Ideas for an action coordination website

in­spired by Inad­e­quate equil­ibria, and fol­low­ing the ‘Kick­S­tarter for Co­or­di­nated Ac­tion’ se­quence.

This is an idea-dump post for a web­site i thought of af­ter read­ing Inad­e­quate equil­ibria.

To­day, tools like Face­book and twit­ter help us co­or­di­nate bet­ter and faster. but still, that is not enough to solve prob­lems of “high-in­ad­e­quacy”—where we’re stuck in bad Nash equil­ibrium, and mov­ing away from it de­mands many things to hap­pen to­gether.

The idea here is to take co­or­di­na­tion much fur­ther, so we can solve as much of the game the­ory prob­lems bound in mov­ing to other Nash equil­ibria.

The goal is to al­low ‘Ex­o­duses’ from bad Nash equil­ibria, to bet­ter a Nash equil­ibria.

Note—none of this (at least as of writ­ing this) is be­ing worked on, nor are there cur­rently plans to do so. Though, of course this doesn’t mean there shouldn’t be, or else i wouldn’t have wrote the post :)

ba­sic concept

big prob­lems can’t be solved by a sin­gle per­son. but some­times even a large group of peo­ple who agree on the prob­lem and the solu­tion, and are even mo­ti­vated to bring the change—can’t do it. That is be­cause some prob­lems are more com­pli­cated than that.

one such com­pli­ca­tion is co­or­di­na­tion prob­lems—where “ev­ery­one is tak­ing some ac­tion A, and we’d rather all be tak­ing ac­tion B, but it’s bad if we don’t all move to B at the same time.” or each in­di­vi­d­ual wouldn’t want to take ac­tion A un­less he knows the rest do too.

solu­tions to co­or­di­na­tion prob­lems in­clude com­mon knowl­edge (Where all know the in­ten­tions of all ac­tors) and pre-com­mit­ment. that’s the site’s goal.

Co­or­di­nated actions

Any user can be­come a co­or­di­na­tor/​ini­tia­tor by cre­at­ing a co­or­di­nated ac­tion (CA), in or­der to solve some prob­lem. Each CA gets its own page (similar to kick­S­tarter). Any user (un­less oth­er­wise speci­fied) can obli­gate to the ac­tion—and here’s the catch—the com­mit­ment is to be re­al­ized if and only if a cer­tain num­ber of other users obli­gated the same. In Kick­S­tarter, we com­mit to pay, here we com­mit to take ac­tion.

Ver­sa­tile co­or­di­na­tion:

Since re­al­ity is very dy­namic, a rigid struc­ture will work for few situ­a­tions and be less use­ful. So to al­low a large ar­ray of pro­jects, op­tions for CAs should be ver­sa­tile. I see it as an on­go­ing pro­ject that will de­velop next to the com­mu­nity’s needs. think of how tesla is im­ple­ment­ing fea­tures in di­a­log with their com­mu­nity’s wishes. here are some ex­am­ples of “con­tracts”:

Ba­sic: all of us obli­gate to some CA if X oth­ers do too.

mile­stones: many ac­tions, listed on the same ini­ti­a­tive, which are taken at differ­ent amounts of obli­ga­tions. It is ei­ther re­quired to obli­gate for all ac­tions, or pos­si­ble to obli­gate to spe­cific ac­tions only.

Obli­ga­tion for obli­ga­tion: a group or in­di­vi­d­ual obli­gate for some­thing, if a differ­ent group or in­di­vi­d­ual obli­gate for the same or a differ­ent ac­tion (only one side can be an in­di­vi­d­ual).

Please com­ment with more types that you can think of, it’s very in­ter­est­ing.

fur­ther support

And some­times just com­mit­ting isn’t enough for you, or you sup­port the ini­ti­a­tive but done that ac­tion (in cases of one time de­ci­sions, like go­ing ve­gan, go­ing zero-waste, get­ting rid of your car, etc...). So we want to give users a way to sup­port fur­ther than just obli­gat­ing. One ex­am­ple is an op­tion for users would to donate money to the cam­paign, which will be used to fur­ther spread it, through some kind of ad­ver­tis­ing.


This is where it gets quite com­plex, but it has to. peo­ple don’t want a thou­sand ran­dom hu­mans around the world to some­thing with them. they want peo­ple from their coun­try, from their city, from their pro­fes­sion/​hobby/​In­ter­est-area, from their so­cial cir­cle or or­ga­ni­za­tion.

That’s where com­mu­ni­ties come in, the ideal is that for ev­ery real-world com­mu­nity you would be able to cre­ate a com­mu­nity on the site to re­sem­ble it, and that if peo­ple from that com­mu­nity are already on the site, they will find out that a new com­mu­nity they be­long to has been opened, and will join. the rea­son is, so it’s pos­si­ble to co­or­di­nate ac­tion with and within cer­tain com­mu­ni­ties.

When i first thought about how com­mu­ni­ties will work in prac­tice it re­minded me of set the­ory, but think­ing about it more, it ended up merely re­sem­bling it (and prob­a­bly breaks some of its laws). still, hope­fully men­tion­ing it helps to vi­su­al­ize.

the com­mu­ni­ties structure

There are many com­mu­ni­ties, it’s easy and ac­cessible to cre­ate new ones, or join ex­ist­ing ones (Un­less said com­mu­nity has some re­quire­ments). any com­mu­nity is a sub and/​or par­ent com­mu­nity of other com­mu­ni­ties. The goal is that the com­mu­ni­ties on the site will be able to re­flect the com­mu­ni­ties in the real world.

some ex­am­ples of com­mu­ni­ties: The earth (The par­ent of all com­mu­ni­ties un­til we colonies mars) and the user base*, all re­gions/​con­ti­nents, and all coun­tries, are some that can be added per-launch. ex­am­ples for user added com­mu­ni­ties: cities, EA, farm­ers is Is­rael, bus drivers in new York, ve­g­ans is the US, LessWrong, etc...

“that’s a lot com­mu­ni­ties..”, you say? “like, a ton of com­mu­ni­ties”, yeah, that’s true. but that’s how it should be.

it might sound a bit like FB, but ex­cept the com­mu­nity group­ing as­pect, there are two more im­por­tant differ­ences. here we don’t want two groups which are ba­si­cally the same (Cause that’s in­effec­tive, if the real world com­mu­nity is di­vided be­tween them on the site), and the user doesn’t have to be aware of all the com­mu­ni­ties they’re part of**, which may feel weird in a on­line plat­form, but that’s how it is in real life.

When­ever a user joins a com­mu­nity he is sug­gested sub-com­mu­ni­ties he might fit in, and is au­to­mat­i­cally added to all par­ent com­mu­ni­ties, which he can man­u­ally ex­clude them­selves from (I bet the set the­o­ret­i­ci­ans winched). part of the ac­count cre­ation pro­cess would be spot­ting the user’s com­mu­ni­ties (easy ones, for ex­am­ple, are coun­tries and cities)


not only in­di­vi­d­ual hu­mans are “agents” in this world, but also some com­mu­ni­ties, like, cor­po­ra­tions, non-prof­its, and any other goal ori­ented group. or­ga­ni­za­tions will be able to cre­ate a com­mu­nity around them­selves, but also act as a user on the site. this is im­por­tant, cause some co­or­di­nated ac­tions seek not only the co­op­er­a­tion of sin­gle hu­mans, but of the groups they make. if ve­g­ans think of tak­ing a co­or­di­nated ac­tion to­gether, they want to know that the busi­nesses and or­ga­ni­za­tions they sup­port will go with them too. some CAs will be rele­vant only for or­ga­ni­za­tions, where it doesn’t mat­ter how many in­di­vi­d­u­als com­mit to take some ac­tion, they have to have co­op­er­a­tion from their own group.

Mo­ti­va­tion and verification

At least un­til all of our so­ciety is nice, ed­u­cated and ra­tio­nal—We need two mechanisms, one that ver­ifies who re­ally co­op­er­ated and who defected, and a mechanism to dis­cour­age defect­ing, and en­courage co­op­er­a­tion.

fulfill­ment verification

Ver­ifi­ca­tion is the much harder of the two. how do you ver­ify that some­one tried ve­g­anism for 28 days? how do you ver­ify that some­one has/​hasn’t posted cer­tain types of posts to FB? Can you ver­ify whether some­one re­ally voted third party? whether they went to work that day? Some stuff are eas­ier to ver­ify, but if we stuck only to ac­tions that are easy to ver­ify, this tool won’t be very use­ful. It’s a hard-shell to crack, but it needs crack­ing—ideas?

user verification

In this sys­tem it’s im­por­tant that we know user = per­son, even bet­ter if we know user = which per­son. It can also help with fulfill­ment ver­ifi­ca­tion. There are many ways to do it—Email, phone, face, PayPal(?), ID. it just needs to be. maybe not all users have to be ver­ified, yet still pos­si­ble for com­mu­ni­ties and CAs to re­quire ver­ifi­ca­tion.


given that we solved ver­ifi­ca­tion—mo­ti­va­tion is sim­pler. a few op­tions:

co­op­er­a­tor/​defec­tor score: users have a pub­li­cly dis­played score that shows how they acted on their obli­ga­tions.

Achieve­ment badges: I en­vi­sion some­thing similar to khan academy’s badges, but harder to get so they’re more mean­ingful. an ex­am­ple of one “good to have: on av­er­age, you referred to each co­or­di­nated ac­tion you obli­gated to, at least 10 fulfilled obli­ga­tions.” You can dis­play these badges on your pro­file to sig­nal how awe­some you are :)

putting your cash where your mouth is: for each CA, ei­ther there’s a set amount or the ini­tia­tor sets it, each user de­posits cash against his co­op­er­a­tion—if he defects, he looses that money, if he co­op­er­ates, if he wins some ex­tra money. This is a pretty much bound-to-work mo­ti­va­tor (un­less bill gates starts us­ing this too), But i’d rather in­cor­po­rate money as a last re­sort, If we find that we re­ally need this ex­tra mo­ti­va­tor, since it makes ev­ery­thing more com­pli­cated. It also makes the ver­ifi­ca­tion task harder, since if peo­ple can use the site as a money-pump, they’ll be more likely to look for a back­door to ex­ploit.

I be­lieve this con­cept, if it was suc­cess­fully re­al­ized, could bring great benefits to the world.

*yeah, both are all users, but i see rea­sons to differ­en­ti­ate, sort­ing wise—com­mu­ni­ties might have a counter fea­ture of how many peo­ple in the real world com­mu­nity are reg­istered to this on­line com­mu­nity, in the users it’s a 100%, in the world.… you’d tar­get differ­ent CAs to the user base and “the whole world”. and, Sub com­mu­nity sort­ing is based on the real world, not the web­site, so the farm­ers of Is­rael are only part of the earth com­mu­nity and not the web­site’s User Base. pos­si­ble sub-com­mu­ni­ties are ac­tive users, con­trib­u­tors, etc...

**Hard to es­ti­mate ex­actly how much, but it’s at least a few dozens and maybe more than a hundred