#social 2017-10-19

2017-10-19 UTC
jaywink, jungkees, timbl, xmpp-social, KevinMarks, KevinMarks__, KevinMarks_, tantek and JanKusanagi joined the channel
#
Gargron is there any prior art on making hashtags in federated networks work like hashtags in centralized ones?
#
cwebber Gargron: what do you mean?
#
cwebber Gargron: like having a shared namespace?
#
cwebber Gargron: and searchability?
#
Gargron like... being able to browse them fully
#
cwebber I think GNU Social let you browse the hashtags of all the objects the server knows about, I assume Mastodon is the same
#
Gargron yes
#
Gargron but i want more
#
Gargron :T
#
sandro Not sure anyone's got the working. Options include: flooding, so the server really does know about it, assuming there's any connection via some user subscription to some other server; having a set of independent-from-each-other directories (search engines), the servers all report to and query for this stuff; using a DHT like Kademlia; ..
#
sandro All of this applies to general message searching as well, which I think is valuable, but I understand is a cultural issues for Mastodon. https://github.com/swicg/general/issues/3
#
Loqi [sandhawke] #3 Socially Acceptable Search
#
sandro Personally, my favorite solution at this point is: semi-flooding, along the social graph. Basically, users "shadow-boost-1" every post from someone they follow and "shadow-boost-1" every shadow-boost-1 from someone they follow.
#
sandro Shadow-boosts are visible to the machine, but not seen by end user. So then every server knows about posts 3 hops out the social graph, not just the usual one. That's not the whole world, but it's probably close enough.
#
sandro You could tweak the number of hops vs performance, of course, and you could let users tag each other with flags about whether the don't actually want to shadow-boost them, or want to extra-shadow-boost them.
#
sandro If I'm fallowing someone I hate, just to keep and eye on them, I don't want to shadow-boost them, etc.
#
sandro (arg, I wrote "shadow-boost-1" when I meant "shadow-boost-2" in the second occrance of that term above.)
#
sandro Hopefully this makes sense @gargron. I learned this approach from to Prof. David Karger, who was doing decentralization algorithms way back when.
#
Gargron i am interesting in this shadow-boost thing
#
Gargron *interested
#
sandro What would be helpful in terms of exploring it more?
#
puckipedia hm, could these posts be sent to an endpoint on the server-wide user thingy?
#
puckipedia also I'm thinking of an AP C2S extension that would allow users to be in control of their encryption keys etc
#
puckipedia basically, just before committing to the database and delivering, etc, the objects are returned to the client (with corresponding hashes, so clients can verify it or trust the server), and the client will then replay the request, but with the hashes attached
#
puckipedia also probably some kind of nonce, to allow for e.g. backdating published etc
#
aaronpk tbh the prior art is search engines
#
aaronpk google alerts and superfeedr trackers both provide a similar end result
#
aaronpk This is how I use superfeedr to run essentially a hashtag search but for my domain name and for “indieweb” https://aaronparecki.com/2016/02/20/5/global-webmentions
#
Loqi [Aaron Parecki] Enabling Global Webmentions with Telegraph and Superfeedr
#
aaronpk I like sandro’s suggestion for growing the network of what’s crawled