#social 2021-05-18
2021-05-18 UTC
xmpp-social joined the channel
# 
michaeljob HIV/AIDS vaccine: Why don't we have one after 37 years, when we have several for COVID-19 after a few months?
# michaeljob https://flip.it/6KSPkh
xmpp-social, timbl and naruto joined the channel
# naruto hi, can I ask what search+discoverability protocols are available for activitypub? As far as I can tell, the spec doesn't say anything. All the implementations that I've seen appear to use nodeinfo and/or webfinger. Is there anything else, for example for searching arbitrary object on a remote instance?
xmpp-social and BitBot| joined the channel
# nightpool[m] naruto: no, nothing else is common
# nightpool[m] Mastodon thought about implementing a search protocol, but we believe universal full text search is too harmful to discussion
# nightpool[m] Therefore, Mastodon only allows you to search posts that you've interacted with before, such as posts you've previously liked or reblogged
# naruto nightpool[m]: thank you a lot for your feedback. Basically, Mastodon allows only searching by direct URL or webfinger (actor@domain.tld), correct?
# nightpool[m] no, mastodon also has full text search, but only for posts you've interacted with before
# naruto yes, sorry, I mean wrt remote objects
# nightpool[m] well, they're still remote objects
# nightpool[m] they're just cached locally
# naruto arbitrary remote objects whether irregardless of their "previously-seen" status :) (I need to be more precise)
# nightpool[m] yes, mastodon generally only allows you to interact with objects it already has in local cache
# naruto nightpool[m]: do you know about other software/implementations? Like, peertube, fixelfed, something else? Do they all behave like mastodon?
# nightpool[m] or look them up by URI or Webfinger, but that's pretty uncommon
# nightpool[m] I know there are a few third party apps that attempt to spider the fediverse and index the content, but they're pretty socially discouraged
# naruto nightpool[m]: what did you mean by "we believe universal full text search is too harmful to discussion"?
# nightpool[m] let me see if I can find the issue threads about it
# nightpool[m] broadly, global full text search is a very, very powerful tool for harassment and surveillance
# naruto nightpool[m]: are you a mastodon dev?
dmitriz joined the channel
# nightpool[m] yes, although not a very active one :)
# nightpool[m] from our docs: https://docs.joinmastodon.org/user/network/#search
# nightpool[m] > Admins may optionally install full-text search. Mastodon’s full-text search allows logged-in users to find results from their own toots, their favourites, their bookmarks and their mentions. It deliberately does not allow searching for arbitrary strings in the entire database, in order to reduce the risk of abuse by people searching for controversial terms to find people to dogpile.
# naruto nightpool[m]: I understand the motivation for that but I don't think it's a good reason to not develop some kind of activitypub search functionality. There are plenty of good cases for searching something that hasn't been already cached by the local instance, or that isn't in my toots/fav/mentions
# naruto just my ¢2
# nightpool[m] I understand that, but you're going to get significant pushback from the current user base, which has strong expectations of privacy and local context. there's this fundamental notion of context collapse that comes with a global search tool that brings in potentially hostile or harmful audiences
# nightpool[m] see https://www.semanticscholar.org/paper/I-tweet-honestly%2C-I-tweet-passionately%3A-Twitter-and-Marwick-Boyd/0ee2da37957b9d5b3fcc7827c84ee326cd8cb0c3, which has been really informative of my thinking around this topic.
# nightpool[m] I don't think it's entirely an accident that the largest growing social technologies in the world are group chats and Discord servers, which have a very strict separation of contexts and no "global" search
# cjslep[m] In addition to nightpool's excellent commentary, i just want to add that: even if full-blown search is implemented, it would only ever be implemented in some software and not all, so any assumptions made building off of that functionality will always be "non-global, incomplete, partially-inaccessible". There is simply no such thing as a "global & consistent" view in federated networks.
dmitriz and Grishka joined the channel
# Grishka I just want to say that there can't be any privacy on viewing something in a distributed system like this. Global search is very much needed, both for posts and for users.
# Grishka The "local context" is also something that only Mastodon seems to want. In Smithereen, I want to blur the lines between instances as much as technically possible. The whole idea "instance is a community" won't and can't work.
# Grishka Switching instances is a costly endeavor. Participating in multiple different instances is too much hassle. Smithereen groups, on the other hand, have zero cost to join and leave, and you could be in as many of them as your server could handle.
# GregoryKlyushnikov Or, to put it simpler: you can't expect any "privacy" for something that is available without authentication.
xmpp-social, Grishka, lanodan and dmitriz joined the channel
# nightpool[m] anyway naruto (IRC) none of this is to discourage you from working on applications in this space! I'm just trying to explain some of the previous work and potential roadblocks to adoption. I definitely think discoverability and recommendation is a big need for Mastodon right now, but I think it's really important to address this in a privacy-preserving and context-sensitive way, which is something we're trying to do with our new user
# nightpool[m] recommendation features.
xmpp-social, hdenskove, getpaidyn81, GregoryKlyushnikov, ginahicks09, iwanaucamp, Jay, humont, SouthernWeldingRebel, bengo, daniel-uhlik, JayWake, salvehh30, enick_580, stoic_troll, SocioProphet, unnotchslayer, michaeljob and telnetcookie joined the channel