ascent logging grades when route does not have a grade

[scd]

How should we handle the rather long list of grades when route does not have a grade in the ascent logging process.

Ideas:

• similar to #color in the route description, we could use #ewbanks
• try and work this out by using the grade context and gear style of route- This will not be definitive.

I like the #ewbanks, because it is only really a concern for gyms and for routes without a grade.

[brendanheywood]

Does an ascent need a grade? In the context of gyms, and quite possibly more widely I think the full list of grades is overkill and we can filter it based on the gear type and context. Additionally in gyms, we could have a flag on the gym node, which when set forces people to choose an ascent grade (assuming they aren't required to start with), and then after a few ascents the consensus grade emerges.

[scd]

Technically an ascent can be logged without a grade, but then stats on level of difficulty will be missed.

I like the idea of unrated routes getting a rating from the first couple of ascents in a gym. Although, personally I think that only a small number of climbers will feel confident enough to rate the grade. I recon that almost everybody would be able to say whether they thought it was hard, average or easy. If we had that information for plus knowing what level they climb at, we could have a pretty good shot at getting a reasonable grade.

[brendanheywood]

hard, average or easy

I've had so many discussions with so many climbers over the years about exactly this, and going down this way is where madness lies IMHO. You just end up inventing yet another grade system, except it only applies to one person, and then have to map it back to an existing grade system, and you lose accuracy at both steps.

The only two realistic ways I can see a consensus grade system working is people either give it a subjective grade directly, ie 'It is a 22 (or unknown) but it felt to me like a 21' or use a relative ranking system like Elo in chess 'This route felt harder than X, but easier than Y' and then anchoring the resulting consensus list of ranked routes against known problems at particular grades, eg we could assume that the routes with the most ticks have the strongest consensus grade, and the less ticked ones get regraded up or down relative to these. This is how I've graded stuff in the guides I've written and I think a similar process is pretty common with other coordinators.

I've mused for a while over the idea that we could have a game sorta like 'whoishotter.com' that people can play with their local crag which then generates a list of potential grade corrections for that crag. We could even build this into the ticking page, it picks at random a route at the same gym or crag, which the user has done before, and is of a roughly similar grade and asks them simply which one is harder or are they the same.

Another random idea I just had, is that if we keep going down the 'it felt like...' method then even as the first couple of tentative grades are added, we could use them right away, and hone them over time:

So person A grades X as 22 -> shows up as 20-24? So person B grades X as 21 -> shows up as 20-23? So person C grades X as 22 -> shows up as 21-23? So person D grades X as 24 -> shows up as 22-23? So person E grades X as 22 -> shows up as 22

So after a number of ticks, if the stdv of the tick grades falls within a single then we lock the mean / mode in as the real grade.

[scd]

Without doing some experimentation I am not so sure which way would be better at establishing a grade. I don't agree with your madness hypothesis :)

If a grade 16 climber says it is easy then it is probably less then a grade 14. Later if a grade 12 climber says it is hard then it is probably more than a grade 14. Apply some histogram function and you come up with a grade 14.

I think just asking people what grade it is would be better if people were comfortable with assigning a grade. To me this is the big unknown. If only 1 in 10 gym climbers are comfortable assigning a grade then it will be a while before we can put any grade on the route. We are not the average gym climbers, so it is going to be impossible for us to judge this without either doing a trial, or asking some gym climbers.

Also I think that people are really good a discriminating route grades around their level, but as you get away from their level they become inaccurate. Crags put up by hard climbers often have many very difficult for the grade routes at the lower grades.

I like the idea of asking if it is easier or harder then another climb they did. But I think this is a better technique for fine tuning grades over the long term.

[brendanheywood]

If you want to experiment with some histogram based method you probably don't need to ask them anything, you could just base it of whether they onsighted / flashed == easy, and down the spectrum to 10 attempts or dog == hard. If someone onsights it then that is probably more meaningful data to an algorithm than if they ticked it after 3 attempts but then still called it easy. You would need to account for a users strength / technique improving and waning over time, and this varies quite a lot. I know if I climb 3 times a week for 2 weeks I easily improve a grade or 2, and then it will fade slowly.

Agree that getting a good feedback loop with real gym users is the best way

[brendanheywood]

This thread got a bit distracted but the original issue was around grade input which would be solved by grade widget #1180.

[Mdemaillard]

closely linked to #1400

[scd]

This is resolved in the new tick flow release