# SoerenMind

Karma: 884
Page 1
• That fair, I made a math­e­mat­i­cal er­ror there. The cluster headache math con­vinces me that a large chunk of to­tal suffer­ing goes to few peo­ple there due to lop­sided fre­quen­cies. Do you have other ex­am­ples? I par­tic­u­larly felt that the rel­a­tive fre­quency of ex­treme com­pared to less ex­treme pain wasn’t well sup­ported.

• Your 4 cluster headache groups con­tribute about equally to the to­tal num­ber of cluster headaches if you mul­ti­ply group size by # of CH’s. (The top 2% ac­tu­ally con­tribute a bit less). That’s my en­tire point. I’m not sure if you dis­agree?

• To the sec­ond half of your com­ment, I agree that ex­treme suffer­ing can be very ex­treme and I think this is an im­por­tant con­tri­bu­tion. Maybe we have a mi­s­un­der­stand­ing about what ‘the bulk’ of suffer­ing refers to. To me it means some­thing like 75-99% and to you it means some­thing like 45% as stated above? I should also clar­ify that by fre­quency I mean the product of ‘how many peo­ple have it’, ‘how of­ten’ and ‘for how long’.

“the peo­ple in the top 10% of suffer­ers will have 10X the amount, and peo­ple in the 99% [I as­sume you mean top 1%?] will have 100X the amount”

I’m con­fused, you seem to be sug­gest­ing that ev­ery level of pain ac­counts for the _same_ amount of to­tal suffer­ing here.

To elab­o­rate, you seem to be say­ing that at any level of pain, 10x worse pain is also 10x less fre­quent. That’s a power law with ex­po­nent 1. I.e. the lev­els of pain have an ex­treme dis­tri­bu­tion, but the fre­quen­cies do too (mild pains are ex­tremely com­mon). I’m not say­ing you’re wrong—just that I’ve seen also seems con­sis­tent with ex­treme pain be­ing less than 10% of the to­tal. I’m ex­cited to see more data :)

• Aside from my con­cern about ex­treme pain be­ing rarer than or­di­nary pain, I also would find the con­clu­sion that

″...the bulk of suffer­ing is con­cen­trated in a small per­centage of ex­pe­riences...”

very sur­pris­ing. Stan­dard com­pu­ta­tional neu­ro­science de­ci­sion-mak­ing views such as RL mod­els would say that if this is true, an­i­mals would have to spend most of their ev­ery­day effort try­ing to avoid ex­treme pain. But that seems wrong. E. g. we seek food to re­lieve mild hunger and get a nice taste and not be­cause we once had a an ex­treme hunger ex­pe­rience that we learned from.

You could ar­gue that the learn­ing from ex­treme pain doesn’t track the sub­jec­tive in­ten­sity of pain. But then peo­ple would be choos­ing e. g. a sub­jec­tively 10x worse pain over a <10x longer pain. In this cause I’d prob­a­bly say that the sub­jec­tive im­pres­sion is mis­guided or eth­i­cally ir­rele­vant, though that’s an eth­i­cal judg­ment.

• Thanks. I was ac­tu­ally ask­ing about a differ­ent fre­quency dis­tri­bu­tion. You’re talk­ing about the fre­quency of ex­treme pain among peo­ple with ex­treme pain which has no bear­ing on the quote above. I’m talk­ing about the fre­quency of ex­treme pain ex­pe­riences among all pain ex­pe­riences (i. e. is ex­treme pain it lmuch less preva­lent). Hence the ex­am­ple about mild dis­com­fort.

• Great anal­y­sis!

″...the bulk of suffer­ing is con­cen­trated in a small per­centage of ex­pe­riences...”

This seems like your core im­pli­ca­tion. But it re­quires an ar­gu­ment about in­ten­sity dis­tri­bu­tion and fre­quency dis­tri­bu­tion. There’s only ar­gu­ments about the first one if I haven’t missed any­thing? To illus­trate, I have mild dis­com­fort about 8000s/​day on av­er­age but ex­treme pain per­haps 0.02s/​day, if I get 1h of ex­treme pain in my life (and many peo­ple don’t get any at all).

• Great work! I won­der if there are any ways to track qual­ity ad­justed en­gage­ment since that what we’ve mostly been op­ti­miz­ing for the last few years. E. g. if low-qual­ity page views/​joins/​listen­ers are go­ing down it seems hard to com­pen­sate with an equal num­ber of high qual­ity ones be­cause they’re harder to cre­ate. 80k’s im­pact ad­justed plan changes met­ric is the only suit­able met­ric I can think of.

• PMed you the pay­walled re­view. There seems to be some agree­ment that ev­i­dence trans­fers be­tween differ­ent ten­dons FYI, e. g. some stud­ies are about Achilles ten­dons. The spe­cific re­view on golfer arm (seen by my doc as nearly equiv­a­lent to RSI on the hand-fac­ing ten­dons) is also in my mes­sage. If you want to talk to an ex­pert about the ev­i­dence you can prob­a­bly ask to skype him for a fee.

• PMed, and yes. The ex­er­cise the doc gave me was to hold it with both hands fac­ing down and then al­ter­nat­ingly bend into an in­verted /​ nor­mal u-shape. This hits both flex­ors and ex­ten­sors and it’s both ec­cen­tric and con­cen­tric com­bined.

• Many poli­cies are later re­voked and aren’t about trad­ing off pre­sent vs fu­ture re­sources (e. g. in­come re­dis­tri­bu­tion). So those who are still al­ive when a policy’s effects stop got more than their fair share of vot­ing power un­der this pro­posal if I un­der­stand cor­rectly. E. g. if I’m 80 when a policy against re­dis­tri­bu­tion comes into effect, and it’s re­voked when I die at 84, my 1x vote weight­ing seems un­fair be­cause ev­ery­one else was also just af­fected for 4 years.

• Very nicely writ­ten! Typo: “and the tak­ing ac­tion on that ba­sis”

• This post seems to be miss­ing the ther­apy with the best ev­i­dence ba­sis—heavy loaded ec­cen­tric train­ing. See e. g. https://​​www.up­to­date.com/​​con­tents/​​overview-of-the-man­age­ment-of-overuse-per­sis­tent-tend­inopa­thy (pay­walled). The com­bi­na­tion with con­cen­tric train­ing is al­most as well sup­ported and eas­ier to do. The only tool needed is a flexbar. 3x 15 reps twice daily for 3 months should bring re­sults.

The web­site painscience.com is a great SSC-like read but I’ve found it lack­ing from time to time, for in­stance by omit­ting ec­cen­tric train­ing.

I can also recom­mend a pro­fes­sor in Ger­many who spe­cial­izes in ten­don prob­lems and charges ca 150eur for a 30-60m ses­sion plus email sup­port. I could even imag­ine him do­ing a skype ses­sion with some con­vinc­ing but he’ll want to get an ul­tra sound and strength test. He was recom­mended to me by a paid ser­vice in Ger­many (bet­ter­doc) that asks a coun­cil of med­i­cal ex­perts for the lead­ing ex­pert for a de­sease. The pro­fes­sor web­site: http://​​www.sport­praxis-knobloch.de/​​

• Pos­si­ble am­bi­guity in the sur­vey ques­tion: If the per­son stops work­ing “for you or any­one for 3 years” that plau­si­bly negates most of their life’s im­pact, un­less they find a great way to build ca­reer cap­i­tal with­out work­ing for any­one. So with this in­ter­pre­ta­tion, the an­swers would be some­thing close to the NPV of their life’s im­pact di­vided by 3 (ig­nor­ing dis­count­ing).

Also, did you con­trol for will­ing­ness to ac­cept vs pay?

Sorry if this ad­dressed, I skimmed the post.

• To cover more con­tent that’s not new but im­por­tant, you could use a new source on one topic to sum­ma­rize the state of that topic. I like that pa­pers do this in the in­tro­duc­tion and liter­a­ture re­view and I think more posts and the like should do it.

• Google scholar also lists recom­mended new pa­pers on its home­page.

• Why not just pay Rus­sia an (ar­guably fair) repa­ra­tion?

• What sort of de­ci­sion timeline can ap­pli­cants ex­pect? The ex­ist­ing op­por­tu­ni­ties are of­ten slow com­pared to e.g. VC fund­ing which is bad for plan­ning.

• Re 4): Cor­re­la­tion or similar­ity be­tween agents is not re­ally nec­es­sary con­di­tion for co­op­er­a­tion in the open source PD. LaVic­toire et al. (2012) and re­lated pa­pers showed that ‘fair’ agents with com­pletely differ­ent im­ple­men­ta­tions can co­op­er­ate. A fair agent, roughly speak­ing, has to con­form to any struc­ture that im­ple­ments “I’ll co­op­er­ate with you if I can show that you’ll co­op­er­ate with me”. So maybe that’s the mea­sure you’re look­ing for.

A pop­u­la­tion of fair agents is also typ­i­cally a Nash equil­ibrium in such games so you might ex­pect that they some­times do evolve.

Source: LaVic­toire, P., Fallen­stein, B., Yud­kowsky, E., Barasz, M., Chris­ti­ano, P., & Her­reshoff, M. (2014, July). Pro­gram equil­ibrium in the pris­oner’s dilemma via Löb’s the­o­rem. In AAAI Mul­ti­a­gent In­ter­ac­tion with­out Prior Co­or­di­na­tion work­shop.