This was originally a slideshow presentation, but my slides weren’t anything special so I’ve left most of those out of this re-run of the talk’s presentation in favor of running the text itself. If this reads a bit odd in spots, it’s partly a reflection of that transition.

On Metafilter

I’m Josh Millard, aka “cortex“, and I’ve worked for the last four and a half years or so for Metafilter.com, where I’m a moderator and community manager and data nerd. I’m going to talk today about a data project called the Metafilter Infodump, but since all of us on this panel are discussing work related to Metafilter I thought I’d start things off by giving a little background on the site itself.

Metafilter is a community website started by Matt Haughey in 1999, as a place to share interesting links from the web and discuss them with other Metafilter users, or “mefites”. It’s a generalist site with no given topical, ideological, or demographic focus or mission statement. Joining the site has, since 2004, cost a one-time $5 fee. The userbase is majority North American but significantly international as well. English is the common language on the site.

The front page of the Metafilter functions as a sort of low volume link aggregator, with about thirty posts a day made by individual users of the site. Posts range in scope and tone from funny cat videos to encyclopedic examinations of medieval sacred architecture, from election scandals to indie comic art. Each post features a comment thread, which was a definitive feature of the site back in 1999 and remains the fundamental venue for members to interact—while anyone can read the site, only members can comment in threads, and for most regulars it’s the discussion in threads on the front page and other parts of the site that makes Metafilter an interesting place to hang out.

Aside from the front page and it’s link-sharing ethos, we’ve built other subsets as well; there’s Music, where users can post their own recorded work; Jobs, where users can list job openings or note their own availability for work; IRL, a meetup organization area; and Projects, a site for users to post about creative projects they’ve undertaken, since self-linking elsewhere on the site is discouraged.

But most importantly there’s Ask Metafilter and Metatalk, the two other big subsites; including Metafilter proper, these three subsets account for the bulk of the posts and discussion that takes place in the community. The subsets are often referred to by color nicknames, based on their background colors: Metafilter is “the blue”, Ask Metafilter is “the green”, and Metatalk is “the grey”.

Metafilter I’ve talked about already. In addition to “the blue”, it’s also often called “the front page” or “mefi”, all of which are useful in discussing that link-sharing part of the site specifically as distinct from the overall entity that is Metafilter-as-a-community-and-collection-of-subsites.

Ask Metafilter, “the green”, is a question-and-answer site launched in late 2003 after it became clear that there was a lot of desire within the community to have a place where mefites could specifically interact with one another as a relatively trusted group. Users can ask at most one question a week but can answer as many as they like; on the order of a hundred questions and two thousand answers go up daily.

Metatalk came along much earlier, in early 2000, about the time that Metafilter was making the transition from being Matt’s de facto personal link blog to being a small community with several hundred registered accounts and an active core group of dozens of daily participants. People were talking about the site on the front page — making feature requests or suggestions, reporting bugs, discussing site policy or user behavior — and there was a clear need to segregate all that metacommentary from the actual “here’s a neat link” business of the front page.

Today, Metafilter has about 70,000 registered user accounts, about ten thousand active users depending on how you define your terms (more on that below), fields about three thousand comments a day from logged in users, and sees something like 25 million pageviews on a typical month from users and web readers in general.

The Metafilter Infodump

So that’s Metafilter in a nutshell. Let’s talk about the Infodump.

The Infodump is well-structured, weekly-updated collection of data files that distill the bulk of user activity on Metafilter down to a skeleton of activity records. It contains information like the datestamp, posting user, and unique IDs of every post and comment made to each subsite, as well as category information where applicable and whether a given post was closed to new comments or deleted by a moderator. It also contains raw data about the use of the favorites system, a multi-use and sometimes contentious bookmark-slash-thumbs-up features that allows users to mark individual posts and comments as a “favorite”.

The Infodump contains mostly numeric data; it elides the actual text of comments and posts, partly to keep the size of the data files a lot more manageable and their structure simpler, and partly to get away from simply reproducing the entire content of the database in slightly altered form. It does however contain some text-related content, including information about every folksonomic tag associated with a post on the site, the title text of posts, the character count of every comment, and the actual userids and usernames of registered users.

(A parallel project I’ve been working this year, the Metafilter Corpus, is more explicitly focused on the text content of the site; currently it consists of a variety of 1-gram frequency tables for comment text across the site over the first eleven years of comments, which is about 457 million words of source text. I plan to expand that into higher order n-gram files in the near future.)

The files that make up the Infodump are generated by some fairly simple perl scripts I wrote that pull data directly from our database every Sunday night and sort and clean that data a bit before writing it out to tab-delimited, zipped-up text files. It’s a straightforward, fire-and-forget procedure that our code wizard Paul Bausch has running as a service behind the scenes, which is good for everyone because I’m pretty distractible and back when I originally had to run it manually it would get pretty out of date until someone poked me over email.

Before I came to work for Metafilter, I’d been fascinated for years by questions about how the Metafilter functioned, about what the actual quantitative answers might be to questions otherwise only answerable with speculation when they came up in discussions on Metatalk. I had pestered Matt Haughey on a few occasions for behind-the-scenes data, and had done some scraping of the site to collect data manually, but both felt like impositions and I’d hoped there was a better way to get at that stuff.

Other folks had been in the same boat as well; academic mefites and even folks outside the community had done research projects about the site, some trivial and some as serious as graduate-level theses, and much of that work was complicated by the need to either beg for or scrape data as step one.

The Infodump was an attempt to answer that need, to simplify data collection and make it easier for people, whether serious researchers or curious community members, to dig into this stuff via an accessible and reliable outlet straight from the site itself.

My co-panelists here will all have some interesting stuff to talk about along the lines of that serious research side of things, but I’ll give you some quick examples here of the more informal intracommunity navelgazing stuff our users have done with this data over the years.

Datawankery In Action

When should you ask a question?

I mentioned that Ask Metafilter has a limit on questions a user can ask: one question per week per user at a maximum. This helps keep the volume of questions manageable, but it also makes a question feel a little precious sometimes — you don’t want to waste it, you want to get the most out of it. And so one of the recurring questions in Metatalk is “when is the best time to ask a question?” Surely some days are better than others, or some times are better than others? There was a lot of speculation and folk wisdom about this for years, but the Infodump made it easy for some data-minded users to actually answer the question quantitatively.

And it turned out that the answer, to pretty much everyone’s surprise, was that it really doesn’t matter. If you’re looking to maximize answers given to your question, you can post literally any time of the day, any day of the week, and expect an average of 14 answers or so. This despite the Metafilter community’s majority demographic being in North American timezones and weekends being significantly quieter than weekdays.

One partial explanation: Metafilter has a heartbeat. Folks who visit the site more or less daily are likely to look at the things that interest them whenever they get around to visiting, and to catch up on what they’ve missed since the day before, whether it’s something posted noon yesterday or three o’clock this morning.

Beyond that, fewer people visiting during weekends or quiet hours might imply fewer answers, but it also implies fewer questions; while both kinds of activity drop measurably on Saturday and Sunday, a lower number of questions asked means that it takes fewer total answers to get a fair number or responses to all the questions that have been posted.

How many active users does the site have?

How many active users does Metafilter have? If I wanted to brag the numbers up I’d just point out that we assign userids sequentially and we’re inches away from userid 140,000 right now, but in actual practice that number is pretty much bullshit. For one thing, not everyone who starts signing up completes the process, and we reserve those partially-registered accounts in case the inchoate user decides to come back and finish signing up later. About half our signups are false starts, so we’ve got maybe 70,000 active users.

Except a lot of those folks don’t do much on the site. A large chunk signed up but never did anything; many others have only commented a few times and then disappeared; others still were once very active but have since closed their accounts or walked away or even in some cases been banned.

So how many for real, ACTIVE active users do we have? One of our members, FishBike, put together a nice graph that uses the benchmark that a user is “active” if and only if they’ve had some sort of site activity (posting, commenting, favoring) in the last month; by that metric, there were on the order of 7000 active Metafilter members at the time that he did these calculations, with steady slow growth in the active user population over time even as there was a smaller level of steady slow growth in attrition.

It’s not a perfect metric — folks taking a months-long break from the site or who only comment sporadically but long-term are treated as having left, hence the big red spike at the right end of the graph, and a month is an arbitrary choice in any case — but thanks to the Infodump it at least gave us an actual figure based on data and made it possible for a user to answer the question himself instead of relying on a mod running the numbers. And it’s not a terrible estimate, though it fails to account for lurkers; looking at internal site logs, we can tell that in fact it’s about twelve thousand distinct logged-in users who at least visit the site each month, so Plutor’s metric mostly just misses the large chunk who are actively around but don’t leave any traces in the Infodump by leaving comments or so on.

Are there cliques on Metafilter?

I said our favorites system is contentious; people disagree about what they’re for and how they should be used, and beyond that some folks worry that for example they encourage cliquish behavior. A clique as a qualitative thing is hard to test for, but numerically the Infodump makes it possible to at least look at who favorites who to look for interesting patterns.

And so we’ve had a few different looks at some of these ideas. One user analyzed mutual favoriting behavior to see what pairs of users favorited one another’s content more often than average. It turns out that, for whatever reason, there are definite pairings and groupings of people who favorite one another more often than the baseline average for a user who engages in favoriting, often correlating to notable real-life relationships or strong shared topical interests.

We also have a contact feature on the site, where users can explicitly set up one-directional relationships with other users’ accounts, with modifier tags like “spouse” or “muse” or “met” as in met in real life. That data’s in the Infodump as well. And so another user compared favoriting behavior to the user contacts data and established that in fact folks are far more likely to favorite content by people who they have listed as a contact on the site, by a factor of thirteen or so.

And lots more

Mefites have even built tools to make the Infodump data more accessible to folks not as practically equipped to get into hardcore DIY datawankery (which is the site jargon for this kind of info wrangling stuff). One user, Bill Zeller who went by “null terminated” on the site, had maintained for a while a web-facing SQL query interface for the Infodump data; more recently, Combustible Edison Lighthouse built the Infodumpster, a friendly fill-in-the-blanks engine for a variety of common query templates that lets folks play with the basics without needing to know anything about data manipulation.

There’s a wealth of further examples but I can’t go through them all; if you’re interested in seeing more, there’s a page on the Metafilter wiki that rounds up a lot of the user-driven research over the years and points to a number or discussion threads on the site; user FishBike maintains his own charts and graph page as well.

How To Build An Infodump

As far as I have seen, there’s not a lot of stuff out there like the Infodump, which is a shame because it’s not necessarily a difficult thing to produce. I’m barely literate with SQL and ask our coder Paul for help with basic stuff like JOINs a lot of the time; if I can manage to hack out some perl and some basic SELECT statements to accomplish this, pretty much anybody who runs a database-driven website can, and if a site wants to accomplish this but doesn’t feel confident about creating a data dump procedure internally it’s a pretty simple consulting gig to hire out.

The big challenge is more in deciding what data to make available and thinking about how that impacts your users. Private stuff needs to stay private — we don’t put the content of our users’ profile pages or their email or payment details into the Infodump, we don’t include internal stuff like data about the flagging system folks use to report problematic comments or posts, and we do provide a munge feature so that users who would rather not have their ID directly correlated with their site activity in the Infodump files can have that dissociated automatically. Ultimately, everything in the Metafilter Infodump is already publicly viewable on the site, and not everything publicly viewable or derivable is in the Infodump, but it’s still important to think about what your userbase might expect to be out there and in what form. In the case of Metafilter, we’ve talked about this stuff at length with the Metafilter community in Metatalk and provide a clear channel for feedback from users about any ideas or concerns about the data, and that communication is key to not having a lot of folks feel like they’ve had the rug pulled out from under them.

Communities Are Great Research Venues

It really is striking to me that even as cohesive social communities have become a very significant part of the landscape of the web, things like the Infodump seem to be uncommon at best. There’s just a lot to recommend it.

A tool like this is a benefit to the community itself in terms of transparency in administration and giving folks back what they’ve invested — users are the lifeblood of a community site, this is their activity and their data in the first place and they should have access to it. At Metafilter we offer users direct access to a flat dump of their own commenting activity and internal mail as well, for the same reason.

An Infodump is also good for a community’s sense of self-awareness and involvement in its own state of being, of its context and history — a data dump like this gives them the ability to get their questions answered, to explore the various introspective “who”s and “how”s and “what if”s that naturally arise in any self-organizing group of smart people, and it gives them the autonomy to pursue those answers without waiting on the ability or availability of the site’s administrators to do analytical work for them.

As a practical logistical benefit, it can help a site avoid extra unplanned load from the methodical (and sometimes poorly-throttled) full-page scraping that has traditionally been the sole avenue for data collection for curious users and researchers, and to host the data independently from the main site to reduce the impact of large downloads.

And the data itself is cleaner and more reliable — scraping is often a miserable experience for the person collecting the data, as what’s useful or pragmatic for live webpage layout in a cross-browser, cross-device context is often a hard-to-clean-up mess for the person only looking for skeletal data. A conference like this is a testament to the fact that people have been, and will more and more as time goes by be, looking seriously at what people are doing with their time on the internet as individuals and groups, and if a community can help make that process more practical by offering up tools like the Metafilter Infodump or the Metafilter Corpus I think that’s an important thing to do.

It’s my hope that more sites will embrace this idea; there’s lots of smart, introspective communities on the web worth taking a close look at and I think if the folks running those sites and the researchers who see the value in those close looks can talk to each other about these ideas, this sort of thing can become less a rarity and more just an obvious and sensible way to deal with this wealth of data.

But I play a lot of games and don’t mention it here much at all, so why am I bringing up this one? Because of a plate of beans, is why.

It goes like this:

1. I found out about Glitch because I work for and hang out at Metafilter, and a lot of folks on Metafilter are playing it. And so while my little yellow dude has been mining beryl and cavorting with pigs in exchange for steaks and harvesting bubbles off of bubble trees, I’ve been chatting with my fellow Mefites. And not only are the are a lot of us playing the game, there’s even a couple folks who work for Tiny Speck, the company that makes Glitch.

2. There’s a long-running joke on Metafilter about how we’re the kind of people who could overthink a plate of beans. It started as a joke someone made during a thread years ago in which people were earnestly deconstructing the performative elements of Alanis Morissette’s cover of Black Eyed Peas’ My Humps, and it caught on, to the point where “beanplating” and “to beanplate” are a commonly understood derived verb forms used to describe maybe-needlessly-in-depth analysis of one thing or another. There’s even a song.

3. Glitch has bean trees, on which grow beans. One can harvest beans, and eat them or use them as constituents in recipes. What Glitch hasn’t had is beans arranged tastefully on a plate.

4. Except, well, that now exists, thanks to one of those Mefite Glitchers who happens to work on the game.

Here it is:

Click on it and you get, as with most items in Glitch, a context menu that gives you some basic options and a special option unique to this item: overthinking.

So what happens when you decide to overthink your plate of beans? Well, overthinking isn’t something you want to just jump into without a little bit of consideration, so the game helpfully asks if you’re sure you want do this:

But being sure about overthinking isn’t something to settle on trivially, so after a moment it asks if you’re really sure:

Any seasoned overthinker knows that settling on being sure about something takes serious contemplation, and so, well:

And, well:

Given that energy slowly goes down over time in Glitch, I had to resort to frequent bouts of meditation with my Focusing Orb just to keep myself from starving to death while waiting for the eleventh confirmation screen:

I have no reason to believe this would end. But I haven’t got the patience to find out; in fact, I don’t have the lifespan to find out, because the pause between each confirmation step grows literally exponentially: the wait between confirmation prompt n and confirmation prompt n+1 appears to be 2ⁿ seconds, which means that while it only took one second for the “really sure” prompt to appear, it took four seconds for the “really, really, really sure?” to appear.

And because I don’t want to keep tying out big strings of “really”s, let’s just note that the number of times “really” appears in a confirmation prompt can also be expressed mathematically as n – 1, and call this value really_n.

So, as above, the wait on really₁ is 2^{(n – 1)} = 2⁰ = 1 second; the wait on really₃ is 2² = 4 seconds.

The wait on that last one pictured above, really₁₀, was 2¹⁰ = 1024 seconds, or about seventeen goddam minutes. Or about one and three quarter hours of in-game time.

If you wanted to get as far as really₃₁, you’d have to play Glitch for about 50 years straight, every day, without sleeping. And that’s a lot of bubbles and pig steaks to eat.

You’ll never see the really₃₂ milestone yourself; you’ll need to establish a trust for that.

I can’t think of a better in-game manifestation of this little bit of Metafilter in-joke humor, and it’s just a goddamned magical bit of internet wonderfulness that I have this silly item in my inventory on this game I didn’t even know about until I blundered into it a few days ago and lost my mind to it.

Thanks for the plate of beans, ericost. You’re a mensch.

The Dune Abides.

You can read through the thread to see the raw output as folks put it together. Fun real-time riffing, people pushing in a few different directions with it over a couple of hours.

Here’s a neatened-up arrangement of my take on it, in the general style of an IMDB “Memorable Quotes” digest:

The Big Atreides

PRINCESS IRULAN [voiceover]: Way out in the stars there was this fella… fella I wanna tell ya about. Fella by the name of Paul Atreides. At least that was the handle his loving parents gave him, but he never had much use for it himself. Mr. Atreides, he called himself “The Dib”. Now, “Dib” – that’s a name no one would self-apply where I come from. But then there was a lot about the Dib that didn’t make a whole lot of sense. And a lot about where he lived, likewise. But then again, maybe that’s why I found the world so darned interestin’.

They call Arrakis “the Spice planet”. I didn’t find it to be that, exactly. But I’ll allow there are some colorful folks there. ‘Course I can’t say I’ve seen Caladan, and I ain’t never been to Ix. And I ain’t never seen no Reverend Mother in her damned undies, so the feller says. But I’ll tell you what – after seeing Dune, and this here story I’m about to unfold, well, I guess I seen a purpose every bit as turrible as you’d see on any of them other planets. And in Galach, too. So I can die with a smile on my face, without feelin’ like the God Emperor gypped me.

Now this here story I’m about to unfold took place back in the early 10190s – just about the time of the conflict with Vlad Harkonnen and the I-treides. I only mention it because sometimes there’s a man… I won’t say a madhi, ’cause, what’s a madhi? But sometimes, there’s a man. And I’m talkin’ about the Dib here. Sometimes, there’s a man, well, he’s the man for the place the Bene Gesserit dare not look. He fits right in there. And that’s the Dib, on Dune. And even if he’s a prescient man – and the Dib was most certainly that. Quite possibly the most prescient in Arakeen, which would place him high in the runnin’ for the most prophetic galaxywide. But sometimes there’s a man, sometimes, there’s a man. Aw. I lost my heighliner of thought here. But… aw, Shaitan. I’ve done introduced him enough.

—

THE DIB: I’m the Muad’dib. So that’s what you call me. You know, that or, uh, His ‘Dibeness, or uh, ‘Diber, or El Muad’diberino if you’re not into the whole my-name-is-a-killing-word thing.

—

THE DIB: Yeah, well, that’s just, like, my terrible purpose, man.

—

CHANI: What do you do for recreation?

THE DIB: Oh, the usual. I jihad. Ride worms around. The occasional prescient spice trance.

—

DUNCAN IDAHO: Facedancers! Fuck me. I mean, say what you like about the tenets of the Bene Gesserit, ‘Dib, at least it’s an ethos.

—

FEYD-RAUTHA [to THE DIB]: What’s this day of rest shit? What’s this wormshit? I don’t fuckin’ care! It don’t matter to Feyd. But you’re not foolin’ me, man. You might fool the fucks in the naib, but you don’t fool Feyd. This bush league wheels-within-wheels stuff. Laughable, man – ha ha! I would have killed you Saturday. I kill you next Wednesday instead. Wooo! You got a date Wednesday, baby!

—

THE DIB: Mind if I do a water of L?

—

STILGAR: You have got to buck up, man. You cannot drag this negative energy in to the jihad!

THE DIB: Fuck the jihad… Fuck YOU, Stilgar!

STILGAR: Fuck the jihad? All right, I can see you don’t want to be cheered up here, ‘Dib. Come on Guerney, let’s go get us a worm.

—

IRULAN [voiceover]: I guess that’s the way the whole durned Golden Path keeps perpetuatin’ itself.

—

CHANI: Do you like sex, Usul?

THE DIB: ‘Scuse me?

CHANI: Sex. The physical act of love. Coitus. Do you like it?

THE DIB: I was talking about the waters of my homeworld.

—

STILGAR: Jamis was a good fighter, and a good Freman. He was one of us. He was a man who loved the desert… and killing Harkonnens, and as a wormrider he explored the sands of the Great Flat, from Tuono Basin to Habbanya Ridge and… up to… Gara Kulon. He died, like so many young men of his generation, he died before his time. In your wisdom, Shai-Hulud, you took him, as you took so many bright flowering young men on Salusa Secundus, on Giedi Prime, on Bele Tegeuse. These young men gave their lives. And so would Jamis. Jamis, who loved fighting. And so, Jamis of Sietch Tabr, in accordance with what we think your dying wishes might well have been, we commit the water of your body to the tribe, which you loved so well. Good night, sweet prince.

—

THE DIB: Just say the Litany Against Fear, man.

STILGAR: I’m perfectly calm, ‘Dib.

THE DIB: Yeah, waving the fucking crysknife around?

STILGAR: Calmer than you are.

—

[The Baron Vladimir Harkonnen and his mentat aid Piter De Vries interrogate Duke Leto Atreides.]

VLADIMIR: Is this your uniform, Leto? Is this your uniform, Leto?

DE VRIES: Look, Baron…

VLADIMIR: Piter, please? Is this your uniform, Leto?

DE VRIES: Just ask him about the signet ring.

VLADIMIR: Is this yours, Leto? Is this your uniform, Leto?

DE VRIES: Is that your thopter out front?

VLADIMIR: Is this your uniform, Leto?

DE VRIES: We know it’s his fucking uniform! Where’s the fucking signet ring, you freaking’ duke?

VLADIMIR: Look, Leto. Have you ever heard of Giedi Prime?

DE VRIES: Oh, for CHOAM’s sake, Vlad…

VLADIMIR: You’re entering a world of pain, Duke. We know that this is your uniform. We know that you had a signet ring.

DE VRIES: And your fucking heir.

VLADIMIR: And your fucking heir. And, we know that this is your uniform.

DE VRIES: We’re going to feed you your uniform, Leto.

VLADIMIR: You’re killing your lady, Leto!

—

IRULAN: I like your style, ‘Dib.

THE DIB: Well, I dig your style too, man. Got the whole galactic princess thing goin’.

IRULAN: Thankee.

—

IRULAN [voiceover]: Sometimes you ride the worm and sometimes, well, he rides you.

It’s been a busy, Metafilter-centric three months — after finishing up that trip (which was fantastic), we did a raffle (complete with wierd Paypal headache) that raised over $12K for charity, we ran an informal UI experiment with Metafilter favorites (complete with much arguing), Matt found out he had a great big pituitary tumor and scared the shit out of us, and said tumor has ended up responding very well to non-surgical treatment. Plus a dozen other slightly less landmark-worthy things I’m forgetting to mention.

And two mefites, newish member Bageena and long-timer and blogosphere-coiner Brad “bradlands” Graham, both passed away suddenly this last week. Sobering start to the new year.

I think I’m going to dodge the formal Resolutions thing again this year and just state a general intent to get my shit in gear.

Whole bunch of content up on Meeting Metafilter, more to come as the trip continues.

I’ve set up a site to track it all: Meeting Metafilter. I’ll be posting text, pictures, and even some audio and video as I can manage it.

Thanks to Jetblue’s stuntish (and sold-out-early) All-You-Can-Jet promotion, I’ll be spending the month from September 8th through October 8th flying around the continental United Stated, hitting a new city just about every other day and partying my ass off with mefites.

It’s an aggressive schedule; I’ve never done anything quite like this before. I’m excited and a little terrified, and I think it’ll be a hell of a lot of fun.

My city-by-city itinerary looks like this:

NYC
Boston
Portland, ME
Randolph, VT
New Orleans
Philly
Baltimore
DC
Raleigh/Durham
Las Vegas
LA
San Francisco
Salt Lake City
Chicago
Denver

I plan on documenting the trip heavily as I go, and I’ll be documenting the run-up to the 8th as well as I figure out how, exactly, I’m going to pull this thing off.

I celebrate by recycling, so here’s two (among several) songs I’ve written about the place over the years:

Happy [8th] Birthday, Metafilter, from a couple years back, and
Metafilter’s Down, from our long dark downtime of the soul earlier this year.

In an age where websites are always hunting for the next great trick for keeping unruly commentors in line – voting mechanisms? User scores? Enforced real names? – MetaFilter seems to be held together by the sheer stubbornness of its culture.

From (mefi’s own) Ivor Tossell at the Globe & Mail, here’s one of the best short explanations of Metafilter I’ve yet read.

And I’m not just saying that because it quotes me.

2. For example, this was a double post, and those are routinely removed — once something has had a shot on Metafilter, that’s pretty much it.

3. When we remove posts, we leave a deletion reason explaining to whatever degree with think is necessary what’s going on with the deletion, so that the poster and any curious readers have a little context for why the post went away.

4. (Posts that are deleted remain viewable at their original url, as in (2) above, but don’t show up to search engines or get listed in the site archives, etc.)

5. If the reason for deletion is pretty much self-explanatory and there’s nothing really at stake emotionally with the deletion (i.e. it’s not a contentious issue, no malicious intent on the poster’s part, no apparently newbie confusion, etc), I might provide a sillier or jokier reason than normal.

6. For example, again, the deletion reason in (2), where I went sort of meta-memetic, as such:

This post was deleted for the following reason: now we just need someone to combine this with that old tapestry-art meme and call it “thru thou” and the last track will be called “Play Him Hence, Lute Cat” and we’ll basically be all set. — cortex

7. Sometimes, people will read these deletion reasons and laugh to themselves and then that’s all that happens.

8. But sometimes, they will draw an awesome picture of Lute Cat.