robryk @robryk@qoto.org
I enjoy things around information theory (and data compression), complexity theory (and cryptography), read hard scifi, currently work on weird ML (we'll see how it goes), am somewhat literal minded and have approximate knowledge of random things. I like when statements have truth values, and when things can be described simply (which is not exactly the same as shortly) and yet have interesting properties.
I live in the largest city of Switzerland (and yet have cow and sheep pastures and a swimmable lake within a few hundred meters of my place :)). I speak Polish, English, German, and can understand simple Swiss German and French.
If in doubt, please err on the side of being direct with me. I very much appreciate when people tell me that I'm being inaccurate. I think that satisfying people's curiosity is the most important thing I could be doing (and usually enjoy doing it). I am normally terse in my writing and would appreciate requests to verbosify.
I appreciate it if my grammar or style is corrected (in any of the languages I use here).
Joined Nov 2020
It seems to me that you're saying that one can read indirect implications from the phrasing. I would like to posit that being more terse makes this problem worse.
When I joined fedi ~2 years ago, I suffered from the standard decision paralysis about which instance to join. In the end I followed a friend's recommendation, but before that I read rules (and federation policies/server blocklists) of a few instances and in many cases couldn't really tell what the rules mean. I can't point to exact instances I've looked at then, but I remember seeing rules essentially paralleling Stewart's definition of pornography (e.g. "do not make us add a rule"), which made me think that typical interpretation of anything banned in the rules was very expansive and that I was expected to read between the lines that anything that resembles a banned thing is also banned.
If you put yourself in such a position, you can see how such apparently-obvious clarifications are not necessarily that ridiculous. Perhaps the phrasing of this one is terrible, but then I expect it could be made better by making it more verbose and not terser.
@FrankSonntag @jsdodge @VergaraLautaro
Ah: the preview works _in the logged in view_, but also doesn't on the publicly-available single-thread page. So, there's probably no inter-instance difference here, but only some very weird difference in how Mastodon renders previews in different UIs.
robryk
boosted
has anyone seen a really good guide to writing good code reviews? someone just asked me and I couldn't think of anything
@FrankSonntag @jsdodge @VergaraLautaro
qoto.org uses Glitch + some other customizations, so I would assume that this changes _something_ in the preview generator.
I've emailed someone from that publisher (they don't have a published webmaster contact) about that issue. I would have ccd you, but couldn't find an email address.
You are right. I was conditioned to expect the name to be a link to the person's profile page and looked for that link either under the timestamp or under the "link" link (the one that points to the anchor of this comment in the very same page).
@jsdodge @VergaraLautaro @FrankSonntag
It seems that the problem is that the actual page declares that the broken URL is its canonical URL in HTML:
```
$ curl -s https://www.rutgersuniversitypress.org/the-cold-wars/9780813532950/ | grep canonical
<link rel="canonical" href="https://www.rutgersuniversitypress.org/book-details/" />
```
So, you're arguably doing the correct thing and the website is broken.
If you're curious, a request for an HTTP{S,} page specifies the part of the URL after the domain and a set of headers[1]. These headers make the request more precise (e.g. "what's the part of URL before the /", "what formats can the client handle" or "what (human) languages the client desires to see", "what cookies does the browser have for this server"), specify housekeeping-style things about the connection (e.g. "can compression be used", "is the client going to send another request on the same connection") and pieces of information that the server can make use of (in particular, User-Agent: "what is the software running in the client" and "what page did the user come from"[2]).
One of the possible responses means "this is a valid URL, but the thing it points to has a different canonical URL: please fetch that one instead (and remember this relationship)". It's called "permanent redirect" or "301".
I suspect that fediscience's previewer is making its request in a way that's different from a browser and e.g. qoto.org's previewer. I suspect that the website reacts to that difference in a silly way, by responding with a redirect to that broken URL.
What can be the difference that matters? The two most obvious ones are User-Agent header and the IP the request is coming from.
I've tried fetching the correct URL with some totally weird User-Agent and nothing weird happened, so it's either not that, or it reacts to a very specific User-Agent in a weird way.
A totally different hypothesis is that the previewer fetches that page using a simulated browser and some JavaScript in that page redirects the simulated browser to the broken URL.... and now I've found the culprit. Let me post about the actual culprit separately.
[1] I'm talking about "GET requests"; you can think of it as "normal" requests to fetch something as opposed to form submission requests.
[2] This used to be sent ~always by browsers in the past, currently the set of situations when it's sent and the precision of the value are greatly diminished. (FTR this header is called Referer, with the spelling mistake in it.)
Yup. It seems that it's fediscience-specific for some reason. Either a very weird bug, or the target website does the wrong thing when visited by fediscience's link preview generator (maybe if visited by something with a weird user agent? dunno; i'm too lazy at the moment to try figuring out what requests does fediscience's preview generator make).
Yup, repeatable.
@FrankSonntag, could you investigate why, when you look at the previous post at https://fediscience.org/@jsdodge/109485275489976147 the preview links to a different URL than the one mentioned in the post?
If you take a look at https://fediscience.org/@jsdodge/109480278297608998.json, you can see that the wrong link is nowhere present. So, it seems that fediscience.org managed to break the preview. Can you try replying with the same link (e.g. in this conversation) to see if that's repeatable?
(By image you mean preview, right?)
I confirm that this is the case for the post when viewed on fediscience.org, but it does not happen for the post when viewed e.g. on my instance. The broken link is https://www.rutgersuniversitypress.org/book-details/
Each instance generates the preview on its own when it receives a post. I suspect that when fediscience.org did that, the website in question issued a redirect to the broken link for some reason. Let me try to find the raw form of the post to confirm what was actually sent out though.
Would you mind adding links to comments (i.e. to the post on the instance the post is from) on your blog's rendering of fediverse comments? That would make it easier to respond to them.
@lauren The studies I can quickly find tend to look at a single social network and compare things (e.g. how different kinds of news spread) in that single network.
There are things like https://sci-hub.se/https://www.nature.com/articles/s41562-020-0833-x, which try to compare the same pieces of news spreading via various online social networks, or https://www.pnas.org/doi/pdf/10.1073/pnas.2102141118 that try to show how some properties of the network matter by examining toy models that vary along those axes. Sadly I haven't found (after a cursory search of a couple of minutes) anything trying to compare direct gossip with anything else. Do you know of some review publications in this area, or some better starting points for the between-network-comparison facet?
Do you know of anyone trying to study how the mechanisms used to spread information affect that?
I would be curious (and have no predictions) whether the situation is worse in traditional gossip networks, or social-media-style networks (where you see pieces of information published by people you don't know without anyone being an intermediary). I would also expect that we've had systems that were somewhat better at it (e.g. I would guess that late 19th/early 20th century system spreading scientific discoveries was better, even though it had many failures).
(The main reason I'm curious is that I would be very happy if we had a way of making false information not harmful without relying on an authoritative source of truth, so that the same mechanisms would work if no such source is assumed to exist.)
Trying for more verbosity:
I get why it might be bad for Twitter to claim that it's doing something that's e.g. counter to its consent decree.
Would it be bad for Twitter to falsely claim that it's doing something that's not counter to the FTC consent decree and other requirements?
I get that claiming to be doing something that does go against the expectation of what Twitter is going to be doing might be relevant. Do you think that claiming to be doing something that (a) is _not_ against requirements (b) is actually false might be important?
@lauren I thought both of them care ~only about what the company's doing, and not about what it's claiming. Am I wrong?
People in Switzerland who want to estimate anything around #covid risks might wish to be aware of https://ibz-shiny.ethz.ch/wastewaterRe/
@lauren Roughly what kinds of constraints do you see that would incentivize Twitter and/or Musk to constrain what they're saying publicly? I don't see why lying in public would be legally bad for them (I don't see whom this could be defrauding, and don't know of any legal constraints on lying other than[1] fraud or defamation).
[1] Well, and false advertising, which I wouldn't expect to apply here.
@lauren Frankly, I've admitted defeat on understanding Twitter's reasons for doing things, given that Musk does not seem to feel constrained not to mislead people. Thus, I really have no model of the process that generates Twitter's public statements now. Do you think there are some constraints on the relationship between their statements and reality that can be relied upon?
I enjoy things around information theory (and data compression), complexity theory (and cryptography), read hard scifi, currently work on weird ML (we'll see how it goes), am somewhat literal minded and have approximate knowledge of random things. I like when statements have truth values, and when things can be described simply (which is not exactly the same as shortly) and yet have interesting properties.
I live in the largest city of Switzerland (and yet have cow and sheep pastures and a swimmable lake within a few hundred meters of my place :)). I speak Polish, English, German, and can understand simple Swiss German and French.
If in doubt, please err on the side of being direct with me. I very much appreciate when people tell me that I'm being inaccurate. I think that satisfying people's curiosity is the most important thing I could be doing (and usually enjoy doing it). I am normally terse in my writing and would appreciate requests to verbosify.
I appreciate it if my grammar or style is corrected (in any of the languages I use here).
Joined Nov 2020
