Proposal: Anti-Akrasia Alliance

*Re­lated to: [Kick­ing Akra­sia: now or never](http://​​less­wrong.com/​​r/​​dis­cus­sion/​​lw/​​3ko/​​kick­ing_akra­sia_now_or_never/​​); [Tsuyoku Nar­i­tai](http://​​less­wrong.com/​​lw/​​h8/​​tsuyoku_nar­i­tai_i_want_to_be­come_stronger/​​)*

# The situation

I am greatly af­ficted by akra­sia, and in all prob­a­bil­ity, so are you. Akra­sia is a de­stroyer of wor­lds<sup>1</​sup>.

I have come to the con­clu­sion that akra­sia is the sin­gle biggest prob­lem I have in life. It is greater than my im­pend­ing biolog­i­cal death, my im­perfect en­joy­ment of life, or the dan­ger of a car ac­ci­dent. For if I could solve the prob­lem of akra­sia, I would work on these other prob­lems, and I be­lieve I could solve them. Even a big prob­lem like phys­i­cal mor­tal­ity can be mean­ingfully challenged if I spend a life­time tack­ling it. But un­til I solve the prob­lem of akra­sia, I will sit around and *do noth­ing about my mor­tal­ity*.

Solv­ing akra­sia is nec­es­sary: with­out it, we can­not effi­ciently at­tack other prob­lems, and have a high chance of los­ing the fight against dis­ease, war, and UFAI.

Solv­ing akra­sia may also be suffi­cient: if a group of LW read­ers—smart, ra­tio­nal, lu­mi­nous, and rel­a­tively rich peo­ple—was also un­usu­ally ded­i­cated to a pur­pose, it would have a good chance of at­tack­ing Really Big Prob­lems rang­ing from per­sonal satis­fac­tion to Friendly AI<sup>2</​sup>.

Some peo­ple have solved this prob­lem, or never had it. Thus, we know it is pos­si­ble to van­quish akra­sia. How­ever, it is a unique prob­lem that pre­vents its own cure. Be­cause of akra­sia, we don’t spend as much effort as we’d like fight­ing akra­sia.

# Ex­ist­ing efforts

There have been many posts about akra­sia on LW. There are also many meth­ods and work­groups in the world which are ded­i­cated to it. I don’t know very many my­self, and no doubt there are many use­ful ap­proaches I don’t know about. (I know that other LW users have more rele­vant knowl­edge and ex­pe­rience than I do.)

I do know that there are sev­eral com­mon prob­lems with all such efforts:

1. Any given method, and even any well-defined com­pact com­bi­na­tion of meth­ods, works for some peo­ple but not for oth­ers, or works only some of the time.

2. When a method does work, it al­most always stops work­ing for that per­son af­ter a while, and can’t be used again.

3. Most meth­ods have no clear the­o­ries of the mechanisms be­hind them. Those that do, al­most always have bad the­o­ries, which don’t pre­dict why the method some­times fails, or are untestable just-so sto­ries.

4. Most meth­ods have no sci­en­tific back­ing of dou­ble-blind ex­per­i­ments, com­par­i­sons with other meth­ods, rul­ing out other ex­pla­na­tions, etc.

5. Many (most?) meth­ods perform no bet­ter than the placebo of do­ing some­thing that doesn’t ac­tu­ally work, feel­ing “pumped up”, and ex­pect­ing akra­sia to dis­ap­pear.

6. There are many differ­ent meth­ods out there, and a per­son can’t test them all to find one that works for them.

7. Most ex­ist­ing work­shops and groups are formed around a method, rather than around the goal of fight­ing akra­sia, and they ap­ply that method like a ham­mer to the ex­clu­sion of all oth­ers.

8. None of the meth­ods I have tried so far have helped me, per­son­ally.

Many of the meth­ods also share a prob­lem of trans­parency. If a method works for some peo­ple, they may pub­lish it widely, via books, web­sites, and work­shops. But there is usu­ally no-one col­lect­ing re­ports of cases where that method failed, in­ves­ti­gat­ing them, and pub­lish­ing up­dat­eds of the method’s ex­pected effec­tive­ness. In other words, there is a large pos­i­tive se­lec­tion bias for pub­li­ca­tion, and third-party re­ports aren’t re­viewed and pub­lished.

Fi­nally, there are few or­ga­nized at­tempts to col­late knowl­edge of many differ­ent meth­ods and test them in com­bi­na­tion.

# Proposal

I sug­gest that the first step in the fight against akra­sia should be to pro­claim the es­tab­lish­ment of a com­mu­nity ded­i­cated to this fight: the **Anti-Akra­sia Alli­ance**, or **3A** for short. A pub­lic com­mit­ment will help us to keep at­tack­ing the prob­lem.

Com­mit­ments are hard to keep (due to akra­sia). If we com­mit to try­ing spe­cific meth­ods and fol­low­ing var­i­ous reg­i­ments, we *will* of­ten fail. There­fore we should re­serve our willpower for just one Big Com­mit­ment: that of fight­ing akra­sia by par­ti­ci­pat­ing in 3A.

The dan­ger of stop­ping all progress due to akra­sia is big­ger, in the ini­tial stage, than the dan­ger of failing to find good solu­tions. We need mechanisms for mak­ing and keep­ing com­mit­ments, and for giv­ing these com­mit­ments a pos­i­tive af­fect.

We should es­tab­lish a web­site as a meet­ing-point. LW has the right for­mat (group blog + wiki), but if this is judged to be off-topic in the Dis­cus­sion sec­tion, I’ll set up a similar site some­where else.

Next, we should briefly ex­am­ine ex­ist­ing meth­ods and col­lect sug­ges­tions to de­ter­mine the course of ac­tion. Ideally, we should or­ga­nize the ex­ist­ing knowl­edge on the sub­ject, an­a­lyze mem­bers’ akra­sia case-his­to­ries, iden­tify likely the­o­ries, and get enough peo­ple to run the ap­pro­pri­ate ex­per­i­ments.

I be­lieve the ap­pli­ca­tion of ra­tio­nal­ity and the sci­en­tific method will, in it­self, give us a head start over other groups. We should also be trans­par­ent and open to new­com­ers and to im­port­ing new tech­niques. We should make good use of data col­lec­tion and anal­y­sis (nat­u­rally, al­low­ing for anonymity). We should, in short, use all the LW tech­niques.

# The next steps

I would like to hear your re­ac­tions in the com­ments. Do you think there is a bet­ter ap­proach? Do you think we’re bound to fail be­cause I didn’t take prob­lem X into ac­count? Is there a great, ra­tio­nal com­mu­nity fight­ing akra­sia that we should join in­stead of start­ing our own?

*Are you with me?*

I com­mit to post­ing an up­date by Jan 5th, which will take into ac­count the com­ments here. If the con­sen­sus is to es­tab­lish a web­site (and not use LW) I com­mit to set­ting one up by Jan 8th.

## Notes

1. We choose to cre­ate pos­si­ble fu­ture wor­lds. Akra­sia de­stroys our choices.

2. Per­son­ally, I might choose to in­vest effort in things other than FAI. Anti-akra­sia is a nec­es­sary meta-tool for hu­mans to achieve all hard goals.

No comments.