@mmccue Going a step further, Feedbin could do this for relays. Pull in the firehose, run categorization, and surface trending posts to Feedbin users who subscribe to those topics.

This gets into #fedilytics territory, which People Have Opinions about. Best to be transparent and opt-in for publishing instances.

Perhaps Feedbin hosts a relay with the explicit policy that relayed posts will be promoted this way. Useful to have a professionally operated relay with a clear business model.

I want to use #Dolt as the backend for an Fediverse server populated entirely by #bots. The clone/pull semantics will help me keep an analytic copy of the DB.

If it goes well and I can open up the instance, branches will allow people to pull data on their own bots.

The primary goal: experiment with #fedilytics and #GenerativeAI.

Secondary goal: make progress on #Fediverse account portability.

@haxing_ninja

Hey @PeterBronez @jaz I appreciated your thoughts on #fedilytics yesterday at #DecentSocial.

Here's some user feedback on post statistics from @steventdennis .

Was anyone working on these?

https://journa.host/@steventdennis/109853488902246403

Found a great article summarizing the state of #Fediverse #Governance https://notes.smallcircles.work/_FF2CKvcRy62RrygWkOWHA#

I discovered this via the #SocialHub forum, which apparently was a coordination point several years ago. They are considering the role of that forum going forward: https://socialhub.activitypub.rocks/t/poll-socialhub-scope-and-purpose/2843

If you care about #ActivityPub #activitypubdev #fedilytics consider chiming in!

Regarding https://www.theguardian.com/news/datablog/2023/jan/08/elon-musk-drove-more-than-a-million-people-to-mastodon-but-many-arent-sticking-around all I can say is

(1) #Fedilytics are hard, but blithely quoting https://api.joinmastodon.org/statistics is just ignorant & lazy.

(2) “First they ignore you. Then they ridicule you. And then they attack you and want to burn you. And then they build monuments to you.” - Nicholas Klein

https://www.snopes.com/fact-check/first-they-ignore-you/

@slashdot @clmerle I’m not confident in the analytics behind this take.

1) #fedilytics are hard because (a) it’s a distributed system and (b) the users seem to be allergic to all forms for data mining

2) instance admins, who have the best data, are not reporting this drop off. E.g. @DataDrivenMD argues here:
https://mastodon.fedified.com/@DataDrivenMD/109666544283809617

It’s fascinating to watch the backlash against basic Fediverse analytics Specifically, it’s just wild to me that:

(1) A sizable group of Fediverse users feel very strongly that there should be ZERO data analysis, but…

(2) The platforms have piles of APIs, many unauthenticated, and the fundamental ActivityPub protocol is extremely chatty

The technology does not align with the user’s expectations, thus drama. We need @spritelyinst to implement https://gitlab.com/spritely/ocappub #Fedilytics

@timbray @jnm @pixelfed related:

Megaface was a facial recognition dataset created from CC licensed Flickr images. It was ultimately decommissioned due to licensing objections: https://exposing.ai/megaface/

Discovered via this HN discussion https://news.ycombinator.com/item?id=34213036

#fedilytics

@Doug_Bostrom @alex agree that bad actors are doing this already and won’t stop.

Probably the solution will be to have an opt-in at the instance level. Instance gets better local search. Optionally they can join a pool of instances that share a search index. Sponsor reserves the right to do cross-pool analytics, perhaps with some restrictions.

Critical to deliver value to the instances or it won’t fly.

#fedilytics

@hisham_hm @harlanh do either of those options feel viable to you?

What principles would you want to see in a #fedilyticscharter ?

Are there any good analytics focused code of conduct documents we could reference?

#fedilytics

Alternatively, we could anchor a #fedilyticscharter on the consent-based communication model that @cwebber wrote for @spritelyinst

https://gitlab.com/spritely/ocappub

@hisham_hm @harlanh
#fedilytics

One option is to fork the #SlocanStatement and extend it with a section on analytic applications (contrasting with the current focus on transactional applications) to create a #fedilyticscharter

https://slocanstatement.org/

@hisham_hm @harlanh
#fedilytics

@hisham_hm @harlanh not at all, thanks for the thoughtful comment.

Perhaps it would be productive to write a #fedilytics code of conduct? A #fedilyticscharter?

This would provide a nexus for debate about constitutes respectful data science on #Fediverse. Then practitioners can follow the code and we can call out companies that violate it.

There are already a few proposals like this, see https://writer.oliphant.social/oliphant/mastodon-handy-links-page#proposals-and-policies

@harlanh Honestly people are quick to hate on algorithms and #fedilytics in general, but it’s because they’re used to being exploited by them.

Responsible data scientists can build respectful tools that empower principals* with affordances for thoughtful, constructive, and kind participation in authentic communities.

*Like a “User”, but not exploited

@harlanh I haven’t dug in yet, but I believe it’s based exclusively on post data. Adding social network features could be useful.

@bentomn has been thinking about using social networks to predict the information value of boosting posts. That is, has your network already seen this?

A stand-alone tool like Mastodon Digest might be a good place to experiment with this.

Discussion here: https://hachyderm.io/@bentomn/109547838770700017

#fedilytics

@harlanh yup!

👉🏼 https://github.com/hodgesmr/mastodon_digest

by @MattHodges

Background: https://mastodon.social/@MattHodges/109451111674479285

#fedilytics (I am trying to make this hashtag happen, intent is all things Fediverse data science)

@TCsHappyPlace @mattblaze there are a few #fedilytics services that do this:

https://fedi.buzz/ ➡️ trending hashtags + which instances they’re trending on

https://feditrends.com/ ➡️ links that are trending on the Fediverse. Has an RSS feed, which is nice.

Weaknesses:

1) it’s non-trivial to follow the trend upstream to the conversation

2) no published methodology, results could be random or manipulated

a #fedilytics find:

@Chartodon is a bit that graphs reply trees of fediverse conversions.

@Chartodon @ColinTheMathmo @weex nice #fedilytics example!

@digichelle @admiral I just noticed this the other day. Unexpected, but I kinda like it. It’s saying “these people in your neighborhood also like this thing.” It would make sense to load the other followers and perhaps hide them behind a fold.

This starts to touch on #fedilytics … soon enough someone will start doing social network analysis on the fediverse. You can have follow suggestion bots, etc. Might clash with established privacy expectations…