Mountain Project Logo

What is the grade-averaging algorithm?!

Original Post
John RB · · Boulder, CO · Joined Oct 2016 · Points: 194

I'm wondering how a consensus grade is derived from the suggestions users offer.  For example,

https://www.mountainproject.com/route/105752491/jade-gate

Has 11b/c, 11+, 11+ and 11d as the suggested grades.  The derived grade is... 11b/c.  How does that work?

Or perhaps the grade isn't derived until there are more than a few suggestions and it stays at whatever the original contributor specified?

Leron · · Unknown Hometown · Joined Feb 2016 · Points: 1,141

I think the originator gets a weighted vote.  Meaning they count for 4ish votes.  I also think the +'s don't mix with the letters and we should either use only letters or only + and -'s.  If we are using letters I would prefer  an option for 11d/12a to an 11+ ratting as I have seen many different interpretations of the + and -.

Nick Wilder · · Boulder, CO · Joined Jan 2005 · Points: 4,098

I triggered it to recalculate and it's now 11+.  There's a bug somewhere (that I haven't found yet) that is not causing it to recalculate every time we get a new rating.

John RB · · Boulder, CO · Joined Oct 2016 · Points: 194
Nick Wilder wrote: I triggered it to recalculate and it's now 11+.  There's a bug somewhere (that I haven't found yet) that is not causing it to recalculate every time we get a new rating.

Interesting.  I hope you are able to chase that down eventually (although clearly not that critical since I'm sure new ratings are a rare-ish thing).


Is there a way to see what the original rating was from the contributor?
David K · · The Road, Sometimes Chattan… · Joined Jan 2017 · Points: 434
lech wrote: I also think the +'s don't mix with the letters and we should either use only letters or only + and -'s.

Totally agreed on this. I'd strongly prefer letters to the +/-. It seems like + just means "sandbag", and - just means the grade is soft. And I don't mean by a partial grade, I mean 5.6+ could easily be 5.8, and 5.6- could easily be 5.4. +/- grades just don't fit into a spectrum of difficulty well.

John RB · · Boulder, CO · Joined Oct 2016 · Points: 194

Currently 5.11's can be

5.11a, 5.11-, 5.11a/b, 5.11b, 5.11, 5.11b/c, 5.11c, 5.11+, 5.11c/d, 5.11d

I have no idea what internal values MP uses to sort these in the way above nor what values are used internally to distinguish.  It does seem a bit of a mess, but then grades are so subjective that slicing thinly hardly matters anyway.

Derek DeBruin · · Unknown Hometown · Joined Jul 2010 · Points: 1,094

Pretty sure the algorithm for 5.11 is more similar to Brutus of Wyde's definition:

"I've seen 5.11 divided into 11 different grades of increasing difficulty, as follows: 5.11a, 5.10d, 5.11-, 5.11b, 5.11, 5.11c, 5.9 squeeze, 5.11+, 5.10 OW, 5.12a, 5.11d"

Creed Archibald · · Salt Lake City, UT · Joined Apr 2012 · Points: 1,026

+1 for either +/- or letters but not both. Personally I like +/-.

tenesmus · · Unknown Hometown · Joined Jan 2004 · Points: 3,074

I'm a bit worried about this issue. This is kind of changing as I age, but I'm not normally known for rating routes soft. The algorithm is letting the people who actually rate things a bigger voice than those of us who don't bother. Seriously, what is the percentage of users who do or don't grade things? How much of that is ego? "Liking" a climb, ticking it, etc is awesome and theoretically something you do for yourself. Letting the algorithm change the grades could put someone in serious jeopardy.

For instance, two of the routes in one of my newer areas have starts with blind holds that are hard to see. This makes the start of a .10a as hard as .11a if you don't find the right sequence. You could bust your ass quickly if you blow it and I would really love to keep the grade as originally posted for the onsight. I wrote it up that way but fear this system is sketchy.

I seriously don't give a crap if some dude thinks it's only .9+. Of course it's easier once you have it wired. People have castigated me a little over this issue and I don't want to take the blame for the system changing things. The skill vs confidence ratio for climbers at this grade is highly variable, especially when compared to their strength if they boulder or climb at a gym.

I feel like this could get out of hand fast and would love to see the grade as posted by the original person listed and a consensus Mt Project grade beside it.

Finally, the +/- grade thing is always going to be sticky. That debate isn't going to go away. Letting the algorithm interpret it one way doesn't give justice to other views. For instance, the Old-Schoolers I knew would rate something as 12- and mean "anything on the lower half of 5.12." New-schoolers might say .12- and mean "easier than .12a", which could also create a ton of friction and confusion.

Please fix this before the algorithm gets someone in over their head in a bad way.

Chalk in the Wind · · Unknown Hometown · Joined Aug 2014 · Points: 3
tenesmus wrote: I'm a bit worried about this issue. This is kind of changing as I age, but I'm not normally known for rating routes soft. The algorithm is letting the people who actually rate things a bigger voice than those of us who don't bother. Seriously, what is the percentage of users who do or don't grade things? How much of that is ego? "Liking" a climb, ticking it, etc is awesome and theoretically something you do for yourself. Letting the algorithm change the grades could put someone in serious jeopardy.

For instance, two of the routes in one of my newer areas have starts with blind holds that are hard to see. This makes the start of a .10a as hard as .11a if you don't find the right sequence. You could bust your ass quickly if you blow it and I would really love to keep the grade as originally posted for the onsight. I wrote it up that way but fear this system is sketchy.

I seriously don't give a crap if some dude thinks it's only .9+. Of course it's easier once you have it wired. People have castigated me a little over this issue and I don't want to take the blame for the system changing things. The skill vs confidence ratio for climbers at this grade is highly variable, especially when compared to their strength if they boulder or climb at a gym.

I feel like this could get out of hand fast and would love to see the grade as posted by the original person listed and a consensus Mt Project grade beside it.

Please fix this before the algorithm gets someone in over their head in a bad way.

Yes. Thank you. I have the same concerns and have thought about posting this as well.

I have already seen some of these problems.

Example One: a short 2-bolt route with decking potential that originally appeared here as a 5.8+ and now is a 5.6 because of a single user rating. One could debate the +, but it is not 5.6 and is noticeably harder than a 5.7 right beside it. Many new climbers start around 5.6. A new climber using the MP search tool to list all local 5.6 routes and then going out and climbing them could get in real trouble on this route.

Example Two: a TR route graded 5.12 a/b that is now 5.9+ due to user ratings. Since this is a TR route, there is not really a safety issue here, but it is still annoying and indicative of the problems this system can create. The route is an "eliminate" one, and I can tell from comments that people downgrading the route are using "off" holds.

Most people submitting pages are probably using the local guidebooks or word of mouth for the grades. If people feel a listed grade is wrong, there are multiple ways to address it: the comments section, suggested ratings (that don't affect the published grade itself), discussion with the author or an admin.
John RB · · Boulder, CO · Joined Oct 2016 · Points: 194

The bottom line: if you are a climber who puts your life at risk based on something you read in a guidebook or online (including MP), you are being very reckless and should stop leading until you reevaluate how you approach climbing.

"Hey, Bastille Crack is rated 5.7 on MP, and I TRed a 5.7 in the gym, so I'm ready to lead Bastille Crack!"

This kind of thinking has broken many ankles (and worse).

I love that MP lets consensus grades be set by people who've done the route.  Unless there are deliberate trolls (like on Dawn Wall ratings), it works really well.

M Mobley · · Bar Harbor, ME · Joined Mar 2006 · Points: 911

John RB for the win. Learn to read the climb or get really good at TRing.

Petsfed 00 · · Snohomish, WA · Joined Mar 2002 · Points: 989
David Kerkeslager wrote:

Totally agreed on this. I'd strongly prefer letters to the +/-. It seems like + just means "sandbag", and - just means the grade is soft. And I don't mean by a partial grade, I mean 5.6+ could easily be 5.8, and 5.6- could easily be 5.4. +/- grades just don't fit into a spectrum of difficulty well.

The letter system only makes sense if the route is sufficiently sequential or varies little over body size. Otherwise, the variability in grade becomes too big, so the +/- system captures it a little better. Put another way, how well does your hand fit in a #1 camalot sized crack? A pitch of that is hard 10 for me. For my girlfriend, it’s almost 5.8. Would that be 5.10a?

To me, 11- sounds like 11a/b, 11 sounds like 11b/c, and 11+ sounds like 11c/d.
M Mobley · · Bar Harbor, ME · Joined Mar 2006 · Points: 911

get rid of the +, - and 11. stick to a,b,c and d

Leron · · Unknown Hometown · Joined Feb 2016 · Points: 1,141

I would say trad grades are +- and for sport below 5.10 is to.  For sport let's use letters. Also if we can split 12a/b we should be able to split 12d/13a. I am fine with removing the split entirely to. The consensus grades should end up with a good spread between the grades anyway. Unless people let their ego effect their grading. 

tenesmus · · Unknown Hometown · Joined Jan 2004 · Points: 3,074

Ok, so I'm wondering how to best deal with this. Trying hard to laugh at myself and separate the ego of, "I rated it this way for a specific reason" from, "They downgraded my climb."

I posted a sport route FA. In the description, I let people know the route is graded differently, depending on how you climb it.
"If you want to feel good about yourself, climb right of the bolt line at the bottom and onsite it (but its really only 11+). If you stay in the bolt line at the bottom, the pump becomes a little more of a factor and makes it .12a. Also, the climbing is way more fun that way."

Now the route says it's 11, which is true, if you climb it that way. One person rated it 5.11 in their personal ratings, so now the whole thing is 5.11? I'm kind of loathing the fact that this algorithm is changing grades so easily. I tried to re-edit it but the system won't let me. I notice the other climbs at this crag also have a similar pattern. One or two people say it's easier or harder and the whole thing changes.

Am I the only one who thins it's kind of odd that we can do the FA, post it and not be able to re-edit this way? We need a way to show the FA's grade vs consensus grade. OR to allow either the original poster or admin to lock it in.

I know I used this example above but it didn't go away. We graded one of these climbs at middle .10 by several people who climb that grade. We rated it that way because the start is inobvious from below and you could really hurt yourself if you get it wrong. All it took was one person calling it .9 and the grade changed to 9+. A beginning climber might tell you there is a huge gap between .10c and .9+.

This is a sketchy way to do things and it's going to get someone hurt. I don't want my name attached to this kind of thing. What happens when someone goofs off and starts down or up grading climbs as a joke? What happens when someone gets sand-bagged into getting hurt?

Brian in SLC · · Sandy, UT · Joined Oct 2003 · Points: 22,793

I've always like the consensus grading here on the 'proj.  I really don't have much of an issue if folks want to change the rating of a route I've entered in the database...with the exception being that if I thought the route was PG or R (or even X rated), then, to me that was an important part of the rating but could also be explained in the description.

So...ten...you're bummed cause your 5.11+ got downrated to 5.11?  And you think someone might get hurt on it?  Put in another bolt, man!  Its a sport route!  But, just one vote tanked the rating?  That seems off.

Does sound like the algorithm has got a glitch.  Every new person adding a new rating shouldn't influence the rating that much.

Maybe get a few of your homies to add their two cents and bump the rating to where it should be?  

Ratings tend to be all over the map.  Tall folks, savvy folks, different styles, strengths, weaknesses...all make it difficult to accurately rate a route.  Which is why on popular routes, having 30-50 people add their opinion, some with many years and mucho mileage, makes the consensus rating a winner in my book.

I think its a very useful function of this site.  And...I peruse the ratings especially by folks I know.  Helps keep me grounded in reality (ha ha).

But, I really don't get that worked up over it.  Bottom line, if you think folks will get hurt on your route because of the rating, and, its a sport route, then, add a bolt.  Or change the bolt spread.

Folks get in over their head for a number of reasons.  Its a learning process.  And, some people learn the hard way.  If a lower consensus grade lures in the unsuspecting victim...well...caveat emptor.

What would be useful, is, to see the original rating maybe on the stat's page.  Because if tenesmus  rates a route as hard for me...and someone downgrades it...I'd still want to know the original rating unless I wanted to get my arse kicked.

John RB · · Boulder, CO · Joined Oct 2016 · Points: 194

Can't you still edit the description? You can write anything you want in the description, including your feelings about the grade that will never be affected by whatever consensus grade it lands at.  You can warn people about risk if there is a runout.  I always read route descriptions before choosing to go do a route on MP.  

Andy Eiter · · Madison, WI · Joined Jul 2014 · Points: 276

What if the grade remained locked at the FA grade until a certain number of grades is submitted?

Chalk in the Wind · · Unknown Hometown · Joined Aug 2014 · Points: 3

I really don't see why a page can't have an FA/guidebook grade that only the submitter or an admin can change and a consensus grade. I'm glad I'm not the only one who sees problems with the current system.

The solution doesn't have to be either/or.

Petsfed 00 · · Snohomish, WA · Joined Mar 2002 · Points: 989
Robert Michael wrote: I really don't see why a page can't have an FA/guidebook grade that only the submitter or an admin can change and a consensus grade. I'm glad I'm not the only one who sees problems with the current system.

The solution doesn't have to be either/or.

I've always been rather critical of the fact that you can't submit a route with a historical or "guidebook" grade and still voice your opinion on the "true" grade of the route.

Guideline #1: Don't be a jerk.

Help
Post a Reply to "What is the grade-averaging algorithm?!"

Log In to Reply
Welcome

Join the Community! It's FREE

Already have an account? Login to close this notice.