Ben, Hmm... I agree that the two approaches could work together, but, I'm not sure there would be very much benefit.
My thinking is that initially, your interests are determined by what feeds you explicitly follow. If it is a fb/g+/twitter - like service, then you'd normally add some friends right away. Things might get a little complex behind the scenes to make this work as smoothly as possible. Intuitively, there is a difference between seeing what a person shares and seeing what they follow. In this system, though, a "like" might be the same as a "share", and a "follow" might be the same as a "like" (so, liking a person or a feed might be how you follow them). This is the pure, elegant version. In practice, it might be better to separate these concepts somewhat, to optimise behaviour. (I don't have any examples in mind, but I realise that making all these concepts equivalent might cause some problems.) Another possible improvement is that people could maintain multiple feeds (similar to making g+ pages for others to see, in addition to your initial, personal feed). This would be similar to a tagging system. This way, liking something from Ms. X would not automatically cause you to see all material Ms. X likes; instead, the system could notice that you tend to like what Ms. X shares on her "technology" feed (alternatively, under the "technology" label). "Groups" would be the same thing, except with multiple administrators for the feed. I prefer this kind of user-driven clustering to automatic clustering. I'm not sure what the advantage to your approach would be. On Fri, Nov 9, 2012 at 6:04 PM, Ben Goertzel <[email protected]> wrote: > Abram, > > I think your suggestion is also good, but what it doesn't provide is a > partitioning into coherent, comprehensible groups of the comments on a > given topic. So your approach works for a user who is reading a topic > such that they've already rated a lot of stuff on related topics. Whereas > my approach creates a clean body of info for a new user approaching a topic > and its comments. > > For a brand new user with odd tastes, in your approach, they might need to > fish a lot (through very low ranked comments) to find the stuff they were > willing to rank highly... right? > > I think the two ideas could be used together... > > ben > > On Fri, Nov 2, 2012 at 12:17 PM, Abram Demski <[email protected]>wrote: > >> Mark, Ben, >> >> @Mark, >> >> Sorry I didn't see this sooner! >> >> This reminds me of an idea my brother had, which was that a feed should >> include some randomness in their ranking algorithm. You should not *only* >> be presented with things similar to what you've previously liked, because >> this leads to stagnation. That isn't to say that totally random articles >> should be thrown in every 10 items; the randomness should be integrated in >> a sane way, continuous with the rest of the ranking algorithm. But you need >> to see a variety. He justified this on evolutionary grounds, similar to >> your argument. >> >> It also reminds me of an idea I had about cell-phone autocomplete: if you >> base autocomplete on your data for texting habits, you get suggestions "for >> free", but they may be low quality (because the crowd's spelling accuracy >> in texts is low). It is beneficial to include higher-quality data (such as >> well-spelled text), and also to hand-craft suggestions (my cell phone's >> suggestions when I have not typed anything are all good-quality polite >> beginnings for messages, which I doubt are pulled from statistics). Making >> suggestions which are well-spelled and polite will increase the general >> quality of texts. Sometimes it is good to allow the crowd to speak, but >> sometimes it is good to guide them in the right direction! >> >> @Ben, >> >> I think you're actually making it more mathematically sophisticated than >> necessary, by talking about finding clusters in graphs. I proposed an >> algorithm (as a diaspora feature request, which got closed for being too >> speculative, understandably) which works more like this: >> >> You can upvote or downvote items in your feed as much as you want. (No >> 1-vote limit like on Reddit.) This alters a matrix of the people you >> subscribe to, so that we know how many up/down votes you've hit a >> particular person with. (Items could be ranked by more than just who posted >> it, but a source-based ranking would be fairly good I think...) Ordering in >> the feed is similar to the Reddit formula, (upvotes - downvotes)/age. >> However, the upvotes and downvotes are weighted according to how much >> you've liked a particular person (so if you "like" the same thing as Ms. X >> a large number of times, your feed will rank things based mainly on how Ms. >> X votes.) >> >> This could use some refinement, I'm sure, but the point is that I think >> we can work in a local way (essentially looking at dot products between my >> preferences and other people's) rather than looking at the network to find >> clusters. People can form their own clusters if several people mutually >> like each other's posts. >> >> More importantly, a "like" will now directly tune your personal feed >> (without a global crowd vote like on Reddit). >> >> Best, >> >> Abram >> >> On Mon, Oct 29, 2012 at 2:59 AM, Ben Goertzel <[email protected]> wrote: >> >>> Hey Mark Nuzz... >>> >>> I thought about the issue you raised (problems w/ Reddit comments and >>> their dynamics) in this email back in August, and here is my >>> suggestion... what do you think? >>> >>> >>> http://multiverseaccordingtoben.blogspot.co.uk/2012/10/avoiding-tyranny-of-majority-in.html >>> >>> -- Ben G >>> >>> On Mon, Aug 27, 2012 at 10:23 AM, A. T. Murray <[email protected]> wrote: >>> > Mark Nuzzolilo wrote on 27 August 2012: >>> > [...] >>> >> 1) Reddit is where the masses (at least in my part of >>> >> the world) go to talk about things. >>> >> >>> >> 2) Comments are artificially moved by a computer algorithm. >>> >> People influence the movement by clicking on buttons to >>> >> approve or disapprove of something that is said. It's often >>> >> done without much thought or >care, just a first impulse. >>> >> >>> >> 3) The voting tends to favor a strong bias toward certain >>> >> patterns, which are not directly designed or programmed >>> >> by Reddit, but are rather an emergent consequence >>> >> indirectly resulting from the site's >design. >>> >> >>> >> 4) The results in the voting determine who has a stronger >>> >> voice. The stronger the voice, the more people see it. >>> >> >>> >> 5) Based on these votes, the site gives people rewards >>> >> for being popular, and punishes them for being unpopular. >>> >> People who might have slightly opposing cultural beliefs >>> >> could become assimilated and become thinking more similar >>> >> to those who earn the most votes. People who disagree with >>> >> those who carry the most votes may find themselves without >>> >> many people to talk to, since people don't use forums anymore, >>> >> they use Reddit more. Less people to talk to means these >>> >> people do not get to share their ideas quite as much. [...] >>> > >>> > I take a lot of flak (Fliegerabwehrkanone) for my AI ideas, >>> > but I persist in "Redditing" because of useful sub-Reddits: >>> > >>> > http://www.reddit.com/r/artificial >>> > >>> > is the sub-Reddit on artificial intelligence. See also: >>> > >>> > http://www.reddit.com/r/programming >>> > >>> > http://www.reddit.com/r/technology >>> > >>> > Mentifex (Arthur) >>> > -- >>> > http://www.scn.org/~mentifex/AiMind.html >>> > >>> > >>> > ------------------------------------------- >>> > AGI >>> > Archives: https://www.listbox.com/member/archive/303/=now >>> > RSS Feed: >>> https://www.listbox.com/member/archive/rss/303/212726-11ac2389 >>> > Modify Your Subscription: https://www.listbox.com/member/?&; >>> >>> > Powered by Listbox: http://www.listbox.com >>> >>> >>> >>> -- >>> Ben Goertzel, PhD >>> http://goertzel.org >>> >>> "My humanity is a constant self-overcoming" -- Friedrich Nietzsche >>> >>> >>> ------------------------------------------- >>> AGI >>> Archives: https://www.listbox.com/member/archive/303/=now >>> RSS Feed: >>> https://www.listbox.com/member/archive/rss/303/7190161-766c6f07 >>> Modify Your Subscription: https://www.listbox.com/member/?&; >>> >>> Powered by Listbox: http://www.listbox.com >>> >> >> >> >> -- >> Abram Demski >> http://lo-tho.blogspot.com/ >> >> >> *AGI* | Archives <https://www.listbox.com/member/archive/303/=now> >> <https://www.listbox.com/member/archive/rss/303/212726-c2d57280> | >> Modify<https://www.listbox.com/member/?&;>Your Subscription >> <http://www.listbox.com> >> > > > > -- > Ben Goertzel, PhD > http://goertzel.org > > "My humanity is a constant self-overcoming" -- Friedrich Nietzsche > > *AGI* | Archives <https://www.listbox.com/member/archive/303/=now> > <https://www.listbox.com/member/archive/rss/303/7190161-766c6f07> | > Modify<https://www.listbox.com/member/?&;>Your Subscription > <http://www.listbox.com> > -- Abram Demski http://lo-tho.blogspot.com/ ------------------------------------------- AGI Archives: https://www.listbox.com/member/archive/303/=now RSS Feed: https://www.listbox.com/member/archive/rss/303/21088071-c97d2393 Modify Your Subscription: https://www.listbox.com/member/?member_id=21088071&id_secret=21088071-2484a968 Powered by Listbox: http://www.listbox.com
