Been trying this out secretly for the past few weeks — a "catch-up" timeline page. Did whole bunch of iterations, where I myself is the guinea pig, but this time reaching a point where I just got to share it out in the public.

- NOT final design, still very experimental, only on my local machine
- Instead of algo timeline, this is focused on data-grid-like UI for configuring exactly what I want to see.
- This honestly could be another full-fledged app/client by itself 🫠

#Mastodon

Here's the screen *before* this. Specify how long back I want to fetch the Home timeline posts.

After a while, my usage slowly leaning towards this: grouping by authors. Feels refreshing browsing it this way. Faster to scan too.

Oh yeah, this is how it looks like on wider viewport. Kinda email-like.

#PhanpySocialDev #sneakpeek (not available on dev site yet)

Not sure where to put these, so experimentally placing them at the top for now, reusing the styles from trending news in Trending page. It's based on the post's `card` so some links are actually "quote posts" 😬

#PhanpySocialDev (not on dev site yet)

🤫 Quietly dropped on dev site #PhanpySocialDev . Feedback/bug reports welcomed 🙇‍♂️

Cool to see @laurenshof the first one to post about it https://indieweb.social/@laurenshof/111997146906858162

Laurens Hof (@[email protected])

Attached: 2 images holy shit catch-up just dropped on dev.phanpy.social this is soooo fucking good @[email protected]

Indieweb.Social
@cheeaun curious, is this more, less, or doesn't a make difference to resource usage on an instance? I guess highly dependent on someone's mastodon usage? Also, have you played around with different types of 'popularity' metrics? e.g. very rough one off of the top of my head, a post's likes / their follower count, as a way to not have things skewed to just large accounts.

@paulcox re: resource usage, you mean the API calls and the rate limit?

As for 'popularity' metrics, I didn't really play around with them. I don't think anything is really skewed here as you have complete control on how to filter or sort the timeline. Though the idea of sorting posts by follower count seems interesting 🤔

@cheeaun I guess both things like API calls, plus things like CPU on a server i.e. curious the impact on a server's hosting costs.
@paulcox for API calls, there's still rate limit and pagination limit set by the server. As for CPU usage, this is probably not much compared to the federation stuff 😬
@cheeaun I was thinking more of an aggregate popularity sorting e.g. giving a post a popularity score based on the number of likes divided by the number of followers. Kind of like a relative popularity of a post for the user. Interested if this is a better metric than sorting based on number of likes as posts from larger accounts are more likely to appear at the top and so I might miss particularly interesting posts from smaller accounts I follow.

@paulcox in my opinion, regardless of whatever scoring logic, there's always a chance or possibility that you might miss something.

So instead of intelligently rearranging the posts based on some score, Catch-up shows everything, provides email-like interface for fast scanning and lets you sort and filter in any way you like. It's more manual work but you have high visibility over the whole list and not having to worry that some posts are hidden from you.

@cheeaun yeah fair. I think an extra sort option based on a score would be useful, but I understand your reasoning.
@paulcox I know there are other attempts, not sure if you're aware of them. Fediview https://fediview.com/ (open-source) does this score-based sorting. There's also fedialgo https://github.com/pkreissel/fedialgo (open-source) and Murmel https://murmel.social/ (paid, free-trial).
fediview - The algorithmic timeline for Mastodon

Generate an algorithmic summary for Mastodon timelines