A New Kind of Imageboard

I’ve been using im­age­boards for a long time, and I ac­cept that they have prob­lems. Total an­onymity brings out the worst in people: any con­ten­tious issue at all be­comes a hot button is­sue, where ci­vility of dis­cus­sion rap­idly de­grades. It doesn’t help that many of the more vocal im­age­board users are trolls, in­cels, or other sim­ilar groups. Politics and re­li­gion, even ex­er­cise and re­la­tion­ships, can cause huge ar­gu­ments. The simple solu­tion is to say “an­onymity has failed, let’s just use ac­counts with per­sistent handles and a repu­ta­tion system like everyone else”.

But I don’t like that solu­tion. Ac­counts raise the bar­rier to entry, which helps to curb drive-by trolling, but not sig­ni­fic­antly. People even act ter­ribly when their ac­count is as­so­ci­ated with their real name and em­ployer, as we have seen time and time again on Face­book and Twit­ter.

An­other problem with im­age­boards is stag­na­tion. They’re pretty much all the same. They have dif­ferent ap­pear­ences and dif­ferent boards, but that’s it. Some do de­part from the standard fea­ture-­set: live­boards, where you can see a post as it is typed, and have a more rap­id-fire dis­cus­sion, are a great in­nov­a­tion. Part-way between real-­time and asyn­chronous com­mu­nic­a­tion, I think they are the fu­ture of im­age­boards.

Live posting is a big change, both tech­nic­ally and cul­tur­ally, but live­boards are still clearly the same sort of thing as im­age­boards. I think we can go fur­ther.

This memo is an ex­plor­a­tion of ideas I have for some­thing which is re­cog­nis­ably “im­age­board-­like”, but which comes with some new stuff which I hope will have a cul­tur­ally sig­ni­ficant ef­fect. At the same time, I’ll dis­cuss “A Group Is Its Own Worst En­emy”, as my ideas are part-­mo­tiv­ated by the prob­lems raised in that.

A group is its own worst enemy

I re­com­mend reading the ori­ginal art­icle by Clay Shirky, but I’ll present a sum­mary here, which will be useful to bear in mind in later sec­tions.

There are three things we have to ac­cept about on­line com­munit­ies:

  1. You cannot com­pletely sep­arate tech­nical and so­cial is­sues. See “LambdaMOO Takes a New Dir­ec­tion” for one failed at­tempt at this.

  2. Mem­bers are dif­ferent than users. Some group of users (the “core group”) will arise that cares more than av­erage about the in­teg­rity and suc­cess of the group as a whole.

  3. The core group has rights that trump in­di­vidual rights in some situ­ations.

And there are four things we have to design for:

  1. A no­tion of iden­tity, be­cause knowing who said what is the min­imum re­quire­ment to have a con­ver­sa­tion. The simplest is a per­sistent handle.

  2. A no­tion of mem­bers in good stand­ing, so good deeds can be re­cog­nised. The simplest is, again, a per­sistent handle.

  3. Bar­riers to par­ti­cip­a­tion, to pro­tect the group from drive-by har­ass­ment and dam­age.

  4. A way to spare the group from scale, as it’s harder to have a good con­ver­sa­tion in a large group.

A plan for the fu­ture of im­age­boards

I’ll drill into this more deeply in the sub­sec­tions that fol­low, but here’s the high-­level over­view:

  • The basic model is a live­board.
  • With lim­ited opening hours.
  • Mod­er­a­tion is by vot­ing, by eli­gible on­line posters.
  • A mail sys­tem: users have any number of named mail­boxes, you can mail the au­thor of a post or a par­tic­ular mail­box, and re­spond to such mes­sages.
  • Mail is just live­posting but with a re­stricted set of users.
  • No en­forced vis­ible iden­tity: you can’t know that two posts are by the same person (or would send mail to the same per­son).

I’m taking care to pre­serve an­onymity to some de­gree, as I think it’s an im­portant part of im­age­board cul­ture. An im­age­board which ties every post to a named user is too much of a change: at that point you just have a weird forum.

Let’s go through these in or­der.

Lim­ited opening hours

This cur­rently ex­ists in some live­boards, and is typ­ic­ally called a “curfew” (though that’s not quite the right word). The idea is that the board is only “open” for some period of time, not only can you not post out­side of the opening hours, you can’t even browse.

It works ex­tremely well for fos­tering dis­cus­sion when the com­munity is small, be­cause everyone makes an ef­fort to show up at the same time. Even if there are only ten people, all live­posting at once feels a lot busier than people being spread around the clock.

This is often com­bined with auto­matic de­le­tion of posts when the board closes. Posts only exist within their one ses­sion.

Poster iden­tity

Shirky goes quite far, and says that weak pseud­onymity doesn’t work well, as people need to con­nect the cur­rent con­ver­sa­tion to prior con­ver­sa­tions. I don’t fully agree with that: I think con­necting to prior con­ver­sa­tions is im­portant for a more mean­ingful dis­cus­sion, but I don’t think al­lowing users to change their handles easily is much of a prob­lem.

One model I have seen work well, al­beit in a small com­munity where all of these prob­lems are sim­pler, is that most users vol­un­tarily as­sume a con­sistent handle. But the handle is not part of the soft­ware, it’s just a thing ex­pressed through choice of im­ages.

People oc­ca­sion­ally change handle, but not fre­quently. When someone does, one of three things hap­pen:

  • They ex­pli­citly con­nect their new handle to their pre­vious handle, by making a post.

  • They do not con­nect their handles, but people figure it out any­way, be­cause of how they write and what they post about.

  • They do not con­nect their handles, and their posting style is not dis­tinctive enough for people to con­fid­ently con­nect the two handles.

So two out of three out­comes result in the com­munity knowing that the poster has just changed their handle. Would man­aging ac­counts and handles in soft­ware change this? I don’t think so: in the third case, where the user wants to use a new handle un­con­nected to their old one, they would just create a new ac­count.

Fur­ther­more, there is a so­cial cost to chan­ging your handle: un­less the whole com­munity is on at the same time, people will be asking for weeks af­ter­wards “Is X gone? Or did they just change their name?” This tends to dis­courage fre­quent changes, but still per­mits them for users who like to change and don’t mind the hassle.

So, in sum­mary, I agree with Shirky that con­sistent handles are needed for in­ter­ac­tions which last longer than a single con­ver­sa­tion, but I don’t think they need to be part of the soft­ware.


Im­age­boards at­tract trouble, and a per­sistent problem is that there are never enough mod­er­at­ors. I have written be­fore about im­age­board staffing, and the system is quite heavy­weight. Here’s a sum­mary:

They can do everything. Some ex­amples of ad­min-only powers are: cre­ating and de­leting users, cre­ating and de­leting boards, modi­fying user per­mis­sions.
The bulk of your typ­ical im­age­board’s staff. On their boards, they can do ba­sic­ally any­thing: de­lete, move, and edit posts and im­ages; ban users; spoiler and un­spoiler im­ages; bumplock threads; sticky threads; and so on.
They can de­lete posts.

Making someone a mod­er­ator is a sig­ni­ficant in­vest­ment of trust. Mod­er­ators are ex­pli­citly cre­ated by the ad­min­is­trator, and have little over­sight. If a mod­er­ator does some­thing bad on their board, you’d better hope you have good backups.

Jan­itors were in­tro­duced as a low-trust stop­gap, but they can’t do much. If someone is spam­ming, and only jan­itors are on­line, they cannot really deal with the spam­mer, only clean up after them.

What we want is a large group of people who can handle an on­going at­tack in the ab­sence of any staff. Who better to form this group than the core users? The core users are the most in­vested users: the ones who show up to every ses­sion, who re­port bad posts, and who try to steer dis­cus­sion away from bad act­ors.

But we don’t want someone who in­filt­rates this trusted core group to be able to cause prob­lems if they turn bad. So we’ll re­strict the power of any one in­di­vidual, by making de­cisions by a simple ma­jority vote of on­line mem­bers. Spe­cific­ally, I think this group should be able to:

  • Elect new mem­bers to the group.
  • Issue a tem­porary ban to the au­thor of a post.
  • Hide an in­di­vidual post.
  • Hide all posts by a banned user.

On the tech­nical side, this could work using pass­words. When a user is elected to the group, they get is­sued a unique pass­word. They can gain their powers by en­tering the pass­word on the web­site. They can also reset their pass­word.

When a vote is triggered, all users who have signed in during the cur­rent ses­sion are eli­gible voters. The vote passes if a ma­jority vote yes.


When a user shows up for the first time, they get a ran­domly gen­er­ated mailbox name. This could be stored in a cookie, with the op­tion to ex­pli­citly enter it so that users don’t lose their mail if they clear their cookies (or use an­other com­puter).

When a user makes a post, the post keeps a re­cord of the user’s mail­box. Other users can send a mes­sage to the poster’s mail­box. However, the mailbox name is not shown (all you know is that you’re sending a mes­sage to the au­thor of that post), so mailbox names can’t be used to link posts to­gether.

A user can also choose to tell other users their mailbox name, in which case those users can dir­ectly mes­sage them. Such mes­sages can be sent to mul­tiple mail­boxes at the same time. In this case, we’ll prob­ably want the ability to pass­word-­pro­tect a mail­box, so we can allow private com­mu­nic­a­tion without also granting read ac­cess.

We’ll prob­ably also want the ability for a user to mon­itor mul­tiple mail­boxes; but one must be the “primary” mail­box, which gets as­so­ci­ated with their posts.

I’m en­vi­sioning mail to be just like reg­ular posts, where all re­cip­i­ents can see as you type. So mail threads are ef­fect­ively one-­thread private live­boards. It should also be pos­sible to add a new re­cip­ient to a thread. By keeping these threads in­vite-only, I think it avoids the mod­er­a­tion night­mare that is 8chan, where any user can create their own (pub­lic) board.

Per­haps mail should be ac­cess­ible at all times, so the curfew is not a bar­rier to these more private dis­cus­sions.

A few more thoughts

If we take the private group mail to its lo­gical ex­treme, then the point of the public posting space is to be, more or less, a lobby in which the whole group gets to­gether, and to act as a place for new mem­bers to ar­rive; dis­cus­sion in this lobby will be fairly su­per­fi­cial and friendly, with more ser­ious dis­cus­sion hap­pening in cul­tiv­ated groups. This gives rise to an in­ter­esting ques­tion:

Are mul­tiple boards, or even mul­tiple threads, ne­ces­sary at all?

I sus­pect not. In which case, we need to be careful with the growing number of posts in the single “thread” (and in each mail group). I think an ex­piry mech­anic would work nicely here: posts can either ex­pire at a fixed time (such as the end of a ses­sion), or when the thread con­tains a given number of posts.

I think this would be a very in­ter­esting com­munity!

  • A single public thread, only open for a few hours a day at fixed times;
  • With a mail sys­tem, ac­cess­ible even when the lobby is closed;
  • Where posts gradu­ally ex­pire as time goes on.

This feels more like a chat­room than an im­age­board!

Open prob­lems

So how does this pro­posal do with re­spect to Shirky’s design re­quire­ments?

  1. A no­tion of iden­tity.

    As dis­cussed, I don’t think this needs a solu­tion in the soft­ware, and any­thing which the soft­ware does im­ple­ment can be easily over­come. The mail system is the closest thing there is to iden­tity, and I am ex­pli­citly al­lowing the same in­di­vidual to have as many mail­boxes as they want.

  2. A no­tion of mem­bers in good stand­ing.

    I’m going to ex­pli­citly re­cog­nise this group, by pro­moting them to mod­er­at­ors. This group will then grow it­self, without the need for staff in­ter­ven­tion. Is there a need for people to be able to cast out mem­bers? Per­haps, I don’t know. Hope­fully this will be a rare oc­cur­rence, but def­in­itely staff should be able to do it.

  3. Bar­riers to par­ti­cip­a­tion.

    This is sort of solved. The core group chooses new mem­bers by vot­ing, which is cer­tainly a bar­rier: you need half of the ones who are on­line at any given time to agree you’re also of good stand­ing. The curfew system im­poses an­other bar­rier, which some­what dis­cour­ages drive-by at­tacks, but does nothing to deter those willing to wait.

  4. A way to spare the group from scale.

    This still needs thought. There are two types of scale prob­lems: bad actors over­whelming the staff, which we have some­what solved with the idea of the core group; but also good actors di­luting dis­cus­sion. How do we solve this? The mail system will help, es­pe­cially with group dis­cus­sions, but I don’t think it’s a com­plete solu­tion.

We also need to think about users not on a tra­di­tional com­puter: the posters who use tab­lets and phones. I have heard (but couldn’t find an ori­ginal source) that moot said that if he was starting 4chan today, he would focus on the mo­bile ex­per­i­ence. That’s where the users are today. Using im­age­boards on a phone is def­in­itely a pain cur­rently, so there are some in­ter­face wins to be made. Does a mo­bile in­ter­face also sug­gest new func­tion­al­ity? I’m not sure. Being able to up­load photos or videos straight from the cam­era, and audio straight from the mi­cro­phone, would cer­tainly be neat.

culture, tech
Target Audience
Imageboard dwellers & cyber-anthropologists.
Epistemic Status
Totally unsubstantiated opinions, but I've been immersed in this culture for a while.
Attention Conservation Notice
This is part feature list / part response to Clay Shirky's "A Group Is Its Own Worst Enemy" essay. Will I ever implement this and test these ideas out? Who knows!