Why artificial optimism?

Link post

Op­ti­mism bias is well-known. Here are some ex­am­ples.

  • It’s con­ven­tional to an­swer the ques­tion “How are you do­ing?” with “well”, re­gard­less of how you’re ac­tu­ally do­ing. Why?

  • Peo­ple of­ten be­lieve that it’s in­her­ently good to be happy, rather than think­ing that their hap­piness level should track the ac­tual state of af­fairs (and thus be a use­ful tool for emo­tional pro­cess­ing and com­mu­ni­ca­tion). Why?

  • Peo­ple of­ten think their pro­ject has an un­re­al­is­ti­cally high chance of suc­ceed­ing. Why?

  • Peo­ple of­ten avoid look­ing at hor­rible things clearly. Why?

  • Peo­ple of­ten want to sup­press crit­i­cism but less of­ten want to sup­press praise; in gen­eral, they hold crit­i­cism to a higher stan­dard than praise. Why?

The parable of the gullible king

Imag­ine a king­dom ruled by a gullible king. The king gets re­ports from differ­ent re­gions of the king­dom (man­aged by differ­ent vas­sals). Th­ese re­ports de­tail how things are go­ing in these differ­ent re­gions, in­clud­ing par­tic­u­lar events, and an over­all sum­mary of how well things are go­ing. He is quite gullible, so he usu­ally be­lieves these re­ports, al­though not if they’re too out­landish.

When he thinks things are go­ing well in some re­gion of the king­dom, he gives the vas­sal more re­sources, ex­pands the re­gion con­trol­led by the vas­sal, en­courages oth­ers to copy the prac­tices of that re­gion, and so on. When he thinks things are go­ing poorly in some re­gion of the king­dom (in a long-term way, not as a tem­po­rary crisis), he gives the vas­sal fewer re­sources, con­tracts the re­gion con­trol­led by the vas­sal, en­courages oth­ers not to copy the prac­tices of that re­gion, pos­si­bly re­places the vas­sal, and so on. This be­hav­ior makes sense if he’s as­sum­ing he’s get­ting re­li­able in­for­ma­tion: it’s bet­ter for prac­tices that re­sult in bet­ter out­comes to get copied, and for places with higher eco­nomic growth rates to get more re­sources.

Ini­tially, this works well, and good prac­tices are adopted through­out the king­dom. But, some vas­sals get the idea of ex­ag­ger­at­ing how well things are go­ing in their own re­gion, while den­i­grat­ing other re­gions. This re­sults in their own re­gion get­ting more ter­ri­tory and re­sources, and their prac­tices be­ing adopted el­se­where.

Soon, these dis­tor­tions be­come ubiquitous, as the king (un­wit­tingly) en­courages ev­ery­one to adopt them, due to the ap­par­ent suc­cess of the re­gions dis­tort­ing in­for­ma­tion this way. At this point, the vas­sals face a prob­lem: while they want to ex­ag­ger­ate their own re­gion and den­i­grate oth­ers, they don’t want oth­ers to den­i­grate their own re­gion. So, they start form­ing al­li­ances with each other. Vas­sals that ally with each other promise to say only good things about each other’s re­gions. That way, both vas­sals mu­tu­ally benefit, as they both get more re­sources, ex­pan­sion, etc com­pared to if they had been den­i­grat­ing each other’s re­gions. Th­ese al­li­ances also make sure to keep den­i­grat­ing those not in the same coal­i­tion.

While these “praise coal­i­tions” are lo­cally pos­i­tive-sum, they’re globally zero-sum: any gains that come from them (such as re­sources and ter­ri­tory) are taken from other re­gions. (How­ever, hav­ing more praise over­all helps the vas­sals cur­rently in power, as it means they’re less likely to get re­placed with other vas­sals).

Since praise coal­i­tions lie, they also sup­press the truth in gen­eral in a co­or­di­nated fash­ion. It’s con­sid­ered im­po­lite to re­veal cer­tain forms of in­for­ma­tion that could im­ply that things aren’t ac­tu­ally go­ing as well as they’re say­ing it’s go­ing. Pry­ing too closely into a re­gion’s ac­tual state of af­fairs (and, es­pe­cially, shar­ing this in­for­ma­tion) is con­sid­ered a vi­o­la­tion of pri­vacy.

Mean­while, the ac­tual state of af­fairs has got­ten worse in al­most all re­gions, though the re­gions prop up their lies with Potemkin villages, so the gullible king isn’t shocked when he vis­its the re­gion.

At some point, a sin­gle praise coal­i­tion wins. Vas­sals no­tice that it’s in their in­ter­est to join this coal­i­tion, since (as men­tioned be­fore) it’s in the in­ter­ests of the vas­sals as a class to have more praise over­all, since that means they’re less likely to get re­placed. (Of course, it’s also in their class in­ter­ests to have things ac­tu­ally be go­ing well in their re­gions, so the praise doesn’t get too out of hand, and crit­i­cism is some­times ac­cepted) At this point, it’s con­ven­tional for vas­sals to always praise each other and pun­ish vas­sals who den­i­grate other re­gions.

Op­ti­mism isn’t ubiquitous, how­ever. There are a few strate­gies vas­sals can use to use pes­simism to claim more re­sources. Among these are:

  • Blame: By claiming a vas­sal is do­ing some­thing wrong, an­other vas­sal may be able to take power away from that vas­sal, some­times get­ting a share of that power for them­selves. (Blame is of­ten not es­pe­cially difficult, given that ev­ery­one’s in­flat­ing their im­pres­sions)

  • Pity: By show­ing that their re­gion is un­der­go­ing a tem­po­rary but fix­able crisis (per­haps with the help of other vas­sals), vas­sals can claim that they should be get­ting more re­sources. But, the prob­lem has to be solv­able; it has to be a tem­po­rary crises, not a per­ma­nent state of de­cay. (One form of pity is claiming to be vic­tim­ized by an­other vas­sal; this mixes blame and pity)

  • Doom­say­ing: By claiming that there is some threat to the king­dom (such as wolves), vas­sals can claim that they should be get­ting re­sources in or­der to fight this threat. Again, the threat has to be solv­able; the king has lit­tle rea­son to give some­one more re­sources if there is, in­deed, noth­ing to do about the threat.

Pity and doom­say­ing could be seen as two sides of the same coin: pity claims things are go­ing poorly (but fix­ably) lo­cally, while doom­say­ing claims things are go­ing poorly (but fix­ably) globally. How­ever, all of these strate­gies are limited to a sig­nifi­cant de­gree by the over­all praise coal­i­tion, so they don’t get out of hand.

Back to the real world

Let’s re­late the parable of the gullible king back to the real world.

  • The king is some­times an ac­tual per­son (such as a CEO, as in Mo­ral Mazes, or a philan­thropist), but is more of­ten a pro­cess dis­tributed among many peo­ple that is eval­u­at­ing which things are good/​bad, in a pat­tern-match­ing way.

  • Every­one’s a vas­sal to some de­gree. Peo­ple who have more power-through-ap­pear­ing-good are vas­sals with more ter­ri­tory, who have more of an in­ter­est in main­tain­ing pos­i­tive im­pres­sions.

  • Most (al­most all?) coal­i­tions in the real world have as­pects of praise coal­i­tions. They’ll praise those in the coal­i­tion while den­i­grat­ing those out­side it.

  • Po­lite­ness and pri­vacy are, in fact, largely about main­tain­ing im­pres­sions (es­pe­cially pos­i­tive im­pres­sions) through co­or­di­nat­ing against the rev­e­la­tion of truth.

  • Main­tain­ing us-vs-them bound­aries is char­ac­ter­is­tic of the poli­ti­cal right, while dis­solv­ing them (and pun­ish­ing those try­ing to set them up) is char­ac­ter­is­tic of the poli­ti­cal left. So, non-to­tal­iz­ing praise coal­i­tions are more char­ac­ter­is­tic of the right, and to­tal ones that try to as­similate oth­ers (such as the one that won in the parable) are more char­ac­ter­is­tic of the left. (Note, to­tal­iz­ing praise coal­i­tions still den­i­grate/​at­tack ones that can’t be safely as­similated; see the para­dox of tol­er­ance)

  • Coal­i­tions may be frac­tal, of course.

  • A lot of the dis­tor­tionary dy­nam­ics are sub­con­scious (see: The Elephant in the Brain).

This model raises an im­por­tant ques­tion (with im­pli­ca­tions for the real world): if you’re a de­tec­tive in the king­dom of the gullible king who is at least some­what aware of the re­al­ity of the situ­a­tion and the dis­tor­tonary dy­nam­ics, and you want to fix the situ­a­tion (or at least re­duce harm), what are your op­tions?