Page 8 of 10
Posted: Wed Apr 15, 2009 1:05 pm
by twan
pkananen wrote:I think Danny makes sense. Although the difficulty is linear, the distribution of climbers' ability would be close to a bell curve. Which is why going from 15a to 15b may be the same net difference in terms of theoretical absolute difficulty compared to 10a to 10b, but making the leap to 15b will require much more training (and/or good genetics), hence why fewer people can do it, and why some would argue the linear point distribution doesn't make sense.
But this is all kind of silly, anyway. Let's just go climb rocks.
Im going to agree with that last part.. What started out as me joking with BCombs about catching him has turned into a conversation that I am now lost on. My brain now hurts... Joking about points is fun and jokes are jokes and joking is a good thing for jokes when people are joking about jokes that joke... no joke... fo real....
Posted: Wed Apr 15, 2009 1:15 pm
by pigsteak
is there such a grade as 7b+ ? maybe it would fit in there.
Posted: Wed Apr 15, 2009 1:37 pm
by bcombs
So when you do the conversions it looks pretty much the same as what you have... just higher point values. There are weird gaps (as you noted), but it's pretty much an even scale. Two 12a's in a day is still more points than a 5.14.
8b+ 5.14a
1150 1150
8b 5.13d
1100 1100
8a+ 5.13c
1050 1050
8a 5.13b
1000 1000
7c+ 5.13a
950 950
7c 5.12+
900 900
7b 5.12-
800 800
7a 5.11+
700 700
6b+ 5.10+
550 550
6a 5.10-
400 400
5a 5.7
250 250
Posted: Wed Apr 15, 2009 1:43 pm
by Barnacle Ben
It's much harder to go from running a 5 minute mile to a 4 minute mile than it is to go from running a 8 minute mile to a 7 minute mile. But it's still just 60 seconds.
Posted: Wed Apr 15, 2009 2:00 pm
by SCIN
Well now I'm thinking for sure you guys are right and we should consider making the points exponential. I'm really surprised that 8a.nu hasn't done it since they're the "leader" in the entire lameness of the points system.
By the way, twan, we all know the points are gay, retarded, and all other politically incorrect words but it's just another stupid internet toy to get us through the work day. I know plenty of climbers who don't keep tick lists that could post their hit list up here and obliterate all of us with about 100 routes. Also, by no means do points signify the good climbers. They signify the old climbers with no children or no other hobbies.
Posted: Wed Apr 15, 2009 2:06 pm
by ynot
Danny. Don't you have atoms to smash or something important like that?
Posted: Wed Apr 15, 2009 9:25 pm
by Danny
SCIN wrote:Well now I'm thinking for sure you guys are right and we should consider making the points exponential.
One way you might want to scale the points would be to use empirical data from your online database. You could, for example, look at the number of sends of a particular grade and scale them by the total number of routes at that grade. The reason for the scaling is that if there are more routes at a particular grade there is a higher chance they will get climbed. There are many other considerations but this is the simplest thing to do. The plot below shows the total number of sends (green line right axis) for routes between 5.5 and 5.14. Grades with less then 30 total routes in the red were excluded. A grade like 5.8 gets 8.5. for -abcd+ add .1 .2 .4 .6 .8 .9 to the grade. The blue line shows the total number of routes at that grade. The red line shows the ratio. The purple line is a linear fit to the red line from 5.10 to 5.13. The linear fit is surprisingly good over this range. Based on this the point system should be linear. I know there are problems with this. If anyone has better ideas I can try and modify the measure.
Posted: Wed Apr 15, 2009 9:46 pm
by pkananen
why did you take out grades with less than 30 routes? That would provide data on both sides of the bell curve that I think should exist. It also would provide the most evidence for giving points on a curving scale, as that would include many hard routes that see few ascents, and are thus supposedly worth more points.
Posted: Wed Apr 15, 2009 10:08 pm
by Danny
I removed grades that have few routes because the small sample size creates statistical noise and makes the plot more messy than it already is. I could still include them in the calculation of a curve fit and weight them by their standard error. There may be some non linear behaviour beyond 5.13 but the linearity between 5.10 and 5.13 seems interesting.
I just looked at it for the higher grades and it seems to stay on the linear line until 13d. Then 14a is larger than you might expect, more sends per number of 14a's. Here is the raw data:
Code: Select all
-1 4th 11.600000 58 5
1.5 5.1 13.000000 169 13
2.5 5.2 14.833333 356 24
3.5 5.3 23.842105 453 19
4.5 5.4 25.545455 843 33
5.5 5.5 20.379310 591 29
6.5 5.6 35.098361 2141 61
7.5 5.7 45.539683 5738 126
8.1 5.8- 50.500000 1111 22
8.5 5.8 32.939850 4381 133
8.9 5.8+ 26.543478 1221 46
9.1 5.9- 51.750000 1449 28
9.5 5.9 33.897810 4644 137
9.9 5.9+ 51.642857 3615 70
10.2 5.10a 62.245283 6598 106
10.4 5.10b 40.564815 4381 108
10.5 5.10 47.750000 382 8
10.6 5.10c 42.500000 3570 84
10.8 5.10d 51.428571 3960 77
10.9 5.10+ 9.200000 46 5
11.1 5.11- 5.000000 10 2
11.2 5.11a 38.192771 3170 83
11.4 5.11b 41.543689 4279 103
11.5 5.11 2.666667 16 6
11.6 5.11c 28.445946 2105 74
11.8 5.11d 27.310811 2021 74
11.9 5.11+ 5.555556 50 9
12.2 5.12a 24.509259 2647 108
12.4 5.12b 26.809524 1689 63
12.5 5.12 0.500000 6 12
12.6 5.12c 12.468750 798 64
12.8 5.12d 16.935484 525 31
12.9 5.12+ 1.000000 7 7
13.2 5.13a 8.793103 255 29
13.4 5.13b 6.068966 176 29
13.6 5.13c 3.714286 26 7
13.8 5.13d 1.400000 7 5
14.2 5.14a 0.857143 6 7
21 A1 2.000000 2 1
22 A2 3.000000 12 4
24 A4 2.000000 2 1
30 V0 22.600000 113 5
31 V1 3.333333 10 3
32 V2 7.750000 31 4
33 V3 17.000000 85 5
34 V4 14.000000 28 2
35 V5 7.000000 7 1
36 V6 2.000000 2 1
38 V8 5.000000 5 1
Posted: Wed Apr 15, 2009 11:11 pm
by woman
if you ask me...this thread is HIGHJACKED.