Speaking for myself (re: how the LW2.0 team communicates)

Posts made by mem­bers of the LessWrong 2.0 team are typ­i­cally made from the per­spec­tive of the in­di­vi­d­ual—even when they are writ­ing in their ca­pac­ity as LessWrong team mem­bers. My (Ruby’s) model of the team’s rea­son for this is that even if there are col­lec­tive de­ci­sions, there are no col­lec­tive mod­els. Not real mod­els.

When the team agrees to do some­thing, it is only be­cause enough of the in­di­vi­d­ual team mem­bers in­di­vi­d­u­ally have mod­els which in­di­cate it is the right thing to do. Our mod­els might be roughly the same at a high-level such that you can de­scribe a “com­mon de­nom­i­na­tor” model, but this isn’t an ac­tual model held by an ac­tual per­son. I think such “com­mon de­nom­i­na­tor” group mod­els are un­de­sir­able for at least the fol­low­ing rea­sons:

  1. Pres­sure to form con­sen­sus re­duces the di­ver­sity of mod­els, some­times go­ing from mul­ti­ple mod­els per per­son to only a sin­gle model for the group. This can then re­sult in over­con­fi­dence in the sur­viv­ing model.

  2. There might be no group model. The group might have agreed on a de­ci­sion, but they never reached con­sen­sus on the rea­sons for it.

  3. It is costly to de­scribe group mod­els. Either some­one has to draft the model, get feed­back, make re­vi­sions, re­peat, un­til even­tu­ally it is “good enough” or some­one de­scribes a model pu­ta­tively held by the group, but which is not ac­tu­ally rep­re­sen­ta­tive of the group’s think­ing.

  4. In fact, no in­di­vi­d­ual might en­dorse the group model as be­ing their own.

  5. The per­son de­scribing the group model doesn’t nec­es­sar­ily un­der­stand things they’re in­clud­ing which came from oth­ers.

  6. In av­er­ag­ing mul­ti­ple mod­els, de­tail is lost and you no longer have a model which can use­fully gen­er­ate pre­dic­tions.

  7. No in­di­vi­d­ual owns the model, mak­ing it hard for any one per­son to elu­ci­date, defend, or be held ac­countable for it.

  8. Reluc­tance to speak on the be­half of oth­ers means that very lit­tle gets said.

Cru­cially, group mod­els which get shared ex­ter­nally are very of­ten not the mod­els which were used to make de­ci­sions. If you want to un­der­stand a de­ci­sion, you want the ac­tual model which gen­er­ated it.

Given the goal of shar­ing our ac­tual true think­ing with the out­side world (rather than nicely cu­rated PR an­nounce­ments), the LessWrong team has the rough policy that we speak from our per­sonal point of view and avoid com­mit­ting to an im­per­sonal, au­thor­i­ta­tive, “offi­cial view of LessWrong.”

I sus­pect (and I be­lieve the team gen­er­ally agrees) that in­di­vi­d­ual team mem­bers post­ing from their own points of view will ul­ti­mately re­sult in the out­side world hav­ing a bet­ter un­der­stand­ing of our think­ing (in­di­vi­d­ual and col­lec­tive) than if we at­tempt to ag­gre­gate our in­di­vi­d­ual mod­els into the “or­ga­ni­za­tion’s mod­els”. Or­ga­ni­za­tions don’t have mod­els, peo­ple do.

That said, we talk a lot to each other and our mod­els are cor­re­lated. We tend to agree on the broad out­line of things, e.g. we agree at the crud­est level that LessWrong is about ra­tio­nal­ity and in­tel­lec­tual progress, even if we don’t agree on more de­tailed fram­ings and rel­a­tive em­pha­sis. We think roughly like each other, but don’t be sur­prised if a differ­ent team mem­ber says about a high-level vi­sion post I wrote that it’s not their model of it or that they don’t agree with ev­ery de­tail.

Seem­ingly, this com­mu­ni­ca­tion policy might al­low us (the LessWrong team) to weasel out of our pub­lic state­ments. “Oh, that’s just what Ruby said—the rest of us never said that.” This is far from the in­ten­tion. This policy is fo­cused on how we com­mu­ni­cate our rea­sons for do­ing things rather than state­ments about our com­mit­ments or ac­tions. If a LessWrong team mem­ber says the LessWrong team plans to do some­thing (es­pe­cially ma­jor di­rec­tions), it’s fair game to hold the en­tire team ac­countable for do­ing that thing.