You are not logged in.

falcotron

Professional

  • "falcotron" started this thread

Posts: 786

Location: San Francisco

Occupation: software developer/musician

  • Send private message

1

Thursday, September 17th 2009, 1:32am

Birthdays

I've moved this from the Suggestion thread Modify the Opponent Search when attacking to exclude Horde Members & Allies because I didn't want to keep derailing that thread...
I did have a hunch my answer might be wrong about the birthdays, though I couldnt work out how - people only ever pose those sort of questions if baiting for the wrong answer
Well, yes, especially since the point I was trying to make is that people have terrible intuitions about probability and statistics, and common sense is a very bad guide.
all very interesting - if based on the fact that birthdays tend to be clumped in certain months is that due to randomness, or is it due to a human biological thing
No, no, it's not based on that. That clumping increases the chances from 50.7% to... I think it was a bit under 52%. But the 50.7% is there even with a purely random distribution. (Llook up "birthday attack," a crypto attack on digital signatures that takes advantage of exactly this fact.)

This fact about random distribution is why it always looks like some people are unfair lucky or unlucky. It's also part of the reason coincidences seem to be more common than they should. (The other part is selection bias--you notice coincidences, you don't notice things that happened exactly as expected.) The human brain assumes randomly-distributed means evenly-distributed, and this is nowhere near the truth.

But the clumping of birthdays is an interesting question anyway. And, at least when I was in school, it was an unsolved problem. Actually, there were a lot of published theories that explained why birthdays tend to clump in April and May--but the problem with these theories is that once someone actually did the research, May turned out to be the least common month for birthdays, and April the third least, while August and October the most common. (At least in the US.)

Even among professional sociologists, common sense loses out to the facts....

(Birthdays also tend to clump on weekdays, and slightly more on Friday, Monday, and Tuesday. But this one has been explained. In the US, something like 1 in 5 births are a result of induced labor and C-sections. Hospitals try to schedule induced labor and C-sections on weekdays, when they have more staff and fewer patients.)
Top accounts for sale: TG1 (or the gear) and AI1
Already sold: HW1/2/3/4 TG1DE

2

Tuesday, September 22nd 2009, 4:59pm

Well yeah Falcotron,

I reject your reality and substitute my own. If you think that over explaining statistical anomalies makes you better qualified to comment on the 'randomness' of this game go ahead. Pls share the code with us all. I'm not whining about rares, glowies etc, but I've learnt from experience that when I use the random search I keep getting the same person time after time. I don't have the time to record results and conduct power analysis. Pls don't reply with an essay, just saying that random search is definitely not random.

3

Tuesday, September 22nd 2009, 5:11pm


...
I reject your reality and substitute my own.
...

Mythbusters rules.


...
just saying that random search is definitely not random.
In order to have "definite" in that statement you'd need to back it up with facts. Meaning you'd have to spend the time to take statistics, then experiment, then analyze. Falcotron did that.

4

Tuesday, September 22nd 2009, 5:16pm

falcotron said he did that. I generally trust peer reviewed journals (depending on impact factor). Do we even need to go there Virgo? I'm convinced that the random element isn't random. :) (regarding the search factor)

5

Tuesday, September 22nd 2009, 5:26pm

sorry Falcotron & Virgo,

that comment was ridiculous. I do however maintain my stance on the TAG random issue :)

falcotron

Professional

  • "falcotron" started this thread

Posts: 786

Location: San Francisco

Occupation: software developer/musician

  • Send private message

6

Wednesday, September 23rd 2009, 12:03am

falcotron said he did that. I generally trust peer reviewed journals (depending on impact factor). Do we even need to go there Virgo? I'm convinced that the random element isn't random. :) (regarding the search factor)
I will submit my results to the Journal of Patently Obvious Results That Need No Proof and see what the reviewers say.

I'm also writing a followup article for them which proves that people who don't understand how randomness works tend to not understand how randomness works.
Top accounts for sale: TG1 (or the gear) and AI1
Already sold: HW1/2/3/4 TG1DE

7

Wednesday, September 23rd 2009, 1:49am

I'm also writing a followup article for them which proves that people who don't understand how randomness works tend to not understand how randomness works.
lol, I see, and I'm guessing that the only one who understands the concept is......you? :sleeping:

falcotron

Professional

  • "falcotron" started this thread

Posts: 786

Location: San Francisco

Occupation: software developer/musician

  • Send private message

8

Wednesday, September 23rd 2009, 2:43am

I'm also writing a followup article for them which proves that people who don't understand how randomness works tend to not understand how randomness works.
lol, I see, and I'm guessing that the only one who understands the concept is......you? :sleeping:
No, there are millions of people with a rudimentary grasp of elementary probability, statistics, and/or information theory, and all of them get it.

It should have been clear from the name of the journal that these things are not particularly difficult for people who aren't obstinately proud of their lack of knowledge. But apparently, it wasn't.

My next paper will attempt to prove that patently obvious results that need no proof are only obscure or difficult to understand for people who are willfully ignorant.
Top accounts for sale: TG1 (or the gear) and AI1
Already sold: HW1/2/3/4 TG1DE

LaidBeatle

Intermediate

Posts: 238

Location: Berlin

Occupation: Bartender

  • Send private message

9

Wednesday, September 23rd 2009, 10:02am

My next paper will attempt to prove that patently obvious results that need no proof are only obscure or difficult to understand for people who are willfully ignorant.


Hey! I subscribe to that! Makes good bathroom reading material...

falcotron

Professional

  • "falcotron" started this thread

Posts: 786

Location: San Francisco

Occupation: software developer/musician

  • Send private message

10

Wednesday, September 23rd 2009, 1:43pm

My next paper will attempt to prove that patently obvious results that need no proof are only obscure or difficult to understand for people who are willfully ignorant.


Hey! I subscribe to that! Makes good bathroom reading material...
And it's printed on such soft paper, too! I used it for months in the bathroom before I realized anything was printed on it....
Top accounts for sale: TG1 (or the gear) and AI1
Already sold: HW1/2/3/4 TG1DE

11

Thursday, September 24th 2009, 11:08am

No, there are millions of people with a rudimentary grasp of elementary probability, statistics, and/or information theory, and all of them get it.

It should have been clear from the name of the journal that these things are not particularly difficult for people who aren't obstinately proud of their lack of knowledge. But apparently, it wasn't.

My next paper will attempt to prove that patently obvious results that need no proof are only obscure or difficult to understand for people who are willfully ignorant.
hehe Falco, seems like another of your threads has degenerated into a quasi-intellectual flame fest. I'm sure you're also a leading authority on pattern recognition analysis, notice any patterns? For a purported advocate of this forum being a place to "exchange ideas" seems like you're doing a fair amount of brow beating and not much exchanging ;)

falcotron

Professional

  • "falcotron" started this thread

Posts: 786

Location: San Francisco

Occupation: software developer/musician

  • Send private message

12

Thursday, September 24th 2009, 10:06pm

notice any patterns?
Yes, I've noticed that you like to come into threads and derail them just to get another chance to show off your own ignorance, and your pride in it. And, since I'm a nice guy, I will continue to give you those chances.
Top accounts for sale: TG1 (or the gear) and AI1
Already sold: HW1/2/3/4 TG1DE

Lady K

Professional

Posts: 777

Location: Stockport, UK

Occupation: System Developer

  • Send private message

13

Wednesday, September 30th 2009, 5:41pm

I think that this thread, as it appears to have become very personal, should be closed.

Take it to Pm's