2016 LessWrong Diaspora Survey Results

Fore­word:

As we wrap up the 2016 sur­vey, I’d like to start by thank­ing ev­ery­body who took­
the time to fill it out. This year we had 3083 re­spon­dents, more than twice the
num­ber we had last year. (Source: http://​​less­wrong.com/​​lw/​​lhg/​​2014_sur­vey_re­sults/​​)
This seems con­sis­tent with the hy­poth­e­sis that the LW com­mu­nity hasn’t de­clined
in pop­u­la­tion so much as mi­grated into differ­ent com­mu­ni­ties. Be­ing the *di­as­pora*
sur­vey I had ex­pec­ta­tions for more re­sponses than usual, but twice as many was
far be­yond them.

Be­fore we move on to the sur­vey re­sults, I feel obli­gated to put a few af­fairs
in or­der in re­gards to what should be done next time. The copy­right situ­a­tion
for the sur­vey was am­bigu­ous this year, and to pre­vent that from hap­pen­ing again
I’m pleased to an­nounce that this years sur­vey ques­tions will be re­leased jointly
by me and Scott Alexan­der as Creative Com­mons li­censed con­tent. We haven’t
fi­nal­ized the de­tails of this yet so ex­pect it some­time this month.

I would also be re­miss not to men­tion the large amount of feed­back we re­ceived
on the sur­vey. Some of which led to ac­tion­able recom­men­da­tions I’m go­ing to
pre­serve here for who­ever does it next:

- Put free re­sponse form at the very end to sug­gest im­prove­ments/​com­plain.

- Fix metaethics ques­tion in gen­eral, lots of op­tions peo­ple felt were miss­ing.

- Clean up defi­ni­tions of poli­ti­cal af­fila­tions in the short poli­tics sec­tion.
In par­tic­u­lar, ‘Com­mu­nist’ has an overly ag­gres­sive/​nega­tive defi­ni­tion.

- Pos­si­bly com­pletely over­haul short poli­tics sec­tion.

- Every­where that a non-an­swer is taken as an an­swer should be changed so that
non an­swer means what it ought to, no an­swer or opinion. “Ab­sence of a sig­nal
should never be used as a sig­nal.”—Ju­lian Bigelow, 1947

- Give a defi­ni­tion for the sin­gu­lar­ity on the ques­tion ask­ing when you think it
will oc­cur.

- Ask if peo­ple are *cur­rently* suffer­ing from de­pres­sion. Pos­si­bly add more
prob­ing ques­tions on de­pres­sion in gen­eral since the rates are so ex­traor­di­nar­ily
high.

- In­clude a link to what cis­gen­der means on the gen­der ques­tion.

- Spec­ify if the in­come ques­tion is be­fore or af­ter taxes.

- Add char­ity ques­tions about time donated.

- Add “in­el­i­gible to vote” op­tion to the vot­ing ques­tion.

- Ad­ding some way for those who are preg­nant to in­di­cate it on the num­ber of
chil­dren ques­tion would be nice. It might be oner­ous how­ever so don’t feel
obli­gated. (Re­mem­ber that it’s more im­por­tant to have a smooth sur­vey than it
is to catch ev­ery edge case.)

And read this thread: http://​​less­wrong.com/​​lw/​​nfk/​​less­wrong_2016_sur­vey/​​,
it’s full of sug­ges­tions, cor­rec­tions and crit­i­cism.

Without fur­ther ado,

Ba­sic Re­sults:

2016 LessWrong Di­as­pora Sur­vey Ques­tions (PDF For­mat)

2016 LessWrong Di­as­pora Sur­vey Re­sults (PDF For­mat, Miss­ing 23 Re­sponses)

2016 LessWrong Di­as­pora Sur­vey Re­sults Com­plete (Text For­mat, Null En­tries In­cluded)

2016 LessWrong Di­as­pora Sur­vey Re­sults Com­plete (Text For­mat, Null En­tries Ex­cluded)

2016 LessWrong Di­as­pora Sur­vey Re­sults Com­plete (Text For­mat, Null En­tries In­cluded, 13 Re­sponses Filtered, Per­centages)

2016 LessWrong Di­as­pora Sur­vey Re­sults Com­plete (Text For­mat, Null En­tries Ex­cluded, 13 Re­sponses Filtered, Per­centages)

2016 LessWrong Di­as­pora Sur­vey Re­sults Com­plete (HTML For­mat, Null En­tries Ex­cluded)

Our re­port sys­tem is cur­rently on the fritz and isn’t calcu­lat­ing nu­meric ques­tions. If I’d known this ear­lier I’d have pre­pared the re­sults for said ques­tions ahead of time. In­stead they’ll be com­ing out later to­day or to­mor­row. (EDIT: Th­ese re­sults are now in the text for­mat sur­vey re­sults.)

Philos­o­phy and Com­mu­nity Is­sues At LessWrong’s Peak (Write Ins)

Peak Philos­o­phy Is­sues Write Ins (Part One)

Peak Philos­o­phy Is­sues Write Ins (Part Two)

Peak Com­mu­nity Is­sues Write Ins (Part One)

Peak Com­mu­nity Is­sues Write Ins (Part Two)

Philos­o­phy and Com­mu­nity Is­sues Now (Write Ins)

Philos­o­phy Is­sues Now Write Ins (Part One)

Philos­o­phy Is­sues Now Write Ins (Part Two)

Com­mu­nity Is­sues Now Write Ins (Part One)

Com­mu­nity Is­sues Now Write Ins (Part Two)

Re­join Conditions

Re­join Con­di­tion Write Ins (Part One)

Re­join Con­di­tion Write Ins (Part Two)

Re­join Con­di­tion Write Ins (Part Three)

Re­join Con­di­tion Write Ins (Part Four)

Re­join Con­di­tion Write Ins (Part Five)

CC-Li­censed Ma­chine Read­able Sur­vey and Public Data

2016 LessWrong Di­as­pora Sur­vey Struc­ture (Li­cense)

2016 LessWrong Di­as­pora Sur­vey Public Dataset

(Note for peo­ple look­ing to work with the dataset: My sur­vey anal­y­sis code repos­i­tory in­cludes a sqlite con­verter, ex­am­ples, and more com­ing soon. It’s a great way to get up and run­ning with the dataset re­ally quickly.)

In depth anal­y­sis:

Anal­y­sis Posts

Part One: Meta and Demographics

Part Two: LessWrong Use, Suc­ces­sor­ship, Diaspora

Part Three: Men­tal Health, Basilisk, Blogs and Media

Part Four: Poli­tics, Cal­ibra­tion & Prob­a­bil­ity, Fu­tur­ol­ogy, Char­ity & Effec­tive Altruism

Ag­gre­gated Data

Effec­tive Altru­ism and Char­i­ta­ble Giv­ing Analysis

Men­tal Health Stats By Di­as­pora Com­mu­nity (In­clud­ing self dx­ers)

How Di­as­pora Com­mu­ni­ties Com­pare On Men­tal Health Stats (I sus­pect these charts are sub­tly bro­ken some­how, will in­ves­ti­gate later)

Im­proved Men­tal Health Charts By Obor­mot (Us­ing pub­lic sur­vey data)

Im­proved Men­tal Health Charts By Anony­mous (Us­ing full sur­vey data)

Poli­ti­cal Opinions By Poli­ti­cal Affiliation

Poli­ti­cal Opinions By Poli­ti­cal Affili­a­tion Charts (By anony­mous)

Blogs And Me­dia De­mo­graphic Clusters

Blogs And Me­dia De­mo­graphic Clusters (HTML For­mat, Im­pos­si­ble An­swers Ex­cluded)

Cal­ibra­tion Ques­tion And Brier Score Analysis

More com­ing soon!

Sur­vey Anal­y­sis Code

Some notes:

1. FortFore­cast on the com­mu­ni­ties sec­tion, Bayesed And Con­fused on the blogs sec­tion, and Syn­the­sis on the sto­ries sec­tion were all ‘troll’ an­swers de­signed to catch peo­ple who just put down ev­ery­thing. Some­body noted that the three ‘fort­fore­cast’ users had the en­tire DSM split up be­tween them, that’s why.

2. Lots of peo­ple asked me for a list of all those cool blogs and sto­ries and com­mu­ni­ties on the sur­vey, they’re in­cluded in the sur­vey ques­tions PDF above.

Public TODO:

1. Add more in depth anal­y­sis, fix the ones that de­cided to sud­denly break at the last minute or I sus­pect were always bro­ken.

2. Add a com­pat­i­bil­ity mode so that the cur­rent ques­tion codes are con­verted to older ones for 3rd party anal­y­sis that rely on them.

If any­body would like to help with these, write to jd@fort­fore­cast.com