Macroscope Welcome Party

Saturday  13 January  2018  7:00 PM    Sunday  14 January  2018 4:00 AM
Save (77) Saved (78)
Last update 15/01/2018
  3311

Hello all! We’re welcoming many new people in our collective: Amy Zhang, Hunter Glenn, Sivan Altinakar, Mahault Albarracin, Mamal Amini, Ian Oscar Vasquez Gutierrez, Haydn Thomas-Rose and Meghan Elle.

There will be a lot of people for whom it will be their first event at the Macroscope, and will know very few people. So if you’ve been on the fence about coming to one of our events, this would be a great first event to attend. Also feel free to come with friends / invite other people.

You can start arriving at 19:00.

At 19:30, we’ll play a game to get to know each other. To play the game, please come up with 3 facts about yourself: 2 true statements, and 1 false statement. People will have to try to figure out which one is the lie... probabilistically (more info in the comment section).

The party will start after the icebreaker game, around 21:30. Feel free to arrive just then if you don't feel like playing the game. BYOB.

party
Nearby hotels and apartments
Macroscope
810 Avenue Duluth E, Montreal, QC, Canada

Mati Roy Commento inserito tramite Facebook
-- 23/01/2018 05:10
Here's additional information on credence calibration for those interested:
https://docs.google.com/document/d/1sLacMjdm5umu_yfj0QbcdrrcphFY4vh053kqfPd1B74/edit

Mati Roy Commento inserito tramite Facebook
-- 23/01/2018 05:08
Some people asked me about the scoring method. The scoring is: (LOG(credence,number_of_choices)+1)*LOG(number_of_choices,2) Number of choices was 3 in our case because we had 3 statements per person. The credence is the probability we were assigning to the false statement. It gives the amount of bits of information that was included in our answer starting with a prior probability evenly distributed on the answer choices: credence=2^(score)/number_of_choices Where a "bit" is something like the ability to reduce the answer space by a factor of 2. (Once you reduce your number of choices to 1, you're 100% sure it's the answer. You can't reduce the number of choices further.) The baseline score for this scoring method is always 0. Positive values are good (ie. updating in the right direction), and negative values are bad (ie. updating in the wrong direction). To maximize expected value (EV), one should put values for each answer choice corresponding to one's credence. Informal proof with number_of_choices = 3 a*(log(2,x)+1)/log(2,3)+b*(log(2,y)+1)/log(2,3)+(1-a-b)*(log(2,1-x-y)+1)/log(2,3)) max d/dx: y=(a+b*x-x)/a max d/dy: y=b*(x-1)/(a-1) so: x=a ; y=b
Mati Roy
-- 23/01/2018 05:08
Mahault Albarracin, Sivan Altinakar, Mamal Amini

Macroscope Commento inserito tramite Facebook
-- 22/01/2018 23:18
#RationalistGame (sorry, rereposting with the good picture) Here are the results from the Rationalist 2 Truths 1 Lie game. Positive scores mean you did better than random (that is assigning 1/3 probability to each outcome). Negative scores mean you did worse than random. Don't worry if you have a low score, most people are overconfident the first time they play a calibration game. This spreadsheet has 3 sheets (see at the bottom); one for the credence you gave for everyone, one for the score, and one for the ranking: https://goo.gl/XkGjsG. The results (number in parenthesis represent the number of missing results): 1. David Krueger: 2.93 2. Nasim Rahaman: 2.21 (8) 3. Mamal Amini: 2.19 4. Dany Lyth: 2.18 5. Étienne Fortier-Dubois: 2.09 6. Mati Roy: 1.62 7. Philippe Regnier: 1.36 (1) 8. @Nicolas Gagné: -2.13 9. Jess Lozon: -2.28 10. Sivan Altinakar: -2.57 11. MaJo L'Espérance: -3.07 (4) 12. Fares Benslimane: -3.43 13. Mahault Albarracin: -5.88 (1) 14. Hunter Glenn: -6.56 (1) 15. Amy Zhang: -∞ Congratulation to the winner, David! Other fun facts: -Fares had the less well predicted lie. -Amy had the best predicted lie. -Amy was 100% a statement was a lie, and it wasn't. Never be 100% sure ;) Note: I'm missing Haydn Thomas-Rose's results. @Katiane, you only input your odds for the one you were the most confident was wrong, so I cannot score you properly. If anyone want a recount, let me know soon.

Macroscope Commento inserito tramite Facebook
-- 22/01/2018 22:24
#RationalistGame Here are the results from the Rationalist 2 Truths 1 Lie game. Positive scores mean you did better than random (that is assigning 1/3 probability to each outcome). Negative scores mean you did worse than random. Don't worry if you have a low score, most people are overconfident the first time they play a calibration game. This spreadsheet has 3 sheets (see at the bottom); one for the credence you gave for everyone, one for the score, and one for the ranking: https://goo.gl/XkGjsG. The results (number in parenthesis represent the number of missing results): 1. David Krueger: 2.93 2. Nasim Rahaman: 2.21 (8) 3. Mamal Amini: 2.19 4. Dany Lyth: 2.18 5. Étienne Fortier-Dubois: 2.09 6. Mati Roy: 1.62 7. Philippe Regnier: 1.36 (1) 8. Nicolas Gagné: -2.13 9. Jess Lozon: -2.28 10. Sivan Altinakar: -2.57 11. @MJ: -3.07 (4) 12. Fares Benslimane: -3.43 13. Mahault Albarracin: -5.88 (1) 14. Hunter Glenn: -6.56 (1) 15. Amy Zhang: -∞ Congratulation to the winner, David Krueger! Other fun facts: -Fares had the less well predicted lie. -Amy had the best predicted lie. -Amy was 100% a statement was a lie, and it wasn't. Never be 100% sure ;) Note: I'm missing Haydn Thomas-Rose's results. @Katiane, you only input your odds for the one you were the most confident was wrong, so I cannot score you properly. If anyone want a recount, let me know soon.
https://goo.gl/XkGjsG
Macroscope
-- 22/01/2018 22:24
If someone could tag @MJ and @Katiane, that'd be great because I can't.
-- 22/01/2018 22:24
What would be the score of a uniform player? Just to see where the bar is :-)
Macroscope
-- 22/01/2018 22:24
ah, forgot to put the picture!
-- 22/01/2018 22:24
I believe you are missing Amy's results because she scored -infinity and possibly broke Maths.
-- 22/01/2018 22:24
woah, I kicked ass!
-- 22/01/2018 22:24
We should do this again!

Macroscope Commento inserito tramite Facebook
-- 14/01/2018 00:14
Tegan Maharaj is coming back from New York at 5:30am, so we edited the event to end at 7:00am. What could go wrong :P

Nicolas Lacombe Commento inserito tramite Facebook
-- 13/01/2018 15:13
I will probably not come since I am sick.
Sivan Altinakar
-- 13/01/2018 15:13
Remets-toi bien :-)

Macroscope Commento inserito tramite Facebook
-- 13/01/2018 10:30
Just to help us with planning, please let us know when you'll arrive:
-- 13/01/2018 10:30
Jessica and i are going to make it at 8h

Macroscope Commento inserito tramite Facebook
-- 28/12/2017 22:37
We're going to have disco lights, UV lights and a smoke machine! We can transform the upstairs living room in a dance floor! :)
-- 28/12/2017 22:37
This is for real, right?
-- 28/12/2017 22:37
Dance off, 2018

Mati Roy Commento inserito tramite Facebook
-- 27/12/2017 07:02
Looking forward to the rationalist icebreaker game :)
Mati Roy
-- 27/12/2017 07:02
other memes on epistemic rationality: https://anarchyofproduction.wordpress.com/2014/05/09/epistemic-rationality-memes/
Mati Roy
-- 27/12/2017 07:02
source for the game idea: http://lesswrong.com/lw/k8u/credence_calibration_icebreaker_game/
Nicolas Lacombe
-- 27/12/2017 07:02
how do you define "strong beliefs" (in that context)?

Mati Roy Commento inserito tramite Facebook
-- 27/12/2017 06:58
***Litany of Drunk Tarski*** If the beverage contains alcohol, I desire to believe that the beverage contains alcohol; If the beverage does not contain alcohol, I desire to believe that the beverage does not contain alcohol; Let me not become attached to beliefs I may not want.
https://wiki.lesswrong.com/wiki/Litany_of_Tarski

Macroscope Commento inserito tramite Facebook
-- 27/12/2017 06:39
At 19:30, we’ll play a game to get to know each other. Please come up with 3 facts about yourself: 2 true statements, and 1 false statement. People will have to attribute a probability to each statement being false (the three of which must sum to 100%). Here are calibration games if you want to train 😊: http://acritch.com/credence-game/ http://www.2pih.com/caltest/
Macroscope
-- 27/12/2017 06:39
Understanding the scoring method is not necessary to play the game because the best strategy is simply to give your true probability, but if you're interested, here's a short explenation: Given that the prior probability of any given statement being false is 33.3% (since there are 3 options), updates in the right direction (above 33.3%) will be rewarded with positive points, and updates in the wrong direction (below 33.3%) will be penalized with negative points in proportion to the number of bits of information such an update implied, that is, we will use the formula “'=(LOG(credence,number_of_choice)+1)*LOG(number_of_choice,2)”. Here are example of probabilities and amount of points they would give: https://goo.gl/kb52ax
-- 27/12/2017 06:39
"...the three of which must sum to 100%"? I'm confused. This doesn't math up, right? Why should the 3 statements be dependent on one another? If my statements are : 1- My name is Philippe (~1% prob of being false, accounting for me being on acid enough to forget my name) 2- I am 30 y-o (~1% prob of being false) 3- I think about kittens at least once a day. (Not nearly 98% prob of being false, or at least you have no reason to believe it this strongly) It seems like making the probs add up is just forcing bad epistemology; is there something I missed, or am I on acid?
Macroscope
-- 27/12/2017 06:39
let's do this again sometime!!
Macroscope
810 Avenue Duluth E, Montreal, QC, Canada
Is this your event? Claim it now

Make sure your information is up to date. Plus use our free tools to find new customers.