On Expressing Your Concerns

The scary thing about Asch’s con­for­mity ex­per­i­ments is that you can get many peo­ple to say black is white, if you put them in a room full of other peo­ple say­ing the same thing. The hope­ful thing about Asch’s con­for­mity ex­per­i­ments is that a sin­gle dis­sen­ter tremen­dously drove down the rate of con­for­mity, even if the dis­sen­ter was only giv­ing a differ­ent wrong an­swer. And the weari­some thing is that dis­sent was not learned over the course of the ex­per­i­ment—when the sin­gle dis­sen­ter started sid­ing with the group, rates of con­for­mity rose back up.

Be­ing a voice of dis­sent can bring real benefits to the group. But it also (fa­mously) has a cost. And then you have to keep it up. Plus you could be wrong.

I re­cently had an in­ter­est­ing ex­pe­rience wherein I be­gan dis­cussing a pro­ject with two peo­ple who had pre­vi­ously done some plan­ning on their own. I thought they were be­ing too op­ti­mistic and made a num­ber of safety-mar­gin-type sug­ges­tions for the pro­ject. Soon a fourth guy wan­dered by, who was pro­vid­ing one of the other two with a ride home, and be­gan mak­ing sug­ges­tions. At this point I had a sud­den in­sight about how groups be­come over­con­fi­dent, be­cause when­ever I raised a pos­si­ble prob­lem, the fourth guy would say, “Don’t worry, I’m sure we can han­dle it!” or some­thing similarly re­as­sur­ing.

An in­di­vi­d­ual, work­ing alone, will have nat­u­ral doubts. They will think to them­selves, “Can I re­ally do XYZ?” be­cause there’s noth­ing im­po­lite about doubt­ing your own com­pe­tence. But when two un­con­fi­dent peo­ple form a group, it is po­lite to say nice and re­as­sur­ing things, and im­po­lite to ques­tion the other per­son’s com­pe­tence. To­gether they be­come more op­ti­mistic than ei­ther would be on their own, each one’s doubts quel­led by the other’s seem­ingly con­fi­dent re­as­surance, not re­al­iz­ing that the other per­son ini­tially had the same in­ner doubts.

The most fear­some pos­si­bil­ity raised by Asch’s ex­per­i­ments on con­for­mity is the specter of ev­ery­one agree­ing with the group, swayed by the con­fi­dent voices of oth­ers, care­ful not to let their own doubts show—not re­al­iz­ing that oth­ers are sup­press­ing similar wor­ries. This is known as “plu­ral­is­tic ig­no­rance.”

Robin Han­son and I have a long-run­ning de­bate over when, ex­actly, as­piring ra­tio­nal­ists should dare to dis­agree. I tend to­ward the widely held po­si­tion that you have no real choice but to form your own opinions. Robin Han­son ad­vo­cates a more icon­o­clas­tic po­si­tion, that you—not just other peo­ple—should con­sider that oth­ers may be wiser. Re­gard­less of our var­i­ous dis­putes, we both agree that Au­mann’s Agree­ment The­o­rem ex­tends to im­ply that com­mon knowl­edge of a fac­tual dis­agree­ment shows some­one must be ir­ra­tional.1 De­spite the funny looks we’ve got­ten, we’re stick­ing to our guns about mod­esty: For­get what ev­ery­one tells you about in­di­vi­d­u­al­ism, you should pay at­ten­tion to what other peo­ple think.

Ahem. The point is that, for ra­tio­nal­ists, dis­agree­ing with the group is se­ri­ous busi­ness. You can’t wave it off with, “Every­one is en­ti­tled to their own opinion.”

I think the most im­por­tant les­son to take away from Asch’s ex­per­i­ments is to dis­t­in­guish “ex­press­ing con­cern” from “dis­agree­ment.” Rais­ing a point that oth­ers haven’t voiced is not a promise to dis­agree with the group at the end of its dis­cus­sion.

The ideal Bayesian’s pro­cess of con­ver­gence in­volves shar­ing ev­i­dence that is un­pre­dictable to the listener. The Au­mann agree­ment re­sult holds only for com­mon knowl­edge, where you know, I know, you know I know, etc. Han­son’s post or pa­per on “We Can’t Fore­see to Disagree” pro­vides a pic­ture of how strange it would look to watch ideal ra­tio­nal­ists con­verg­ing on a prob­a­bil­ity es­ti­mate; it doesn’t look any­thing like two bar­gain­ers in a mar­ket­place con­verg­ing on a price.

Un­for­tu­nately, there’s not much differ­ence so­cially be­tween “ex­press­ing con­cerns” and “dis­agree­ment.” A group of ra­tio­nal­ists might agree to pre­tend there’s a differ­ence, but it’s not how hu­man be­ings are re­ally wired. Once you speak out, you’ve com­mit­ted a so­cially ir­re­vo­ca­ble act; you’ve be­come the nail stick­ing up, the dis­cord in the com­fortable group har­mony, and you can’t undo that. Any­one in­sulted by a con­cern you ex­pressed about their com­pe­tence to suc­cess­fully com­plete task XYZ will prob­a­bly hold just as much of a grudge af­ter­ward if you say, “No prob­lem, I’ll go along with the group,” at the end.

Asch’s ex­per­i­ment shows that the power of dis­sent to in­spire oth­ers is real. Asch’s ex­per­i­ment shows that the power of con­for­mity is real. If ev­ery­one re­frains from voic­ing their pri­vate doubts, that will in­deed lead groups into mad­ness. But his­tory abounds with les­sons on the price of be­ing the first, or even the sec­ond, to say that the Em­peror has no clothes. Nor are peo­ple hard­wired to dis­t­in­guish “ex­press­ing a con­cern” from “dis­agree­ment even with com­mon knowl­edge”; this dis­tinc­tion is a ra­tio­nal­ist’s ar­tifice. If you read the more cyn­i­cal brand of self-help books (e.g., Machi­avelli’s The Prince) they will ad­vise you to mask your non­con­for­mity en­tirely, not voice your con­cerns first and then agree at the end. If you perform the group ser­vice of be­ing the one who gives voice to the ob­vi­ous prob­lems, don’t ex­pect the group to thank you for it.

Th­ese are the costs and the benefits of dis­sent­ing—whether you “dis­agree” or just “ex­press con­cern”—and the de­ci­sion is up to you.