I can't say that about my current job - but the sign of being an effective anything is that when you leave, on your own terms, folks 1. Miss you 2. You've passed on skills and knowledge to the folks who you trained up to replace you.

I mean, I can't imagine any SE dev or SRE, no matter how introverted go "Nope, I'm not turning up for meetings"

and I don't think anyone knows the entire codebase for anything but the most trivial system

or knows everything

not E90 communicator...

@MetaAndrewT. pretty sure the E series was crackberry clones

Happy 4,000th day anniversary, Stack Exchange!

As far as I can tell, similar-questions ("questions that may already have your answer") look now different then they used to.

IIRC there used to be similar questions in the sidebar and related question - only titles - under the title field. Now I see nothing in the sidebar and the questions shown under title are nt only titles, but more akin to the question list on the site.

I have searched a bit in "Recent feature changes to Stack Exchange" but I did not find when this was changed.

Was that part of responsive-design?

@Martin I think that is an A/B test. Let me find a similar question ....

Oh, I did not expect that.

I saw a recent Glorfindel's question: New “Similar questions” widget is not responsive enough

@Martin yeah, that is the feature

However, I don't see any mention of A/B testing there.

This is how the window with similar question looks to me: i.stack.imgur.com/0DBPt.png

Basically as in Grofindel's screenshot - just that I do not view it in a narrow windows, so I see also the questions.

@Martin found it: meta.stackoverflow.com/questions/384710/…

Thanks! It did not occur to me to search in meta.SO.

No problem, I happen to be active on both.

@Martin oh ... the MathJax ... I'm pretty sure the UX designer didn't account for that ...

Do raise a bug report about it.

There is a somewhat similar report on mso: meta.stackoverflow.com/questions/386796/…

MathJax is displayed fine in the list of similar questions, isn't it?

Yes, but due to its nature it can take up lots of space

I don't think it's avoidable.

That might be the outcome, yes

It's the same on the main site. Since the question preview takes a fixed number of characters, size of the question preview might be rather different depending on the content.

BTW I arrived to this from a question on Mathematics Meta: A suggestion to improve 'similar questions' suggestions..

In fact, I thought that tags are already used there.

But I see that in the post you linked, the most upvoted answer asks to used tags as a criterion.

yeah, tags are not used there. I think I have inspected both network calls and the client side code before on that feature and it only takes the title content as input.

WE have suggested, also in the beta phase of the question wizard to ask for tags first

so you could use the tags in next steps and offer better contextual guidance

Well, if we believe the post on meta, the tags should be used at least for the question in the sidebar. This is a quote from the linked post:

> every 45 seconds while you compose your question, we will query for Similar Questions based on the title, body, and tags you have entered

Hmm, let me check again then

AFAIK, the one under the title only queries based on the title only. The sidebar on the right... not sure how it works, but probably the same as the current related question.

@rene Here is somewhat similar bug report about size due to MathJax in the list of questions: Post with many lines of display math takes up most of the Questions page

But well, the tags are entered last, and the questions are already shown before that...

It depends on askers habits. (In fact, many users use /ask for searching rather than for asking questions. If they think/know that it takes tags into accounts, they probably enter the title and the tags first.)

Oh, the similar questions on the right sidebar has gone, it's moot...

@Martin I only see title go into the ajax call. None of my tags, nor body content is send to the server.

So it's possible that the post on meta is misleading?

Of course, it was posted 8 years ago, some things might have changed.

@MetaAndrewT. You mean as the part of the recent A/B testing mentioned by rene? Or were they completely removed at some point even before that?

And now since the Related Questions was renamed to Similar Questions, and the past Similar Questions was gone... confusion ensues...

I believe since A/B testing

@MetaAndrewT. in the ask question wizard tags are now asked first

@rene ah, that's an improvement

they are not used for similar questions

@rene ah, that's not an improvement...

I don't really now whether MathJax was rendered in the "old" version of "questions that may have answer" or not.

Now I found this post with a screenshot: Why “Search” is inferior to “Ask Question Search”? - so it wasn't rendered.

@Martin I don't see any traffic every 45 seconds so that makes that I think that feature got ditched.

@rene You probably meant to make a post on Meta Stack Exchange, not in the meta.SO thread about A/B testing, right? (This does not really concern SO, only the sites with MathJax enabled.)

@Martin yeah, this involves multiple sites across the network so Meta Stack Exchange is the right venue. SO proper doesn't have MathJax enabled.

I have posted on Meta Stack Exchange, as you suggested: MathJax in the list of similar questions (when asking questions).

So I accidentally found myself watching the entire Princess Bride movie last night. I'm not entirely sure how that happened, but I wasn't complaining.

As in, watching it again.

Once upon a time...

.. and they lived happily ever after. The end.

It's a good movie

One of the best movies.

→ 11 messages moved to Chimney

@rene why is the y in Day not linkified . . . that's so weird.

Is your da 24 hours?

potential indicator of spammy link...

totally intentional ...

Whoa that explains how some of the stuff goes unnoticed

I know they might get me for this, but the flower's scheme is finally revealed and I will denounce him in publ . . .

marshmellow's room gets raided by flowers in baklavas. Or balakavas.

If I had been in a jolly mood I might have linked each letter separately ...

I'm having a bunch of sour cherries. They remind me of you.

bitter sweet

→ 7 messages moved to Chimney

> please clicks here

the spam gets better ...

@rene I clicked there.... nothing happened :(

@Tinkeringbell I hate to tell you, you're now on candid camera ...

Camera? Good! Maybe you'll finally get less blurry, rene ;)

@Olivia it is a cheap camera with even cheaper focal stuff (calling it lens would be rude towards Canon, Nikon, Carl Zeiss)

Cool. A 3D CG recreation...

O.o

@Glorfindel Hey, just wanted to let you know that the Broken Image Repairer script picked up this post, changed the link to the Wayback Machine, and moved the image to imgur. It turns out that the original link still seems to be working fine, so I reverted the main link to the original, but left the image on imgur so that it will stick around if the original ever does break.

@Glorfindel I'm guessing the original link's server probably had a transient outage (it is in Missouri, which currently has a Tropical Storm hitting it, after all,) but it's working now. Nice script, though! The effort is appreciated!

@reirab thanks for thoroughly checking the edit. When I now simulate editing the post, it would leave the link intact. It could very well have to do something with Barry. Good catch!

@Glorfindel I was going to post a comment here that the script should compile a list rather than making the changes immediately and revisit 24 (or longer) hours later so sites that are temporarily down have a chance to recover on their own. In either case (working or not) rehosting is an excellent idea as it would short-circuit that problem.

@Rob that's a decent idea, but the script has made several thousands of edits so far without this being a problem.

An advantage of the Wayback Machine is that it's a snapshot and won't change, even if the site chooses to show other content on that URL. (Some image hosters are known for this behaviour.)

I mean, if a user would have encountered this post manually, they would have replaced the URL; so does my script.

Building a 24-hour queue takes a lot of time and energy, which I can also invest in minor improvements, like not removing URL fragments as in this edit - somehow this solution doesn't work for me

@Glorfindel - First sentence: But each edit was a bump, some unnecessary. 2nd paragraph: Yes, but by the logic of that argument everything should be Waybacked - a reason that is a problem (along with loosing everything if Wayback is unreachable) is that sometimes people edit their Webpages over the decades and provide up to date info, while the Waybacked link stays snapshotted at that one instance in time. 3rd paragraph: A user ought not to make a mistake, I agree your script is more help

than harm. Last paragraph, 1st link: that looks better in the "side-by-side markdown" view. In any event rehosting the images is excellent.

@Rob the script won't bump a post unless an image is broken; what happens to links is a byproduct but no reason in itself to edit a post.

Again, we're talking about 1% edge cases here. Automated moderating has its limits, I'm well aware of them.

K

And it's absolutely fine to discuss them; I've seen lots of suggestions by various community members that were easy to implement. This just isn't one of them :)

@Rob in the infant stages of the script, I noticed people were rejecting the edits, because they had no clue what was going on. Once I included the remark about 'side-by-side' (localized to Spanish, Portuguese and Russian), the rejections dropped by a significant amount.

This has turned out into a journey into human-bot interaction I never expected :)

A more problematic case is when the poster deliberately chose to hotlink the image instead of uploading it to imgur due to... any reasons (particularly: copyright infringement avoidance).

or user tracking

@MetaAndrewT. true, that's why the script adds attribution. I estimate that 90% of the users doesn't do that anyway, and only 1% checks if it's even allowed to upload the image (even with attribution).

Some people argue that they can include such images anyway on the grounds of 'Fair Use'.

If it's not stated otherwise, academic use is allowed with attribution

If you know of any sites with a license which is 'incompatible' with Stack Exchange's, I can modify the script to change the URL to HTTPS instead of HTTP, or link it via the Wayback Machine, instead of uploading them to imgur. All the script cares about is restoring posts to their original state, it doesn't really matter where the images are hosted.

Is there a downside to defaulting to wayback?

> sometimes people edit their Webpages over the decades and provide up to date info, while the Waybacked link stays snapshotted at that one instance in time

that's mostly for links, not images though

It's also slower, 99% of the times.

That's the main reason for using imgur.

No-update also applies to imgur

The main philosophy of the script is to preserve the author's intent. Which is almost always 'I have an image and I want to use it in my post. I don't care how.'

If I can spend a few hours to let the script show how that's done properly, I'm all yours.

I don't think I can help much

You're already helping by being a sounding board.

We were glad to contribute something to an almost perfect script.

@Glorfindel I think building a 24 hour queue wouldn't take "a lot of time and energy". To implement it: Seed a cache file with a single link. Alter script so two instances are started, one reads from the cache file and makes the changes (only one for day one) while the second instance writes a cache file 24 hours long. After 24 hours the script loops, going back to the first instance which reads the cache file as if it were the original feed (checking if the server is back up or if a human has

fixed it ahead of you. Execution continues to the second instance which creates the cache file for the next day. --- Really no changes in logic (algorithm) and simply a matter of spawning one additional thread. --- An improvement realized is that the server can come back up or a human can make a (theoretically) perfect fix.