The Economics of the Asteroid Deflection Problem (Dominant Assurance Contracts)

Imag­ine a world with no ads or pay­walls. A world where open-source soft­ware gets the same level of fund­ing as pro­pri­etary soft­ware. A world where peo­ple can freely reuse ideas and mu­sic with­out pay­ing roy­alties. A world where peo­ple get paid for writ­ing book re­views. A world where Game-of-Thrones-qual­ity shows are freely available on YouTube. A world where AI safety re­search gets the same-level of fund­ing as AI ca­pa­bil­ities re­search. Is this a fan­tasy world? No, this is the world where peo­ple use Dom­i­nant As­surance Contracts

If you are already con­vinced you can make this idea a re­al­ity by donat­ing to cre­ate a Plat­form for Dom­i­nant As­surance Con­tracts. If you are not con­vinced read on.

The Free-rider problem

A few months ago I stum­bled across this video. (I highly recom­mend you watch the video, but if you don’t have time, I’ve sum­ma­rized the video be­low).

Sum­mary of A Deeper Look at Public Goods

A good is ri­val if one per­son’s use of a good diminishes an­other per­son’s abil­ity to benefit from it. Jeans are ri­val. If I’m wear­ing a pair of jeans, you can’t wear it at the same time. As­teroid deflec­tion is non-ri­val. If I deflect an as­ter­oid to pro­tect my­self, you are saved with no ad­di­tional cost.

A good is ex­clud­able if peo­ple who don’t pay can be eas­ily pre­vented from us­ing a good. An ex­am­ple of a good that is ex­clud­able is a pair of jeans. You can ex­clude peo­ple by lock­ing the jeans in your closet. An ex­am­ple of a good that is non-ex­clud­able is as­ter­oid deflec­tion. You can­not pre­vent the peo­ple who did not pay for the as­ter­oid deflec­tion pro­gram from benefit­ing from the as­ter­oid be­ing deflected.

A good which is both ri­val and ex­clud­able is called a pri­vate good. A good which is non-ri­val and non-ex­clud­able is called a pub­lic good.

(Ad­di­tion­ally, goods which are ex­clud­able and non-ri­val are called club goods, and goods which are non-ex­clud­able but ri­val are called com­mon re­sources. We won’t be fo­cus­ing on these types of goods, but I’ve men­tioned them for com­plete­ness)

Mar­kets are good at pro­vid­ing pri­vate goods because

• by ex­clud­ing peo­ple who don’t pay, con­sumers are in­cen­tivized to pay, which in­cen­tivizes pro­duc­ers to pro­duce, and
• since pri­vate goods are ri­val it is effi­cient to ex­clude con­sumers who aren’t will­ing to pay (if the benefit to the con­sumer was greater than the cost, the con­sumer would be will­ing to pay).

Public goods challenge mar­kets because

• con­sumers who don’t pay can’t be ex­cluded, con­sumers are in­cen­tivized in­stead to free-ride (i.e. benefit from the good with­out pay­ing) and thus pro­duc­ers have no in­cen­tive to pro­duce.
• Ad­di­tion­ally, even if we could figure out a way to ex­clude non-pay­ers, (e.g. by ex­e­cut­ing ev­ery­one who doesn’t pay for the as­ter­oid deflec­tion pro­gram), it is in­effi­cient to do so (be­ing non-ri­val means there are no ad­di­tional costs to non-pay­ers benefit­ing).

Ex­am­ples of pub­lic goods

• Information
  • Journalism
  • Pre­dic­tion markets
  • Scien­tific research
  • Ed­u­ca­tional material
  • TV series
  • Movies
  • YouTube videos
  • Books
  • Short-stories
  • Art
  • Podcasts
  • Neigh­bor­hood watches
  • Vac­cines and other pub­lic health interventions
  • AI safety
  • Mili­tary defense
  • Public roads
  • Public parks

  Isn’t there a clever mechanism to solve the free-rider prob­lem?

  This video stuck with me. The fact the pub­lic goods are in­effi­ciently pro­vided by the mar­ket seems like the main is­sue with our civ­i­liza­tion. Heck, AI Safety is a pub­lic good.

  The other thing that stuck is that this seems so solv­able. Surely, there is a clever mechanism that can fix this is­sue? So I went to the Wikipe­dia page of the Free-rider Prob­lem and scrol­led to the bot­tom, and lo and be­hold it was just sit­ting there: Dom­i­nant As­surance Con­tracts in­vented in 1998 by Alex Tabar­rok (yes, the same per­son who pre­sented the video you just watched). There is only one prob­lem: no one is us­ing them.

  What is a dom­i­nant as­surance con­tract?

  (I recom­mend you watch the first 30 min­utes of this video. It’s re­ally good, but if you don’t have time I wrote a short ex­pla­na­tion be­low that you can read in­stead).

  My short ex­pla­na­tion of dom­i­nant as­surance contracts

  Sup­pose there are 10 villagers on an un­paved street. Bob the Builder will pave the street for $90. Each villager is will­ing to con­tribute up to $15 to have the street paved. Ad­di­tion­ally, there is there is a $1 trans­ac­tion cost to con­tribut­ing (e.g. from time spent sign­ing forms, from credit card fees, from op­por­tu­nity cost of not earn­ing in­ter­est on that money, etc.).

  If the Bob the Builder gets ev­ery­one to con­tribute, each villager will only have to pay $90 /​ 10 = $9 and profit $15 - $9 - $1 = $5. How­ever, one of the villagers, Free-rid­ing Frank, is hes­i­tant to con­tribute. Con­sider Frank’s de­ci­sion table:

  (Frank’s Profit, Others’ Profit)	Others Don’t Con­tribute	Others Con­tribute
  Frank Doesn’t Con­tribute	(0, 0)	(15, 4)
  Frank Con­tributes	(-1, 0)	(5, 5)

  • If no­body con­tributes then road doesn’t get paved and Free-rid­ing Frank and the other villagers both profit $0.
  • If Frank con­tributesbut the oth­ers don’t then Frank pledges $15 dol­lars (the max­i­mum he is will­ing to pay) and in­curs a trans­ac­tion cost of $1. Since no one else con­tributed, Bob was un­able to col­lect $90, so re­funds Frank the $15. Frank, how­ever, still in­curred the $1 trans­ac­tion cost.
  • If the oth­ers con­tribute but Frank doesn’t then Free-rid­ing Frank prof­its $15, but the oth­ers pay $90 /​ 9 + $1 = $ 11 and so only profit $4.
  • If Frank and the oth­ers all con­tribute then they all pay $9 dol­lars, in­cur a trans­ac­tion cost of $1 and profit $5.

  Look­ing at the table and hold­ing the oth­ers strat­egy fixed, Free-rid­ing Frank re­al­izes that the dom­i­nant strat­egy in this game is to not con­tribute (0 vs −1 and 15 vs 5). Of course, ev­ery­one else in the village re­al­izes this too and Bob is un­able to raise the funds to pave the road.

  Sup­pose Bob changes the game to a Dom­i­nant As­surance Con­tract. Dom­i­nant As­surance Con­tracts have two con­di­tions:

  • As­surance: Bob will only build the road if all 10 villagers con­tribute.
  • Re­fund Bonus: Bob puts up $20 of his own money as col­lat­eral. If some villagers do not con­tribute then Bob will re­fund all villagers that did con­tribute what they pledged plus an equal share of the col­lat­eral: $2 per villager.

  Frank’s de­ci­sion table now looks some­thing like this:

  (Frank’s Profit, Others’ Profit)	Others Don’t Con­tribute	Others Con­tribute
  Frank Doesn’t Con­tribute	(0, 0)	(0, 1)
  Frank Con­tributes	(1, 0)	(5, 5)

  • If no­body con­tributes then road still doesn’t get paved and Free-rid­ing Frank and the other villagers both profit $0.
  • If Frank con­tributesbut the oth­ers don’t Frank pays the trans­ac­tion cost of $1, but since the road was not built, the Bob re­funds Frank an ad­di­tional $2 [Re­fund Bonus]. Frank prof­its $1
  • If the oth­ers con­tribute but Frank doesn’tthe road does not get built [As­surance]. The other villagers in­cur $1 trans­ac­tion costs and get an ad­di­tional $2 re­fund [Re­fund Bonus]. They profit $1
  • If Frank and the oth­ers all con­tribute then they all pay $9 dol­lars and a trans­ac­tion cost of $1, the road gets built and ev­ery­one prof­its $5.

  Look­ing at the table and hold­ing the oth­ers strat­egy fixed, Free-rid­ing Frank now sees that the dom­i­nant strat­egy in this game is to con­tribute (0 vs 1 and 0 vs 5). Of course, ev­ery­one else in the village re­al­izes this too and the road gets paved!

  Challenges of dom­i­nant as­surance contracts

  If dom­i­nant as­surance con­tracts are so great, why is no­body us­ing them?

  The prob­lem is that the pro­duc­ers of pub­lic goods don’t know about them.

  Ex­am­ple: Ele­ment is build­ing Ma­trix, an open-stan­dards chat app (i.e. a pub­lic good). Their busi­ness model is to give away the code/​stan­dards for free and sell con­sult­ing ser­vices. Un­for­tu­nately, their com­peti­tors have un­der­cut them by sel­l­ing con­sult­ing ser­vices with­out con­tribut­ing back to the code/​stan­dards. On Hack­erNews you can see their CEO com­plain about it as “The Tragedy of the Com­mons”. Ex­cept he is not ex­pe­rienc­ing the tragedy of the com­mons. He is ex­pe­rienc­ing the free-rider prob­lem.

  If he doesn’t even know what prob­lem he has, he is not go­ing to be able to: google “the free rider prob­lem”; open the Wikipe­dia page on “the free-rider prob­lem”; scroll to the bot­tom to find the sec­tion ti­tled “Solu­tions”; and then read about dom­i­nant as­surance con­tracts.

  Peo­ple who are try­ing to pro­duce pub­lic goods don’t seem to know that that’s what they are do­ing, and that dom­i­nant as­surance con­tracts are a way to get fund­ing.

  My main goal is to give the idea enough air­time so that when peo­ple want to pro­duce pub­lic goods, the first tool they reach for is a dom­i­nant as­surance con­tract.

  Do they work in prac­tice?

  I only know of four at­tempts done in the wild (tell me if you know about more).

  • Dona­tion to The Cen­ter for Elec­tion Science.
    • Suc­ceeded
    • Raised $300 in dona­tions from 20 peo­ple.
    • Some in­for­ma­tion was pub­li­cly pro­vided about the num­ber of peo­ple who pledged be­fore the dead­line.
    • Fi­nal 3 pledges came in the last half hour.
    • Failed
    • 9 ⁄₁₀ or $225/​250
    • No in­for­ma­tion pub­li­cly pro­vided about the num­ber of peo­ple who pledged be­fore the dead­line.
    • This one failed, but got re­ally close 9 ⁄₁₀ or $225/​$250 for writ­ing a book re­view! I want to em­pha­size how ridicu­lous it is that it got this close. 9 peo­ple were will­ing to pay $25 for a book re­view of a book that only costs $18! I think if they used a plat­form like Kick­starter, a kind soul would have seen that they were only $25 dol­lars away from suc­ceed­ing and would have donated.
    • Failed(?)
    • Tried to raise $2500 of play money.
    • Pledges were pub­li­cly visible
    • I’m not sure how to count this one. They failed to raise $2500 worth of play money but then they im­ple­mented dark mode any­way. So the pub­lic good was pro­vided, even though the con­tract failed.
    • Failed
    • Asked for $700
    • Pledges were not pub­li­cized be­fore the dead­line.
    • Asked for $700 dol­lars to fund a house din­ner. Ar­jun hasn’t done a write-up of the re­sult, but he failed. Also the in the first few days af­ter pub­li­ca­tion the pay­ment links weren’t work­ing.

    1 ⁄₄ doesn’t seem so good. Dom­i­nant as­surance con­tracts work in the­ory so what is go­ing on?

    Main prob­lems ig­nored by theory

    The the­o­ret­i­cal model of Dom­i­nant As­surance Con­tracts as­sumes away some things that you have to deal with in the real word

    • Perfect in­for­ma­tion about pric­ing: In the ex­am­ple prob­lem above, we as­sumed that 10 villagers were will­ing to pay $15 to pave the road. In real life you do not have that in­for­ma­tion and risk over-pric­ing/​un­der-pric­ing your con­tract. Pre­sum­ably the con­tracts run in the real world that failed were over-priced.
    • Perfect mar­ket­ing: In the ex­am­ple prob­lem above, we as­sumed that villagers who would have benefited knew that Bob the Builder was rais­ing funds. In real life you have to mar­ket to po­ten­tial cus­tomers and many po­ten­tial cus­tomers will prob­a­bly not know about your product. It’s pos­si­ble that the con­tracts run in the real world failed to reach all po­ten­tial cus­tomers who would have benefited.
    • Game the­ory/​psy­chol­ogy: In the ex­am­ple above Free-rid­ing Frank was fully aware of the other villagers strat­egy. In re­al­ity we don’t know how other peo­ple are go­ing to be­have.
    • Shady: Dom­i­nant as­surance con­tracts seems gam­bling-like/​scam-like the same way that crypto-cur­rency does. The top com­ment on the Berkeley House Din­ner Pro­ject is “this is a dol­lar-auc­tion (AKA scam)”, and the sec­ond top com­ment is “Surely this is ille­gal?”

    Solu­tion to pric­ing problem

    I be­lieve the pric­ing prob­lem can be solved by pre­dic­tion mar­kets. Book Re­view, Dark Mode and Berkeley House Din­ner all had pre­dic­tion mar­kets on whether they’d reach their tar­get. The mar­kets were at 90%, 10%, 45%. Th­ese prob­a­bil­ities seem con­sis­tent with how much money each of the pro­jects raised.

    Be­fore we launch a pro­ject we could run a mul­ti­ple-choice pre­dic­tion mar­ket on man­i­fold: “The Pro­ject will raise $X”, “The Pro­ject will not raise $X”, “The Pro­ject will not launch or will launch with a price differ­ent from $X”. Only if “The Pro­ject will raise $X” is suffi­ciently high, will we launch the pro­ject.

    Solu­tion to mar­ket­ing problem

    This prob­lem is not unique to dom­i­nant as­surance con­tracts. So it can be solved in the same way ev­ery­one else does it.

    A cool idea is to part­ner with some­one with ex­per­tise in mar­ket­ing. The ad­ver­tiser can put up the col­lat­eral for the re­fund bonus. If the pro­ject suc­ceeds the pro­ducer can give them a cut of the funds raised. In this way, the ad­ver­tiser is in­cen­tivized to make the pro­ject to suc­ceed.

    Solu­tion to game the­ory/​psychology

    I think hav­ing a progress bar show­ing how many peo­ple have pledged so far is in­cred­ibly im­por­tant to reach the tar­get.

    • If the pro­ject is not on track to reach­ing it’s goal, spec­u­la­tors will pledge in or­der to claim the re­fund bonus.
    • If the pro­ject ison track to reach­ing it’s goal, proso­cial peo­ple are likely to pledge so that the pro­ject gets funded.
    • If the pro­ject is close to reach­ing it’s goal, but time is run­ning out. Po­ten­tial free-rid­ers will wait un­til the very last mo­ment, but even­tu­ally con­cede and pledge their money.

    In some sense, with a progress bar, the re­fund bonus re­wards peo­ple for pro­vid­ing in­for­ma­tion on whether the pro­ject is likely to reach it’s fund­ing goal.

    I sus­pect the Book Re­view pro­ject would have worked if there was a progress bar.

    Solu­tion to look­ing like a scam.

    Per­son­ally I think fram­ing it as “You get a Re­fund Bonus as grat­i­tude for sup­port­ing our pro­ject which we are very sorry we are un­able to de­liver” rather than “If you pledge you could win $5. Buy now. ” can make it seem less like a scam.

    Hav­ing a progress bar also makes it seem less like a scam since it makes it eas­ier for peo­ple to pre­dict if the pro­ject is go­ing to suc­ceed or fail.

    Summary

    If you price your pro­ject cor­rectly (with the help of pre­dic­tion mar­kets), mar­ket it suc­cess­fully, phrase the re­fund bonus in a way that it does not seem like a scam and have a progress bar with a list of peo­ple who’ve already pledged, then it’s very likely that it will reach it’s tar­get be­cause:

    • it’s ra­tio­nal for peo­ple to pledge, fur­ther­more,
    • the re­fund bonus re­wards peo­ple for pro­vid­ing in­for­ma­tion on whether the pro­ject will reach it’s fund­ing goal.

    Other po­ten­tial problems

    • Fraud: It’s pos­si­ble that the pro­ject is just a scam, and that the pro­ducer dis­ap­pears with the money.
    • In­for­ma­tion asym­me­try: It’s pos­si­ble that the pro­ducer is un­able to de­liver the product be­cause of in­com­pe­tence or pro­duces a product of lower qual­ity than cus­tomers ex­pected.
    • Col­lat­eral: The need for col­lat­eral that pro­duc­ers need to put up makes risk-averse pro­duc­ers less likely to try, can­cel­ling out the in­crease in the suc­cess rate.
      • Po­ten­tial solu­tion: If we can get suc­cess rate close to 100% then the plat­form can put up the col­lat­eral.
      • Po­ten­tial solu­tion: En­trepreneurs will be will­ing to in­vest if they think they can make a profit (i.e. some­one who is risk tol­er­ant can un­der­write the col­lat­eral).
      • Pos­si­bly, I think our job is to cre­ate a cul­ture where we re­lease things for free and with­out ads, similar to the cur­rent cul­ture in open source soft­ware.
      • If dom­i­nant as­surance con­tract be­come pop­u­lar there is less of a case for patents and copy­right so it will be eas­ier to con­vince poli­cy­mak­ers that they should no longer ex­ist.

      Okay so what am I buy­ing?

      You’re col­lec­tively pay­ing for $629 dol­lars for 1 month of my time to make this idea a re­al­ity. (I don’t live in the Bay Area so my cost of liv­ing is low). I don’t re­ally need the money. I’m do­ing this as a ex­per­i­ment to test that dom­i­nant as­surance con­tracts work. I’ll likely ask for ad­di­tional fund­ing in the fu­ture to im­ple­ment more fea­tures af­ter the first month, and for spe­cific ex­penses as they come up.

      The plan is to cre­ate a web­site where pub­lic-good pro­duc­ers can cre­ate a page that

      • Has a de­scrip­tion of the pro­ject.
      • Has a progress bar show­ing how much and who have pledged.
      • Han­dles pay­ments with PayPal
      • If the pro­ject doesn’t reach the fund­ing goal, the cus­tomers are au­to­mat­i­cally re­funded with a re­fund bonus. If the pro­ject does reach it’s goal, the pro­ducer gets the money in their PayPal ac­count.
      • The pro­ducer will put up the re­fund bonus as col­lat­eral us­ing PayPal.

      The web­site is es­sen­tially already done (my pro­to­type just has my pro­ject hard-coded). I just need to flesh it out so that other peo­ple can up­load their pro­jects.

      The main thing I’m go­ing to do­ing is look­ing for peo­ple who want to cre­ate pub­lic goods and get­ting feed­back from them on the plat­form (if that’s you please join the Re­fund Bonus Dis­cord or con­tact me in some other way)

      Ad­di­tion­ally I want to try some cool things which might not work out like

      • Use pre­dic­tion mar­kets to price the pro­ject.
      • Bring in in­vestors/​ad­ver­tisers who will put up the col­lat­eral on be­half of the pro­duc­ers and take a cut if the pro­ject suc­ceeds.

      But why can’t I just let other peo­ple pay and free-ride?

      You haven’t been pay­ing at­ten­tion. Un­less I’ve priced this con­tract wrong, if you don’t pay it doesn’t hap­pen. If you don’t pay, I don’t meet the goal, ev­ery­one gets re­funded, I go back to my bor­ing day job, and we con­tinue to live in a world where pub­lic goods are un­der-pro­vi­sioned.

      But won’t some­one else do this?

      It seems un­likely. The Dom­i­nant As­surance Con­tract pa­per came out in 1998. And 25 years later, I seem to be the first per­son to com­mit to get­ting it off the ground. You will prob­a­bly have to wait a while be­fore some­one else comes.

      The fi­nal call to action

      If you want to live in a world with no ads or pay­walls; a world where open-source soft­ware gets the same level of fund­ing as pro­pri­etary soft­ware; a world where peo­ple can freely reuse ideas and mu­sic with­out pay­ing roy­alties; a world where peo­ple get payed for writ­ing book re­views; a world where Game-of-Thrones-qual­ity shows are freely available on Youtube; a world where AI safety re­search gets the same-level of fund­ing as AI ca­pa­bil­ities re­search, then please pledge some money to­wards this.

      Also please join the Re­fund Bonus Dis­cord or con­tact me in some other way if your in­ter­ested in:

      • Find­ing peo­ple who want to make pub­lic goods, putting up col­lat­eral on their be­half, mar­ket­ing their pro­ject, and tak­ing a cut if they reach their fund­ing goal.
      • Re­ceiv­ing fund­ing for pro­duc­ing pub­lic goods. Even if you don’t want fund­ing, but already pro­duce pub­lic goods I’m still in­ter­ested to talk.
      • Buy­ing pub­lic goods on the plat­form.

      Isn’t this just Kick­starter?

      “Don’t Con­tribute” is still an equil­ibrium on Kickstarter

      If we go back to our ear­lier ex­am­ple, with­out a re­fund bonus, due to trans­ac­tion costs, both “Don’t Con­tribute” and “Con­tribute” are both equil­ibriums (0 vs −1 and 0 vs 5). The re­fund bonus re­moves “Don’t Con­tribute” from the equil­ibrium.

      (Frank’s Profit, Others’ Profit)	Others Don’t Con­tribute	Others Con­tribute
      Frank Doesn’t Con­tribute	(0, 0)	(0, −1)
      Frank Con­tributes	(-1, 0)	(5, 5)

      Sure, but this is just a small im­prove­ment on Kick­starter. It doesn’t seem rev­olu­tion­ary.

      I don’t think enough peo­ple are work­ing on this prob­lem. Alex Tabar­rok pub­lish his pa­per on dom­i­nant as­surance con­tracts in 1998. Kick­starter was founded in 2009, 11 years later. To this day their are no plat­forms for dom­i­nant as­surance con­tracts to fund pub­lic goods.

      Most things funded on Kick­starter aren’t even pub­lic goods! Their stated mis­sion is to “help bring cre­ative pro­jects to life”. They are even try­ing to solve the free-rider prob­lem! They just hap­pen to be us­ing a similar mechanism.

      I agree dom­i­nant as­surance con­tracts are a small im­prove­ment, but small im­prove­ments add up and al­most no­body is work­ing on this!

      But why aren’t peo­ple us­ing Kick­starter to fund pub­lic goods right now?

      They are. Here are some suc­cess­ful open-source pro­jects funded by Kick­starter:

      • https://​​www.kick­starter.com/​​pro­jects/​​1681258897/​​its-magit-the-mag­i­cal-git-client
      • https://​​www.kick­starter.com/​​pro­jects/​​krita/​​krita-2016-lets-make-text-and-vec­tor-art-awesome

      Here is an un­suc­cess­ful project

      Ak­ira got 38% of their tar­get fund­ing. I’d be will­ing to bet that if they use dom­i­nant as­surance con­tracts they would have suc­ceeded.

      Ap­par­ently 79% of pro­jects that raised more than 20% of their goal were suc­cess­fully funded. The re­fund bonus pro­vides the nec­es­sary kick (hehe) that gets twice as many pro­jects over the line.

      Again, I think the main prob­lem is that peo­ple don’t know that as­surance con­tracts (dom­i­nant or oth­er­wise) can be used to fund pub­lic goods.

      But you have no moat. Kick­starter will just copy you.

      Yes! I’m not try­ing to make a bazillion dol­lars. I just want more pub­lic goods! If Kick­starter copies this idea that would be great! Less work for me!

      Wait, you said some­thing about AI Safety but this doesn’t ap­ply be­cause.

      … AI Ca­pa­bil­ities just scale with com­pute, where as we don’t know how to scale AI Safety. The prob­lem is not fund­ing, and even if it there was a dom­i­nant as­surance con­tract for “$1 000 000 000 to solve the al­ign­ment prob­lem” there is no guaran­tee that who­ever gets the money will solve the al­ign­ment prob­lem.

      Yes, there is a in­for­ma­tion prob­lem when fund­ing re­search things like AI Safety. We don’t have the in­for­ma­tion on whether the re­searcher’s re­search will be any good. But guess what: in­for­ma­tion is a pub­lic good! That means that it’s cur­rently un­der­funded. Dom­i­nant as­surance con­tracts can bridge that fund­ing gap.

      How do we fund in­for­ma­tion? Of the top of my head a bunch of things come to mind

      • We can fund prizes that will be given to some­one who an­swers a par­tic­u­lar re­search ques­tion.
      • We can fund some­one to do an anal­y­sis of which re­searchers or re­search di­rec­tions seem most promis­ing.
      • We can fund pre­dic­tion mar­kets on some­thing like “P(doom|give Bob $100,000)” vs “P(doom|don’t give Bob $100,000)”.

      Alter­na­tives and their problems

      Of course, there are other ways to fund pub­lic goods, but they all have their own prob­lems.

      Ads

      Prob­a­bly the main method to pri­vately pro­duce pub­lic goods is by run­ning ads on goods you give away for free. The main is­sue with this is that ads pay a frac­tion of a cent per view. That means that you can only fund low-qual­ity/​cheap goods. To see this, com­pare videos on YouTube with se­ries on HBO.

      Taxes

      With taxes you face a sym­met­ric prob­lem to the free-rider prob­lem: the forced-rider prob­lem. Think of some­one who doesn’t use a car, but still pays taxes to main­tain the roads.

      Fund­ing pub­lic good through taxes also suffer from lack of mar­ket-pric­ing mechanism. Govern­ment spend­ing doesn’t have to pass the mar­ket test that a dom­i­nant as­surance con­tract has to pass, cre­at­ing waste­ful spend­ing.

      Im­pact Certificates

      Im­pact cer­tifi­cates suffer from the free-rider prob­lem. There isn’t any in­cen­tive (other than al­tru­ism) to buy an im­pact cer­tifi­cate. One can just free-ride on the re­sults of the im­pact pro­ject, and let the ini­tial in­vestor go bankrupt.

      I think im­pact cer­tifi­cates are try­ing to solve the prob­lem of in­for­ma­tion asym­me­try rather than fund­ing. It’s pos­si­ble that they could be com­bined with dom­i­nant as­surance con­tracts in some way, but I’m not sure how.

      Club Goods

      A Club Good is a good that is non-ri­val but ex­clud­able. An ex­am­ple would be a sub­scrip­tion to HBO or Microsoft Office. It costs prac­ti­cally noth­ing on the mar­gin to give an ad­di­tional per­son ac­cess to HBO or Microsoft Office (all the costs were up­front in pro­duc­tion) so these goods are non-ri­val. How­ever, since HBO and Microsoft charge for these goods they are ex­clud­able.

      Of course, there is piracy. So the first prob­lem is that it’s ac­tu­ally difficult to turn a pub­lic good into a club good, and so the free-rider prob­lem per­sists.

      Se­condly, it’s in­effi­cient to ex­clude peo­ple. If you would pay $5 to watch Game of Thrones but the HBO sub­scrip­tion is $15, then $5 of sur­plus is be­ing lost (it costs HBO $0 to let you watch Game of Thrones).

      Patreon

      This is a type of club good.

      From per­sonal ex­pe­rience, I think this works okay for art where the pa­trons have a paraso­cial re­la­tion­ship with the artist. This seems to work less well for im­per­sonal stuff like soft­ware.

      Ad­di­tion­ally, it’s easy for Buc­ca­neer Bob to sign up to your Pa­treon and then up­load your prod­ucts to a piracy web­site. Free-rid­ing Frank won’t bother to sign up to your Pa­treon, he’ll just down­load your prod­ucts off a piracy web­site.

      If you price a dom­i­nant as­surance con­tract cor­rectly, Free-rider Frank is forced to pay.

      Micropayments

      The idea with micro­pay­ments is some­thing like “cost per marginal unit is a frac­tion of a cent so we should make it pos­si­ble for peo­ple to pay frac­tions of cents.” I think this doesn’t work be­cause the cost of fraud is higher than a frac­tion of a cent. Also, the cost of me think­ing “Is this worth 0.01c?” is higher than a frac­tion of a cent. It’s just bet­ter to treat such things as non-ri­val.

      As­surance Con­tracts/​Kickstarter

      The main is­sue with an as­surance con­tract with­out a re­fund bonus is that “not con­tribut­ing” is an equil­ibrium, es­pe­cially if their are sub­stan­tial trans­ac­tion costs. I already ex­plained this above.

      Summary

      Public goods are un­der-pro­vi­sioned by the mar­ket due to the free-rider prob­lem. Dom­i­nant as­surance con­tracts solve the free-rider prob­lem. No­body is us­ing dom­i­nant as­surance con­tracts. To solve this prob­lem, and si­mul­ta­neously test if dom­i­nant as­surance con­tracts work, I cre­ated a dom­i­nant as­surance con­tract to fund a plat­form for cre­at­ing dom­i­nant as­surance con­tracts.

      • Up­date #2 to “Dom­i­nant As­surance Con­tract Plat­form”: EnsureDone by moyamo ( 28 Nov 2023 18:02 UTC ; 33 points)
      • Up­date to “Dom­i­nant As­surance Con­tract Plat­form” by moyamo ( 21 Sep 2023 16:09 UTC ; 32 points)
      • localdeity 's comment on Don’t En­dorse the Idea of Mar­ket Failure by Maxwell Tabarrok ( 1 Mar 2024 18:13 UTC ; 2 points)
      moyamo 29 Aug 2023 18:28 UTC 71 comments 15 min read LW link
      1. They es­ti­mate with high prob­a­bil­ity the limit will only be reached iff they sign up (but this is the case even with a stan­dard Kick­starter, dom­i­nant as­surance isn’t nec­es­sary).
      1. They es­ti­mate with high prob­a­bil­ity that the limit won’t be reached (in which case they’re on av­er­age just milk­ing the con­tract for free cash)

      There’s still no in­cen­tive for the freerider to sign up if the limit would oth­er­wise be reached.

      In­stead what this solves is the co­or­di­na­tion prob­lem. Proso­cial peo­ple want to fund var­i­ous mar­kets, but it’s a waste of time do­ing this if no­body else joins in. The dom­i­nant as­surance con­tract makes this worth it ei­ther way, so it’s eas­ier for mar­kets to get off the ground to the point where it looks like they’ll prob­a­bly make it.

      • Ir­rele­vant: Ex­cept in small and well defined com­mu­ni­ties, most crowd­fund­ing cam­paigns tar­get a large pop­u­la­tion that’s hard to quan­tify and sin­gle out. And most cam­paigns don’t need ev­ery­one who would benefit to back them, only enough of them. If you tried to do a cam­paign where ev­ery­one who would benefit from the it has to back it for it to suc­ceed, it would definitely fail.
      • Un­de­sir­able: The fact that a good is pub­lic doesn’t mean that ev­ery­one benefits from it the same amount, be­cause not ev­ery­one val­ues it the same amount. In your ex­am­ple, the villagers prob­a­bly value the road differ­ently (per­haps based on how much they drive). It makes sense that peo­ple will pay differ­ently based on how much they value the good. If we fund a Bran­don San­der­son adap­ta­tion, it makes sense that the hard­core fans will pay a lot for it and I would pay lit­tle, be­cause I want it, but not as much as they do.

      Which is why crowd­fund­ing cam­paigns use an amount of money as the goal in­stead of the amount of back­ers. But if you use fund­ing-goal in­stead of a back­ers-goal, and peo­ple col­lec­tively value what you offer more than what you ask for it, then you still have the pos­si­bil­ity of free rid­ing, be­cause it would be pos­si­ble for the cam­paign to suc­ceed with­out ev­ery­body (I think this is fine).

      The goal isn’t to elimi­nate free-rid­ing, it’s to pre­vent pro­jects from failing be­cause peo­ple try to free ride, and these are very differ­ent things, be­cause the lat­ter means you’re will­ing to tol­er­ate free-rid­ers as long as the pro­ject gets funded, and former means you’re un­will­ing to tol­er­ate free-rid­ing even if the pro­ject could oth­er­wise be funded.

      So, though DACs do re­duce free-rid­ing, and do prob­a­bly al­low peo­ple to donate less each and still reach the goal, they don’t com­pletely solve free rid­ing be­cause you’d rarely want to use them in such a way where they do (which is why you im­ple­ment re­fund com­pen­sa­tion but don’t re­quire cer­tain amount of par­ti­ci­pa­tion or set the price so high that ev­ery­one would need to par­ti­ci­pate).

      As @Yair Halber­stadt said, the prob­lem this ac­tu­ally solve is the co­or­di­na­tion prob­lem faced by peo­ple who want to back the pro­ject if it suc­ceeds, but don’t want to pay the cost of back­ing a failed pro­ject. In that sense, re­fund com­pen­sa­tion is a very nat­u­ral ex­ten­sion of of the Kick­starter model, be­cause it sim­ply rec­og­nizes that peo­ple need to be re­funded for their effort and not just their money.

      About your el­e­ment/​ma­trix ex­am­ple, the prob­lem is in­deed free-rid­ing, but DACs would only solve the prob­lem if he wouldn’t keep de­vel­op­ing it un­less he got a cer­tain amount of fund­ing (which would make it a co­or­di­na­tion prob­lem). If he doesn’t want to con­di­tion the de­vel­op­ment on a cer­tain level of fund­ing, then fund­ing will be based only on the good­will of peo­ple. So DACs also don’t solve cases where a re­ward is de­served, but you’re gonna do the thing re­gard­less.

      And be­cause of the ti­tle, would a DAC be a good way to raise money for as­ter­oid deflec­tion? Definitely not as a way to try to make ev­ery­one par­ti­ci­pate in fund­ing the deflec­tion. It’s a good solu­tion only if there’s a min­i­mum amount of fund­ing or back­ers you can do it with, and with any less than that you may as well not bother. But that doesn’t seem like a good idea. If you get $5 or 1 per­son less than you asked for ate you just gonna do noth­ing? Or would you rather try any­way and die with dig­nity?

      Public/​pri­vate goods

      I think the fo­cus on pub­lic goods in the post is a bit con­fus­ing. Not be­cause it’s a prob­lem to have and offer that as the main mo­ti­va­tion for this, but be­cause it makes it seem like it’s only rele­vant for pub­lic goods, where it’s still as rele­vant for pri­vate goods if there’s a big up­front cost to pro­duc­ing any at all, so you have to have a min­i­mum num­ber of cus­tomers. For ex­am­ple, de­sign­ing a product is a large cost that you have to pay up­front, but it also doesn’t scale with the amount of prod­ucts sold. So it makes sense to say ‘I will sell this only if there’s enough buy­ers to cover the de­sign cost’. Most crowd­fund­ing cam­paigns for pri­vate goods are ex­actly of this type, and would benefit al­most as much from re­fund com­pen­sa­tion as pub­lic good cam­paigns (and it should sadded us that they fail due to a co­or­di­na­tion prob­lems as well, be­cause even though it’s a pri­vate good it’s still lost value).

      I agree with this com­ment ex­cept for these two points

      About your el­e­ment/​ma­trix ex­am­ple, the prob­lem is in­deed free-rid­ing, but DACs would only solve the prob­lem if he wouldn’t keep de­vel­op­ing it un­less he got a cer­tain amount of fund­ing (which would make it a co­or­di­na­tion prob­lem). If he doesn’t want to con­di­tion the de­vel­op­ment on a cer­tain level of fund­ing, then fund­ing will be based only on the good­will of peo­ple. So DACs also don’t solve cases where a re­ward is de­served, but you’re gonna do the thing re­gard­less.

      I don’t think he is go­ing to do it re­gard­less. Pre­sum­ably if he doesn’t get enough money he has to shut-down his com­pany com­pletely or re­duce the amount he spends on open-source and fo­cus on the more prof­itable con­sult­ing side of the busi­ness. If DACs were used they could in­vest more in the open source.

      I think the fo­cus on pub­lic goods in the post is a bit con­fus­ing. Not be­cause it’s a prob­lem to have and offer that as the main mo­ti­va­tion for this, but be­cause it makes it seem like it’s only rele­vant for pub­lic goods, where it’s still as rele­vant for pri­vate goods if there’s a big up­front cost to pro­duc­ing any at all.

      Pri­vate goods already have a mechanism to solve this prob­lem: in­vest­ment. That’s why I think DACs are not that rele­vant for pri­vate goods.

      In the case where he has a min­i­mum amount be­low which he is un­will­ing to con­tinue de­vel­op­ment, then I agree a DAC would help raise that amount. In­vest­ment solves the fund­ing prob­lem, it doesn’t solve the prob­lem of mak­ing sure you have enough buy­ers. And if you can’t solve the lat­ter then you might not be able to get in­vestors ei­ther (cause they also want to know you have enough buy­ers). That’s the prob­lem Kick­starter solves, and DACs would solve wven bet­ter, in ad­di­tion to fund­ing.

      Dom­i­nant as­surance con­tracts seems gam­bling-like/​scam-like the same way that crypto-cur­rency does. The top com­ment on the Berkeley House Din­ner Pro­ject is “this is a dol­lar-auc­tion (AKA scam)”, and the sec­ond top com­ment is “Surely this is ille­gal?”

      I don’t see any­where in your post where you ad­dress the ques­tion of whether this is ac­tu­ally le­gal, though? An offer that if some­one gives you money then (de­pend­ing on fac­tors out­side of that per­son’s con­trol) you may give them dou­ble their money back sounds like the kind of thing so­cieties tend to reg­u­late heav­ily. For ex­am­ple, in the US, I think it is pos­si­ble that this would pass the Howey test to be con­sid­ered a se­cu­rity?

      1. An in­vest­ment of money
      2. In a com­mon enterprise
      3. With the ex­pec­ta­tion of profit
      4. To be de­rived from the efforts of others

      I think it fails 3 and 4 si­mul­ta­neously. There is no “ex­pec­ta­tion of profit to be de­rived from the efforts of oth­ers”. If the con­tract suc­ceeds then oth­ers make an effort but you do not make a profit. If the con­tract fails you make a profit, but not from the efforts of oth­ers.

      Not a lawyer, but I’m not sure it’s so clear cut. If I buy be­cause I ex­pect the con­tract not to tip, I’m ex­pect­ing profit (3) de­rived from your efforts in set­ting up the con­tract (4).

      Maybe we shouldn’t call it a Re­fund Bonus but some­thing like “At­ten­tion Com­pen­sa­tion”. I would say that you are ex­pect­ing to profit (3) de­rived from your efforts of read­ing the con­tract and agree­ing to pledge and then not ac­tu­ally get­ting what you paid for (not 4). Per­son­ally I think it’s ir­ra­tional to give money ex­pect­ing a DAC to fail. make-yass made a post about this. I think the long-run so­cial equil­ibrium is that peo­ple re­al­ize that it’s dumb to gam­ble on DACs failing and don’t do it.

      In tra­di­tional char­i­ta­ble en­ter­prises, we just call it “a donor gift.”

      Agree, this is im­por­tant. When I wrote about this I didn’t con­sider that it might be ille­gal. I’m quite ig­no­rant of the law, but I re­cently wrote my thought on le­gal­ity of re­fund bonuses here. tl;dr, I think it’s le­gal, but there’s a bunch of things you have to do for it to be le­gal. And I now think this is the main rea­son no crowd­fund­ing plat­form has ever used re­fund bonuses, not that they don’t know about it or don’t think it’s a good idea. I would re­ally like if some­one law-savvy would write up an ex­pla­na­tion of the le­gal­ity of re­fund bonuses. I’d even be will­ing to con­tribute money to­wards that.

      I am not a lawyer, but I think that some­times the prob­lem can be avoided by re­turn­ing some­thing other than cash. Some­times, just hav­ing an ex­pira­tion date makes the differ­ence. You give me $1; if the pro­ject fails I give you a $2 coupon you can spend in Wal­mart you need to spend be­fore De­cem­ber 2023. (Of course, ask a lawyer about the de­tails. I am just point­ing at a pos­si­ble an­gle of ap­proach.)

      Does “Kick­starter but they re­fund you if the pro­ject doesn’t meet its goal” ex­ist? I think not — that might get you some of the gains of a DAC plat­form. Also, what does PayPal’s TOS say about this? I se­ri­ously con­sid­ered build­ing some­thing similar, and Stripe only ap­proves “crowd­fund­ing” if you have their writ­ten ap­proval.

      Does “Kick­starter but they re­fund you if the pro­ject doesn’t meet its goal” ex­ist?

      That’s just Kick­starter. You already only pay if the pro­ject hits its fund­ing goal.

      This is a in­ter­est­ing idea, but your ex­am­ple over­sells it sub­stan­tially. Fu­ture iter­a­tions need to not do this, or you’ll sound like a huck­ster and not get sup­port. If all ten peo­ple on the street have to con­tribute, ev­ery­one knows for sure that they won’t get what they want if they don’t con­tribute. No pay­back is nec­es­sary; free rid­ing is im­pos­si­ble. That’s the easy case. When you just need some amount of money rather than ev­ery sin­gle in­di­vi­d­ual to con­tribute, it’s quite pos­si­ble to free ride.

      But why can’t I just let other peo­ple pay and free-ride?

      You haven’t been pay­ing at­ten­tion. Un­less I’ve priced this con­tract wrong, if you don’t pay it doesn’t hap­pen.

      This is egre­giously wrong. You can’t pos­si­bly know that this con­tract will fail if each in­di­vi­d­ual reader of this sen­tence doesn’t con­tribute. This is the dis­tinc­tion from the ex­am­ple you’ve used. So I agree with the other com­ments that this does not fix the freerider prob­lem. At all. Which is the big prob­lem, not the fric­tional costs of pledg­ing. I think you’ve got to fix that ter­rible logic in fu­ture ver­sions, or you’ll sound dishon­est. I still pledged fifty bucks, be­cause im­prov­ing crowd­sourc­ing like Kick­starter even marginally is a wor­thy goal! And the fric­tional cost is still a prob­lem, so over­com­ing it will help.

      Par­tial Update

      I’m fully funded! In fact, as of writ­ing. I have nearly 3x more fund­ing than I asked for! The money re­ally started rol­ling in af­ter Alex Tabar­rok plugged my pro­ject on marginal rev­olu­tion, which I did not ex­pect. I’m not sure why peo­ple con­tinued to give me money af­ter the pro­ject was fully funded but I’m very thank­ful. (Of course, the more money I get the longer I can work on the pro­ject in­creas­ing the chance of suc­cess). The fundrais­ing only offi­cially ends 15 Septem­ber, so I’m not go­ing to give a full re­port un­til then. I was hop­ing to do some ne­glected chores while the cam­paign ran, but since I’m already fully-funded, I feel I can’t leave peo­ple hang­ing. So right now I’m col­lect­ing peo­ple who would like to use my plat­form to pro­duce pub­lic goods to so­licit re­quire­ments from them. If you’re in­ter­ested join our Dis­cord if you haven’t already.

      Won­der­ful that you’re work­ing on this! I’m with AI Safety Im­pact Mar­kets, and I sus­pect that we will need a sys­tem like this even­tu­ally. We haven’t re­ceived a lot of feed­back to the effect yet, so I haven’t pri­ori­tized it, but there are at least two ap­pli­ca­tions for it (for in­vestors and (one day, spec­u­la­tively) for im­pact buy­ers/​retrofun­ders). We’re cur­rently ad­dress­ing it with a bond­ing curve auc­tion of sorts, which in­cen­tivizes donors to come in early, so that they’re also not so in­cen­tivized to wait each other out. The in­cen­tive struc­tures are differ­ent, though, so maybe a com­bi­na­tion would be nice. That last link links to an ar­ti­cle that links to GiveWell’s donor’s dilemma. Some peo­ple out­side EA who I’ve talked to were sur­prised by it. They had found that donors flock to un­con­tro­ver­sially good dona­tion op­por­tu­ni­ties and don’t try to save their dona­tions for niche pro­jects. I’m think­ing that might be be­cause they donate to look good to oth­ers, which works bet­ter if they ac­tu­ally look good to oth­ers and not weird. So maybe DACs are mostly needed for co­or­di­na­tion among util­i­tar­i­ans or some range of con­se­quen­tial­ists? Then again these are differ­ent prob­lems – freerid­ing be­cause you want the benefits of the pub­lic good for free vs. freerid­ing so that you can use your funds for a niche pro­ject. Both defec­tions in as­surance games though. Sorry if I missed it in the ar­ti­cle, but who are the par­ties you would like to get on board to put up the re­fund bonuses? Can you boot­strap them by us­ing DACs to fundraise for big­ger and big­ger re­fund bonuses? Are you aware of any is­sues with se­cu­ri­ties law, since peo­ple can make mon­e­tary prof­its off re­fund bonuses? I might be able to get you in touch with some­one who knows these things when it comes to the US. I imag­ine (no idea re­ally) that the law of the coun­try of the per­son who starts a fundraiser will be what mat­ters legally. Can you maybe just pick out a char­ity fundrais­ing plat­form that has fund­ing thresh­olds and re­funds (and is used by con­se­quen­tial­ists), put Pol.is on top to rank the pro­jects by how un­con­tro­ver­sial they are, and then offer pro­por­tional (to the un­con­tro­ver­sial­ity) re­fund bonuses to ev­ery­one who con­tributed to failed fundraisers? Maybe they can ap­ply, prove that they’ve con­tributed to a failed fundraiser, and get the pay­out? Maybe, if you’re com­pletely in­de­pen­dent of the peo­ple who put up the fundraisers, you won’t risk get­ting them in trou­ble with the SEC? More philo­soph­i­cally: I think it’s hard to dis­t­in­guish freerid­ing from di­vi­sion of la­bor in economies of scale when there is no ex­plicit co­or­di­na­tion. DACs are prob­a­bly use­ful in the ma­jor­ity of cases, but there’s a minor­ity of cases where peo­ple (usu­ally rather al­tru­is­tic peo­ple) should pick their bat­tles, fo­cus on the stuff where they can make the great­est marginal con­tri­bu­tion, and “freeride” on the so­cial change that oth­ers effect in other ar­eas. Put differ­ently, I would love a non-speciesist world and a world where fund­ing in AI safety is al­lo­cated effi­ciently. I’m cur­rently work­ing only on the sec­ond prob­lem, so one could ar­gue that I’m freerid­ing on oth­ers’ solu­tions to the first. But in a differ­ent sense it’s just a di­vi­sion of la­bor among peo­ple who want to make the world a bet­ter place in gen­eral. So some seem­ing freerider prob­lems prob­a­bly need fix­ing with DACs, while oth­ers might not be prob­lems at all or might be im­proved with bet­ter com­mu­ni­ca­tion with­out any­one hav­ing to put up a re­fund bonus. (But money is flex­ible, so DACs for fundrais­ing are prob­a­bly ro­bustly use­ful.) Awe­some effort! Please keep us posted of how it’s go­ing! Btw., I think it would be use­ful to men­tion what the ar­ti­cle is about in the ti­tle. I would not have read it (even though I’m very in­ter­ested in the topic) if Dony hadn’t told me that it’s about DACs!

      Thanks for you com­ment, it’s very helpful.

      Sorry if I missed it in the ar­ti­cle, but who are the par­ties you would like to get on board to put up the re­fund bonuses?

      It would the pro­ducer of the pub­lic good (e.g. for my pro­ject I put up the col­lat­eral).

      Can you boot­strap them by us­ing DACs to fundraise for big­ger and big­ger re­fund bonuses?

      Pos­si­bly? I’m not sure why you’d do that?

      Are you aware of any is­sues with se­cu­ri­ties law, since peo­ple can make mon­e­tary prof­its off re­fund bonuses?

      I dis­agree that a Re­fund Bonus is a se­cu­rity. It’s a re­fund. To me it’s when you buy some­thing, but it comes bro­ken, so the store gives you a voucher to make up for your trou­bles.

      I imag­ine (no idea re­ally) that the law of the coun­try of the per­son who starts a fundraiser will be what mat­ters legally.

      I’m in South Africa but from what I can tell, if you work with US dol­lars and do some­thing ille­gal the FBI will come af­ter you, so I wouldn’t be con­fi­dent that only South Afri­can law ap­plies.

      Can you maybe just pick out a char­ity fundrais­ing plat­form that has fund­ing thresh­olds and re­funds (and is used by con­se­quen­tial­ists), and re­fund bonuses to ev­ery­one who con­tributed to failed fundraisers?

      This is ac­tu­ally a cool idea. I don’t know how I’d man­age to get peo­ple’s de­tails for giv­ing re­fund with­out co-op­er­at­ing with the fundrais­ing plat­form, and my im­pres­sion is that most plat­forms are hes­i­tant to do things like this. If you know of a plat­form that would be keen on try­ing this, please tell me!

      Put differ­ently, I would love a non-speciesist world and a world where fund­ing in AI safety is al­lo­cated effi­ciently. I’m cur­rently work­ing only on the sec­ond prob­lem, so one could ar­gue that I’m freerid­ing on oth­ers’ solu­tions to the first.

      I don’t quite un­der­stand this point. You could work on AI Safety and donate to an­i­mal char­i­ties if you don’t want to free-ride.

      Btw., I think it would be use­ful to men­tion what the ar­ti­cle is about in the ti­tle. I would not have read it (even though I’m very in­ter­ested in the topic) if Dony hadn’t told me that it’s about DACs!

      You’re right. I didn’t want the ti­tle to just be “Dom­i­nant As­surance Con­tracts” be­cause I as­sumed that most peo­ple have never heard of them and tried to come up with some­thing more in­ter­est­ing, but maybe enough peo­ple on less­wrong have heard of them so I should prob­a­bly be more straight for­ward.

      I dis­agree that a Re­fund Bonus is a se­cu­rity. It’s a re­fund. To me it’s when you buy some­thing, but it comes bro­ken, so the store gives you a voucher to make up for your trou­bles.

      Then per­haps we should call it ‘Re­fund Com­pen­sa­tion’ in­stead. I’d much rather if re­fund com­pen­sa­tion didn’t fall un­der se­cu­ri­ties laws and ev­ery­one could offer it has­sle-free.

      I re­ally like that. It draws at­ten­tion to the fact that the “bonus” is com­pen­sa­tion for time and effort spent on try­ing to buy a product that was never de­liv­ered.

      It would the pro­ducer of the pub­lic good (e.g. for my pro­ject I put up the col­lat­eral).

      Oh, got it! Thanks!

      Pos­si­bly? I’m not sure why you’d do that?

      I thought you’d be fundrais­ing to offer re­fund com­pen­sa­tion to oth­ers to make their fundraisers more likely to suc­ceed. But if the pro­ject de­vel­oper them­self put up the com­pen­sa­tion, it’s prob­a­bly also an im­por­tant sig­nal or se­lec­tion effect in the game the­o­retic setup.

      I dis­agree that a Re­fund Bonus is a se­cu­rity.

      Yeah, courts de­cide that in the end. Howey Test: money: yes; com­mon en­ter­prise: yes; ex­pec­ta­tion of profit: some­times; effort of oth­ers: I don’t know, not re­ally? The size of the pay­out is not se­cu­rity-like, but I don’t know if that mat­ters. All very un­clear. Profit: I imag­ine peo­ple will col­lect statis­tics on how close to funded cam­paigns can still be a day be­fore they close so that they still fail in (say) 90% of the cases. Then they blindly in­vest $x into all cam­paigns that are still $x away from that thresh­old on their last day. I imag­ine the courts may find that if some­one goes to such efforts to ex­ploit the sys­tem, they were prob­a­bly not tricked into do­ing so. Plus there is the ques­tion of what effort of oth­ers we could pos­si­bly be refer­ring to. But even if the courts in the end de­cide that you’re right and it’s not a se­cu­rity, the le­gal bat­tle with the SEC alone will be very ex­pen­sive… They keep ex­pand­ing their own heuris­tic for what they think is a se­cu­rity (not up to them to de­cide). They’ve even started to ig­nore the “ex­pec­ta­tion of profit” en­tirely (with sta­ble­coins). But per­haps you can find a way to keep the peo­ple who run the fundraisers in the clear and keep your com­pany in South Africa (where I know the laws even less though). If the fundraisers are on a main­stream blockchain, the trans­ac­tions are pub­lic, so you (out­side of the US) could man­age the re­fund com­pen­sa­tion on be­half of the pro­ject de­vel­op­ers and then pay re­funds ac­cord­ing to the pub­lic records on the blockchain. That way, no one could prove that a par­tic­u­lar pro­ject de­vel­oper is a mem­ber in your sys­tem… ex­cept maybe if they make “hon­ey­pot” con­tri­bu­tions I sup­pose. Per­haps you can have a sep­a­rate fund from which you re­ward con­trib­u­tors to pro­jects you like re­gard­less of whether they’re mem­bers. If a hon­ey­pot con­trib­u­tor gets a re­fund, they won’t know whether it’s be­cause the pro­ject de­vel­oper is a mem­ber of your org or be­cause you se­lected their pro­ject with­out them know­ing about it.

      This is ac­tu­ally a cool idea. I don’t know how I’d man­age to get peo­ple’s de­tails for giv­ing re­fund with­out co-op­er­at­ing with the fundrais­ing plat­form, and my im­pres­sion is that most plat­forms are hes­i­tant to do things like this. If you know of a plat­form that would be keen on try­ing this, please tell me!

      Yes… You could talk with Giveth about it. They’re us­ing blockchains, so per­haps you can build on top of them with­out them hav­ing to do any­thing. But what I’ve done in the past is that peo­ple sign up with my plat­form, get a code, and put the code in their pu­bic com­ment that they can at­tach to a con­tri­bu­tion on the third-party plat­form. Then if they want to claim any rights at­tached to the con­tri­bu­tion from me, I check that the code is the right one, and if it is, be­lieve them that they’re ei­ther the per­son who made the con­tri­bu­tion or that the per­son who made the con­tri­bu­tion wanted to gift it to them.

      I don’t quite un­der­stand this point. You could work on AI Safety and donate to an­i­mal char­i­ties if you don’t want to free-ride.

      Well, let’s say I barely have the money to pay for my own cost of liv­ing or that I con­sider a num­ber of AI safety orgs to also be the even-more-cost-effec­tive uses of my money.

      But what I’ve done in the past is that peo­ple sign up with my plat­form, get a code, and put the code in their pu­bic com­ment that they can at­tach to a con­tri­bu­tion on the third-party plat­form. Then if they want to claim any rights at­tached to the con­tri­bu­tion from me, I check that the code is the right one, and if it is, be­lieve them that they’re ei­ther the per­son who made the con­tri­bu­tion or that the per­son who made the con­tri­bu­tion wanted to gift it to them.

      Oh cool, that’s a good idea. Then you can piggy back off ex­ist­ing crowd­fund­ing plat­forms in­stead of mak­ing your own one.

      peo­ple sign up with my plat­form,

      Do you have a link? It sounds cool. I want to check it out.

      Well, let’s say I barely have the money to pay for my own cost of liv­ing or that I con­sider a num­ber of AI safety orgs to also be the even-more-cost-effec­tive uses of my money.

      I think I see your point. I agree that DACs don’t solve this type of free-rid­ing.

      Love the idea (con­tributed 10 dol­lars), but I don’t love the ti­tle—it feels a lit­tle click­baity (I ex­pected an ar­ti­cle about as­ter­oid deflec­tion, not about DACs).

      Yeah, some­thing de­scrip­tive like “A plat­form to fund pub­lic goods us­ing dom­i­nant as­surance con­tracts” or “A re­fund bonus crowd­fund­ing plat­form to sub­si­dize pub­lic goods” would be bet­ter in my opinion (and the jar­gon wouldn’t be a prob­lem in this com­mu­nity).

      I’m some­what flab­ber­gasted that no one has men­tioned rad­i­calxChange or quadratic fund­ing. They’re a solu­tion that doesn’t force the pro­ducer to take on all the risks of failure, at the cost of need­ing a cen­tral­ized pot of money.

      The cost of need­ing a cen­tral­ized pot of money is a big cost, and the re­quire­ment to have sev­eral pro­jects at once, make it a solu­tion to a very differ­ent class of prob­lems. That’s prob­a­bly the rea­son it wasn’t men­tioned, but still, thanks for men­tion­ing it be­cause I didn’t know about it. Upvoted. It would be in­ter­est­ing, though, to try to com­bine the two ideas. You can add an op­tion to the plat­form de­scribed in the post to pledge money to a cen­tral­ized pool that’s used to match pledges peo­ple make (and if they’re re­funded than the re­fund bonus on your match­ing pledge goes ei­ther to the pool or back to you). That way can you sup­port the pro­duc­tion of highly de­sired goods with­out hav­ing to ac­tu­ally choose by your­self which goods.

      1. The cost/​benefit ra­tio for the pop­u­la­tion of po­ten­tial pro­jects is bi­modal. They’re ei­ther so at­trac­tive that they have no trou­ble seek­ing donors and ex­ecu­tors via nor­mal Kick­starter, or so unattrac­tive that they’ll fail to se­cure fund­ing with or with­out a DAC.
      2. Even if DACs were nor­mal, Bob the Builder ex­poses him­self to fi­nan­cial risk in or­der to launch one. He has to in­crease his fund­ing goal in or­der to com­pen­sate, mak­ing the value propo­si­tion worse.

      For these rea­sons, it’s hard for me to get ex­cited about DACs.

      There is prob­a­bly a nar­row band of pro­jects where DACs are make-or-break, and be­cause you’re ex­cited about them, I think it’s great if you get the fund­ing you’re hop­ing for and suc­ceed in nor­mal­iz­ing them. Prove me wrong, by all means!