Some early results for feed reader behavior monitoring | Hacker News Reader

Some early results for feed reader behavior monitoring

rachelbythebay.com

37 pointszdw12 comments

Some background here, http://rachelbythebay.com/w/2024/05/27/feed/

why are all the links to the site http instead of https?

There is no redirect in place. They also have a http rss feed where all links are http. Maybe the poster uses this feed.

PSA: If you change the feed to https, it will also serve all links as https.

So even the well intentioned feed readers misbehave sometimes. I guess there is a spec but no proper test suite ever emerged?

On the flip side should RSS become a thing again thus incentivizing bad actors what recourse is there? Or just from poorly written ones?

Seems to be a missing piece.

With torrenting there is either a tracker enforcing ratio or gossip between clients - so bad actors are quickly flushed out of the system. Maybe that's a fertile place to draw inspiration from. I'm not sure.

If I were betting, I'd bet many of the supposed bugs are readers working around incorrectly-implemented servers. If the feed doesn't update, but it's because of a server bug, your users are still left unhappy. In a perfect world the readers would note which servers are implemented correctly and which you can't trust last-updated / etags etc... that's a lot of work.

What do you mean by "become a thing again?" That seems to imply that RSS went away somehow, but this is very definitely not the case.

I mean a widespread popular mainstream thing like the glory days of Google Reader. We are very very far from the peak.

I'm not sure it was ever a "widespread popular mainstream thing" -- most people never used Google Reader.

There is no ratio gossip in torrents. I'm aware of "private trackers" which track per-user ratio, and of clients that will prioritize sending to peers they've received the most data from, and of "super seeding mode" where the client attempts to track whether each peer is spreading around the data it gets, and won't send data to peers that don't - the latter requires the super seeder to be the only seeder. There are no other mechanisms.

I wrote ratio or gossip. Sorry it wasn't clear.

The private trackers not only keep score of per-user ratio but also have a shortlist of whitelisted torrent clients that are conformant. So you can't easily cheat.

The downsides are obvious. There is super seeder as you describe in an attempt to be decentralized/trackerless. Right?

Additionally, at least I was under the impression, inherit to the protocol itself is not so much the keeping it fair but rather multicasting the least shared chunks as optimally as possible across the swarm.

All of that is to say that a user agent misbehaving, maliciously or otherwise, will have a hard time. Trust but verify.

Whereas RSS appears to just.. trust. It is a more idealistic protocol from a more naive time period of the web. And I argue lacking in some sort of mechanism vaguely along those lines. Torrents came out around the same time but sort of evolved and refined these things.

> Bug: If-Modified-Since is only really valid if you were served it as a Last-Modified value previously. Readers are inventing values, or are sourcing them from the wrong layer of the stack. Don't do this.

This sounds more like a server bug. The RFC mention this behavior, but with the caveat that it doesn't work as well.

https://www.rfc-editor.org/rfc/rfc9110#field.if-modified-sin...

https://archive.is/cXPZs

https://web.archive.org/web/20240614033317/http://rachelbyth...

Reply on news.ycombinator.com