rather than “normal” stars.
The idea, of course, is to inject scarcity back into equation. Maybe ten Favstars is the wrong number. It probably needs some testing. I also think a daily “rollover” is probably appropriate. If I go away for three days, and sit down to get caught up on Twitter, maybe I should have the security of knowing I’ve got a maximum of two or three or four days’ worth of rollover to work with.
And by allowing users to Favstar posts from people they don’t follow, without it counting against their quota, the system discourages star-happy users from following 500+ people in the hopes of “star reciprocation”.
The result, ideally, is what Favrd used to be: A truly useful list of vetted Twitter content, reliably worth reading. In short: a great, alternative Twitter filter.
SUBSCRIBE