How do we determine "Classic"?

Marc801 wrote: Oh, I agree with all that. It's when we get to mercilessly parsing something as intrinsically subjective and amorphous as "classic" that we start having arguments bordering on the absurd. Seeing that there are "classics" I have done and don't think are classic at all, and others that aren't considered classic that I feel are; others that I haven't done that are also unquestionably classic; and yet still others that I haven't done that are called classic but really aren't........ all of which means I can't even agree with myself on what is and is not a classic, let alone invest time in a discussion about it. BTW, High Exposure in the Gunks is the most overrated "classic" on the planet. ;-)

I feel ya' and the biggest trouble for me is that a "classic" (ie 4 stars) on some local mossy choss pebble is barely above "bomb" in a real climbing area.

Fundamentally parochial bias should be out of the equation, if there is to be any objective way of identifying "classics".

I had this argument with a Josh Fan Boi giving Illusion Dweller "classic" status when it would be the biggest POS in the entire Needles of CA. After 40 years that climb has barely cleaned up to "good" and still has grit and loose flakes. It is arguably a classic for Josh, home to choss crack climbs, but that is about it.

Classic: judged over a period of time to be of the highest quality and outstanding of its kind.

Everyone will have a different option on what is the best that is why you take lots of people's options and come to a consensus. It will never be perfect though just like if you climb all over the country you won't agree with alot of grades in different areas.

Does anyone really know how they calculate it on this site? The trump tower one being on top is the pure trolling one. They used to have another not real climb one about a guy getting stuck on a wall before that one.

Question is do you rank a route that has 10 max stars higher than one with 1000 ratings of 3 stars? Do you go by how many people tick it off etc. It isn't like you can prove anyone really climbed anything or if they are just trolling etc. I have plenty of routes on this site that are there just as a pure joke.

I think it's an algorithmic problem, that can have an algorithmic solution. The problems, as I see them, are that number of ticks has a greater impact than it should and that lower-graded climbs, difficulty wise, are rated as far more 'classic' than harder climbs simply by virtue of having more people climb the grade. I'm not sure what the algorithm is right now, but this is roughly how I'd go about fixing it:

Number of ticks should definitely be a factor. A low-tick route is likely to be a local classic, but probably doesn't have the type of exposure required to be considered a world classic. In addition, it makes for a good tie break for climbs with similar star ratings. A 3.8 star climb with 1000 ticks is probably more deserving of its 'classic' rating than a 3.8 star climb with only 100 ticks, all else being similar. I'm not sure how ticks are currently included in classic determination, but my guess is that using a logarithmic term would reduce the impact of having lots of ticks over lots of quality.

For normalizing by difficulty, one approach would be to divide the number of ticks by the number of ticks for that grade as a whole, or by some number proportional to that. For example, if 100,000 5.6's have been ticked, and only 50,000 5.10s have been ticked, the difficulty bias could be eliminated by dividing the tick counts for all 5.6s by 2. Of course, this might fun into problems at the lower grades, since there's probably far fewer 5.4 climbs on the site than 5.8 climbs. A better solution might be to create a bias term using the number of people who have logged a tick of that grade or harder. This would account for differences in number of routes at a grade, while still fixing the bias towards climbs at lower grades getting far mroe ticks due to their ease, as opposed to their classic status.

Lastly, my guess is that minor quality differences get drowned out at the upper end, since everything that's even a contender for 'classic' status gets between 3.5 and 4 stars. Subtracting 3 from the star rating on everything would give differences between these grades more significance in an algorithm for computing classics based on MP data.

A final formula might look something like this:
C = (stars - 3) * log(ticks/(tick bias))

That or give users an option to filter classics by difficulty range. If you want 5.1 to 5.8 or want 5.10 to 5.13 you could set it and the list would show what you want. Would be ideal to have it as a user setting to default to a range that works for you.

ViperScale wrote:That or give users an option to filter classics by difficulty range. If you want 5.1 to 5.8 or want 5.10 to 5.13 you could set it and the list would show what you want. Would be ideal to have it as a user setting to default to a range that works for you.

I think this, along with differentiating based on the type of climbing (boulder, sport, trad, ice, alpine etc.). I mean, all the routes on the top 10 a roped routes. I'd say that a few of these routes are less classic than midnight lightning, for example.

anybody notice that they've changed the list? Now it is the top twenty and they've made some changes to the order and removed the trump tower. Also there are only 19 listed even though it says 20

top 20 list

Aslo if you click on "Find more classics near you!" it takes you to a page where you can sort by route type and location

cool new feature

That's not a new feature. I remember that as long as I've known of mountain project. I wish when you selected "Trad" that it wouldn't include aid routes. An additional aid button would be cool.

I think that there is no point in having the top 20 classics listed on the home page if the algorithm isn't optimized in some way. Thanks to a heated discussion in the comments, a recently climbed chossy tower in the Utah desert that has 4 ticks, is now the most classic climb in the country.

mountainproject.com/v/angel…

Just ignore the "classic" list and go about your business!