Call for participation - Centre for Organisational Rationality

post by whpearson · 2013-04-04T18:59:58.126Z · LW · GW · Legacy · 26 comments

Contents

26 comments

Or something like that1. As per this article on Control Markets I am looking to experiment with them. This requires an organization of some sort. This post is my first step to the creation of the organization.

My skills are in technology and not UI or people particularly. So I have mainly been doing what I know2, working around my day job.

So I decided it probably wasn't the best path, I was likely to get distracted and lose focus by myself. So other people would be useful.

There are three rough ways this can be attempted:

  1. Try and create a swanky high profile website to spark interest: Get a co-founder type person/team.   Stay stealthy and make a nice web app for managing funge for a charity based on exploring control marketers using personal funds to start with, get on social media. Raise money from donations to maintain infrastructure, refine the web app and possibly fund academic research on control markets when people can try it out for themselves. This webapp can then be open sourced for other people to use.
  2. Get the theory started: Start a small organization with volunteer Actors and bootstrap using a google spread sheet to record funge transfers and bids, initially, and then use lessons learnt from that process to inform the design of a webapp. Leading to
  3. Get someone who regularly wants to talk about control markets, just to keep focus.

So what is possible with people from lesswrong? And which would people advise? Or is there a different way to attack this problem.

I've thought about kickstarter, but I don't think it is within the ToS. Also I would want a decent demo/visuals before trying this style of thing.

I suppose there is also.

1 I would want to use a kitty for a logo if we went with this name. This is probably why I shouldn't be allowed to make these sorts of decisions.

2 I've been noodling around with a django backend using tastypie to make a restful API, I was thinking about using AngularJS on the client (which I would need to learn). Still needs a lot of work. I also have a bad habit of wanting to make it highly available/scalable and other such things that aren't appropriate at this point.

26 comments

Comments sorted by top scores.

comment by ModusPonies · 2013-04-04T20:35:51.512Z · LW(p) · GW(p)

Based on this post, I'm not sure what you're after, other than "something to do with this control market thing and a web app." Are you trying to build an app that will let users run their own control markets? Are you trying to use a control market to run an organization that will build an arbitrary web app? (If the latter, what are you building?)

What's your end goal? Are you trying to test your theory? Make money? Build something cool?

This is a "call for participation"—what sort of participation do you want, exactly?

Replies from: whpearson
comment by whpearson · 2013-04-04T20:45:37.925Z · LW(p) · GW(p)

The end goal is to publicise the idea and get people to experiment with it, to establish whether it is a viable way of organising people.

what sort of participation do you want, exactly?

The sort of participation people are comfortable with.

If you wanted to explore the idea of control markets, how would you go about doing it?

Replies from: ModusPonies, maia
comment by ModusPonies · 2013-04-04T22:08:48.408Z · LW(p) · GW(p)

If you wanted to explore the idea of control markets, how would you go about doing it?

I don't understand the difference between control markets and regular markets well enough to answer this question. As near as I can tell (which admittedly isn't very far; I have a hard time understanding your writing), you've independently derived the idea of paying people based on performance metrics, with a dash of prediction markets thrown in for spice. There's a fair amount of literature on both of those already.

Replies from: whpearson
comment by whpearson · 2013-04-04T23:20:16.097Z · LW(p) · GW(p)

Sorry if I am unclear. I tend to communicate with computers more than humans, where you just need a specification and the implications are derived. I should learn that people don't tend to work like that....

Perhaps this will help:

In prediction markets people predict facts about the world they have little influence over.

In control markets each Actor predict how well Stakeholders will evaluate them, if that Actor is given control of certain part of an organisation. This prediction is represented by a bid of funge (an internal currency). Other Actors don't make predictions about that, but upon their own performance. There is no concept of odds that normally go along with prediction markets.

So for example, let us say we have a mailing list resource. Gavin and Brett both want control of the resource. It normally gets a funge feeback of 314. Gavin thinks he has a great idea to reduce the frequency of mailings and can get a feedback score of 360 so bids 200. Brett has no such great ideas so bids his usual 150. So Gavin will bid higher than Brett, but he will only pay what Brett thinks it is worth (due to the auction system chosen), so should make a profit. If Gavin's idea does less well than he thought he will lose funge overall.

There are lots of potentially interesting things I am glossing over here, things like internal resources, e.g. IT systems. Let us say this is being bid over by Bob and Jim. Bob wins. Bob then gets paid funge by people like Gavin, rather than getting an external performance review. If Bob makes a tweak that gets another 50 funge from the mailing list resource for Gavin, then Gavin is incentivized to give say an extra 25 fung to Bob so that Bob can outbid the more incompetent Jim for control of the IT systems and Gavin can ask him to make a different tweak in future.

Most of this isn't original anyway but derived from work on learning classifiers and the like, so works in computers.

I am a bit loathe to go into too much detail in how it will actually work in practice, a lot depends upon the scale it is being used at. Gavin and Brett may be different teams rather than individuals, for example.

(Edited for clarity)

Replies from: Zaine, daenerys
comment by Zaine · 2013-04-05T00:52:44.803Z · LW(p) · GW(p)

Advice on writing for clarity:

Omit pronouns in favour of specific referents - exempli gratia this

In control markets people predict how well people will evaluate them, if they are given control of part of an organisation

becomes

In control markets people [x] predict how well others [people y] will evaluate the former's [people x's] (control market? what you meant by "them" was not clear), if people [x or y (I've no idea which party you meant)] are given control of part of an organisation.

I thoroughly searched Google for 'funge', and found no results that seem to fit. Please define it, and be specific.
If you want to get people interested, you need to tell them what you mean to do; if you are uncomfortable disclosing your idea in public fora to the degree necessary for garnering interest (an ambiguous qualifier, to be sure), then perhaps you might wish to reconsider your approach.

Replies from: JenniferRM, whpearson
comment by JenniferRM · 2013-04-08T03:38:05.656Z · LW(p) · GW(p)

For what its worth, the verb "funge" is what kids these days are funging for phrases like "trades off against". Some supposed benefits of this verb are that it takes one syllable rather than four, and it explicitly primes for the concept of "fungibility" in a way that highlights how often a tradeoff isn't being made against two single things, but rather a tradeoff is being considered between some coherent proposal versus a vast array of other possible plans with similar benefits, and partial fractions and/or combinations of such plans, and so on in a vast optimization space that may contain low hanging fruit... even as considering such tradeoffs often runs afoul of sacredness heuristics.

comment by whpearson · 2013-04-05T06:22:09.616Z · LW(p) · GW(p)

I defined funge in my control market article which I linked to. I was a bit hasty making that comment. I'll edit it to match the nomenclature of that article.

If you want to get people interested, you need to tell them what you mean to do; if you are uncomfortable disclosing your idea in public fora to the degree necessary for garnering interest (an ambiguous qualifier, to be sure), then perhaps you might wish to reconsider your approach.

I wish for Control Markets (and experimenting with alternative organizational structures in general) to be part of the zeitgeist. This needs prominent real world examples of its use, which in turn needs experiments with the market and a platform for easy usage of a market. Being a market it needs participants in order to experiment with it. I am currently trying to figure out how to get participants.

I think I need to create a youtube video to explain how it might work. Linking to an article seemed to be TL;DR.

Replies from: Zaine
comment by Zaine · 2013-04-05T10:22:46.104Z · LW(p) · GW(p)

Create a basic multi-player browser game (simpler than the MS Hearts interface), then ask for people to play it. Make the game robust enough to test your ideas, yet simple enough that participants can quickly learn the rules and play the game properly - a condition that still allows for great complexity, depending upon the participants.

Run many trials, contrast the results of the games with your predictions, and revise or discard your hypothesis as necessary. Slowly introduce confounds into your game to test its real-world viability. Ultimately run a long-term game (a couple of weeks or so) in a real world setting as a mock trial. Scale up from there.

In other words, experiment. Prove the feasibility of the idea then build around it. Perhaps the above isn't even necessary; try your best to falsify the idea with the least possible amount of resources, I believe 'twas recommended. The above suggests one manner in which that may be accomplished.

comment by daenerys · 2013-04-07T03:24:02.035Z · LW(p) · GW(p)

I had suspected you were a sock of someone who had gone to one of CFAR's rationality workshops, and had decided to surreptitiously play the "Be Specific"/"Pitch an Idea to..Paul Graham?" game on LW (perhaps for fun, or to see if anyone would catch on, or as a way to run a "real-world" exercise?)

Then I checked your username, and saw you had 1000+ karma, which probably negates the theory that you are a sock, but doesn't necessarily negate the theory that you are playing the Be Specific game with us....

Replies from: whpearson
comment by whpearson · 2013-04-07T11:45:12.139Z · LW(p) · GW(p)

I've not been to CFAR, it is a bit too far. Mostly I've no idea what types of organizations people would actually want to join, not being overly enamoured with organizations myself. But I need bodies to test my theories about organizations....

Any links to descriptions of the game?

comment by maia · 2013-04-04T21:02:22.382Z · LW(p) · GW(p)

What is "the idea"? Just control markets?

Replies from: whpearson
comment by whpearson · 2013-04-04T21:16:37.416Z · LW(p) · GW(p)

Yes. I have other ideas, but I'm not going to push them for this.

comment by Viliam_Bur · 2013-04-05T07:50:23.527Z · LW(p) · GW(p)

Do you have "something to protect"? Some specific real-world goal that you really want to achieve, preferably with help of other people? If yes, you could try using the Control Market app for this purpose.

Making a web application is relatively easy, you don't need a Kickstarter funding, just a free weekend or two. Implement the critical parts first, don't worry about design and other unessential stuff. Then start using the application. This is how you get feedback.

If the application will help, you have helped the cause you care about and did an experiment with the application. And you also have your first reference.

(As an extreme case -- I am not sure if this is even possible -- could you use the application just for yourself, to help you decide about your personal goals?)

comment by Decius · 2013-04-04T23:01:26.200Z · LW(p) · GW(p)

You haven't defined the organization you are trying to form well enough for me to determine if I am interested or not.

Replies from: whpearson
comment by whpearson · 2013-04-04T23:28:22.861Z · LW(p) · GW(p)

Can you define an organization that you would be interested in being part of? The trouble is that I am not that fussy as long as the Control Markets get tested, so will likely go for what will get me more participants and engagement. People I have talked to previously have suggested that the organization have a goal different from just exploring control markets, but I am at a loss to pick a goal.

Would people in lesswrong be interested in an X-risk focused organization?

Replies from: evand, Decius
comment by evand · 2013-04-05T02:08:35.364Z · LW(p) · GW(p)

It sounds like you want to allocate some resources you don't have in furtherance of a goal you haven't defined and don't care about, and would like other people to be passionate enough about this to help you make a go of it.

Replies from: whpearson
comment by whpearson · 2013-04-05T07:19:59.312Z · LW(p) · GW(p)

I don't care about the goal but I do care about the method that the goal is selected and how the people are encouraged to work towards the goal. A good metaphor would someone wanting to try an organisation controlled by voting, he couldn't say what the organization will end up doing, because it would be dependent upon what the people who voted would make it do. If he had a firm goal that he just had to achieve, then he would probably would be better off without the organization. Voting is a method of combining diverse goals and knowledge into one organization; as are control markets.

If I can't get people interested in the idea now, building a web app by myself won't make any difference. I also understand my motivational structures well enough to know that working with other people helps a lot.

Replies from: ModusPonies
comment by ModusPonies · 2013-04-05T15:12:08.411Z · LW(p) · GW(p)

A good metaphor would someone wanting to try an organisation controlled by voting, he couldn't say what the organization will end up doing, because it would be dependent upon what the people who voted would make it do. If he had a firm goal that he just had to achieve, then he would probably would be better off without the organization.

Well there's your problem. People didn't create democracy just to see if it would work. People created democracy to answer the question "how the heck are we going to run our nation?" There were firm goals. "Guard civilization against bandits, natives, and the French." (Or "bandits, barbarians, and the Spartans," depending on the era.) "Protect our liberty from the British." "Make sure the government stays beholden to the people."

If you want to build something, you need an actual, concrete thing to build. Locke didn't create modern democracy. That took Jefferson and Hamilton and the rest.

If I can't get people interested in the idea now, building a web app by myself won't make any difference.

I agree. You don't actually want a web app that badly, and frankly, neither do any of us. I see two solutions.

1) Apply this to a real problem whose mere existence causes you real emotional pain. (e.g., "open source software projects are shoddily run," "the healthcare system is fucked up and bullshit," "the technology to do this awesome thing I want doesn't exist.") Use the enthusiasm this generates to get others involved. 2) Follow Zaine's advice. Build a toy example of your system that's fun for its own sake. Use that to gather data.

Either way, I would suggest you learn to write better before you go ahead with any project that requires outside participation. Basic things like sentence structure and punctuation are still holding you back. Posts like your stuff here in Discussion are the best way to do that. Keep writing, keep getting feedback, and focus on specific techniques.

Replies from: whpearson
comment by whpearson · 2013-04-05T18:38:28.575Z · LW(p) · GW(p)

Heh. Part of the reason I posted here was that Lesswrong is associated wtih CFAR. So I thought people would be more amenable to meta-improvement organisations. If people are interested in raising the sanity water-line of individuals, why not the sanity waterline of organizations? I see I was miscalibrated.

Zaines advice would most likely end up being a webapp of some flavour anyway :P The game would need to be multiplayer, no download required. And I may as well make it so the control market software in the game has an api and can be easily extracted from the game if people decide they want to use it in anger.

I will try and write more. But limited time will make it probably slip off my agenda in favour of coding.

Thanks for the feedback.

Replies from: gjm
comment by gjm · 2013-04-05T18:54:53.665Z · LW(p) · GW(p)

You might want to distinguish between the following two propositions.

  1. "People here are not interested in raising the sanity waterline of organizations."

  2. "People here don't see how your proposal for experimenting with 'control markets' is actually likely to do much to raise the sanity waterline of organizations."

It seems to me that you've had evidence for #2 and have concluded #1.

Replies from: whpearson, ModusPonies
comment by whpearson · 2013-04-05T19:14:34.703Z · LW(p) · GW(p)

I've been explicitly recommended to get a concrete outcome by villiam_bur and told to I need to focus on one by Modus Ponies. They got up voted. No one has said, "Hey I like your enthusiasm for trying a different organisational structure, would you be interested in helping me try out this type of system first?". This would have been evidence of 2 for me. And I would have evaluated the system and may have decided to help it.

We probably won't hit on the right one straight away, but we won't get anywhere without fostering a culture of experimentation.

Edit: I think most people are interested in improving organisations in the abstract, lots of people complain about governments etc. I'm looking for people that are actively looking for new organisational methods to try.

Replies from: evand
comment by evand · 2013-04-05T21:39:11.040Z · LW(p) · GW(p)

If I had an idea I was passionate about, and was looking for people to help make it work, my first instincts would be to find other people who are passionate about it as well, or people who have a demonstrated history of making ideas like mine work. Which one are you?

comment by ModusPonies · 2013-04-05T19:06:10.620Z · LW(p) · GW(p)

Technically this thread is evidence for both of those. If whpearson concluded #1, then that was probably because of bad priors (most likely, failing to consider #2 due to ugh fields and/or illusion of transparency), not because of misinterpreting evidence.

Replies from: gjm
comment by gjm · 2013-04-05T22:36:08.974Z · LW(p) · GW(p)

Almost everything is evidence either for or against almost everything else. I agree that this thread is some evidence for both #1 and #2, but I suggest it's better evidence for #2 than for #1 -- and, as you suggest, #2 has higher prior probability than #1.

comment by Decius · 2013-04-05T04:18:32.066Z · LW(p) · GW(p)

Now it sounds like you don't want to create an organization, you want to use an existing organization. What problem is your app intending to solve, and what groups currently have that problem?

comment by novalis · 2013-04-05T08:17:20.263Z · LW(p) · GW(p)

Control markets are a totally untested way of running an organization. Wouldn't a better name be "center for experimental organization"? That doesn't presuppose that the proposed way of doing things will actually work.