FP %age on a cache page

[+Deepdiggingmole]

I noted a cache that I have which up until recently had 100% (FP / finders) and though not a great number of finders this was a nice stat to have - namely ALL who have found it have given an FP. 
However - today I noted that it has dropped to 94% - on checking I spotted that there were still X number of finders which matched the no. of FPs given 
I couldn't understand why and then I spotted the possible reason. One of the cachers is now only a basic member 

I believe I understand how the algorithm works in that it doesn't include any lapsed membership count thus if a premium member is now a basic this will affect the %age. 
My cache is not PMO and so will at some point get found by basic members but up till now all finders have been premium hence the 100% - though if I am right only finds by PM will count in the FP %age figure anyway
The reason the one member who was premium and is now basic is because they died and their membership lapsed only when the next renewal was due and it didn't occur. 

However I find it slightly erring that that FP is now ignored despite the fact that it was given when that member was premium 
Additionally it would seem that despite their FP now being discounted their find isnt ?? - surely if you dont count their FP in the algorithm, you don't count their find either (in the algorithm) 
That aside I do wonder why the decision to discount FPs for members who changed from premium to basic - at the end of the day they were premium when they awarded the FP (and would have had to have been as they wouldn't be able to award an FP) so why is the fact that they changed membership type is it then penalised in the %age figure. I appreciate the overall FP figure doesn't change but the %age figure is a nice stat to look at as a CO and through no fault of theirs. 

Edited February 26 by Deepdiggingmole

[+TeamRabbitRun]

23 minutes ago, Deepdiggingmole said:

I noted a cache that I have which up until recently had 100% (FP / finders) and though not a great number of finders this was a nice stat to have - namely ALL who have found it have given an FP. 
However - today I noted that it has dropped to 94% - on checking I spotted that there were still X number of finders which matched the no. of FPs given 
I couldn't understand why and then I spotted the reason. One of the cachers is now only a basic member 

Now before all the 'this is how it works' replies come in - I do understand how the algorithm works - it doesn't include any lapsed membership count thus if a premium member is now a basic this will affect the %age. 
My cache is not PMO and so will at some point get found by basic members but up till now all finders have been premium hence the 100% 
The reason the one member who was premium and is now basic is because they died and so their membership lapsed only when the next renewal was due and it didn't occur. 

However I find it slightly erring that that FP is now ignored despite the fact that it was given when that member was premium 
Additionally it would seem that despite their FP now being discounted their find isnt ?? - surely if you dont count their FP in the algorithm, you don't count their find either (in the algorithm) 
That aside I do wonder why the decision to discount FPs for members who changed from premium to basic - at the end of the day they were premium when they awarded the FP (and would have had to have been as they wouldn't be able to award an FP) so why is the fact that they changed membership type is it then penalised in the %age figure. I appreciate the overall FP figure doesn't change but the %age figure is a nice stat to look at as a CO and through no fault of theirs. 

 

Wait, shouldn't it work the other way?

 

If 10 PMs find a cache and each awards a FP, then the calculation is '10/10', or 'FPs / PM Finders', which would be '1', or '100%'.

 

If one of those PMs drops to Basic Member, then the calculation becomes '10/9' which is roughly 1.1, or '110%', not DOWN into the .90s.

 

Either way, I agree that the FP should count in the stat if the finder was a PM when the FP was awarded.

---------------------------------------------------------------

Take it up a notch. What happens if everyone stays at PM level, but one of those PMs REMOVEs a FP. Should the stat go down to 90%?

 

This sounds just like the discussions about COs dropping the D/T ratings on a cache. Should it affect previous finders' stats? Cancel Fizzy Grids? Dequalify challenge fulfillment?

 

I don't envy those folks at GS.

 

[+GeoElmo6000]

As a software engineer it would seem logical that the FP % would be calculated as (# FP / # Premium Finders @ time of find) (with 0 Premium Finders being set to 0% to avoid a divide by 0 error).

 

This would avoid any changes in membership though it could confuse people later (like if someone who found it as a Basic member becomes Premium but didn't add a FP later, the FP % wouldn't change but the "now" numbers wouldn't align with the FP %).

 

I'm don't know the actual algorithm but I'd love to hear what it is.

[+TeamRabbitRun]

1 minute ago, GeoElmo6000 said:

As a software engineer it would seem logical that the FP % would be calculated as (# FP / # Premium Finders @ time of find) (with 0 Premium Finders being set to 0% to avoid a divide by 0 error).

 

This would avoid any changes in membership though it could confuse people later (like if someone who found it as a Basic member becomes Premium but didn't add a FP later, the FP % wouldn't change but the "now" numbers wouldn't align with the FP %).

 

I'm don't know the actual algorithm but I'd love to hear what it is.

 

Not necessarily accurate algorithm documentation, but...

 

[+GeoElmo6000]

1 hour ago, TeamRabbitRun said:

Not necessarily accurate algorithm documentation, but...

 

Right, but the OP was saying that his % changed when one finder's account went from premium to basic due to their passing.  I'm curious about the backend algorithm, how it accounts for membership changes and such.

[+barefootjeff]

The algorithm has been broken for the last two and a half years (see this thread), with it not properly counting the most recent log. One of my more recent hides (GCAHPD9), with 4 FPs from 5 PM finds, shows 75% FPs.

 

All its finders were PM when they found it and are still PM, so a change in membership status isn't the cause in this case.

Edited February 18 by barefootjeff

[+baer2006]

5 hours ago, GeoElmo6000 said:

As a software engineer it would seem logical that the FP % would be calculated as (# FP / # Premium Finders @ time of find) (with 0 Premium Finders being set to 0% to avoid a divide by 0 error).

 

This would avoid any changes in membership though it could confuse people later (like if someone who found it as a Basic member becomes Premium but didn't add a FP later, the FP % wouldn't change but the "now" numbers wouldn't align with the FP %).

Also a software engineer here , I would not base the algorithm on "time of find". Because FP can be awarded/retracted at any time, and this should be reflected in the percentage. If a PM doesn't award an FP, because they have none left, the FP % of the cache will go down. But if the add the FP weeks or months later, it should go up again.

 

That said, my algorithm would look like this: FP % = (#FP given by those finders, who are currently PM) / (total # of finders, who are currently PM)

 

A few corollaries of this: 

- If a cache has 100% FP, and one finder loses their PM status, the FP % stays at 100.

- If a non-PM finder becomes a PM, and they don't retroactively award an FP to a cache, that cache's FP % drops (only very slightly, if the cache has been found often).

 

Basing the FP stats on current PM status and current FP count is also quite easy to implement (because there is no need to store any past data), and easy to verify for the users (lke "my cache was found by 20 PMs, 18 of which have given an FP -> 90% FP").

[+Viajero Perdido]

23 minutes ago, baer2006 said:

Also a software engineer here

 

Yet another software engineer here (retired), trying my best to just keep my mouth shut.  

 

Said my bit years ago.  Best of luck. 

[+barefootjeff]

59 minutes ago, baer2006 said:

That said, my algorithm would look like this: FP % = (#FP given by those finders, who are currently PM) / (total # of finders, who are currently PM)

 

I think that's how it used to work before they broke it in 2022. For caches that get few finds, such as high D/T ones, the displayed percentage is now pretty meaningless.

 

I've also noticed that in the search results, if you click on the number of FPs it used to show the percentage of FPs but now it just says 0% regardless of how many FPs the cache has received.

 

 

This cache has 7 FPs from 11 PM finds so it should be 64% but the cache page is also wrong, showing 55%.

 

Edited February 19 by barefootjeff

[+funkymunkyzone]

Yet another software engineer here... lol

 

Wouldn't it be simplest to just count the FPs awarded by current premium members against finds by current premium members? As per baer2006.

 

Sure, it might miss some nuances, but really is it so important that it be super duper accurate?  Or is it good enough that it provides a guide as to caches that people enjoy or not.

 

In the case of the OP, they'd retain their 100% ratio as all current PM finds are accompanied by an FP.

Edited February 19 by funkymunkyzone

[+MNTA]

12 hours ago, funkymunkyzone said:

Yet another software engineer here... lol

 

Wouldn't it be simplest to just count the FPs awarded by current premium members against finds by current premium members? As per baer2006.

 

Sure, it might miss some nuances, but really is it so important that it be super duper accurate?  Or is it good enough that it provides a guide as to caches that people enjoy or not.

 

In the case of the OP, they'd retain their 100% ratio as all current PM finds are accompanied by an FP.

 

Not a software engineer but a hardware deign engineer that codes regularly.

 

The algorithm might get complicated due to the fact that folks let their premium membership lapse at times and some may add premium membership after finding the cache. Also remember that favorite points can be added and removed at any point including post archival. Sounds to me anytime a FP is awarded or removed the listing should update the small database the controls the FP%. This may be challenging but don't implement a feature you don't wish to support properly. Now throw in privacy concerns so do you encrypt the data or do best effort (This seems to be the current state)

 

Personally I am not a fan of FP% due to the variability of how folk use FP across the game. FP awarding a lot of times has nothing to do with the greatness of the cache but more on the experience of the day. I did not award GC12 a FP most do with a 53% but when I found it was a five gallon bucket full of smelly stinky water and half the swag needed to be thrown away as an example. Add variability in awarding between regions and tourist draws skew things a lot. Looking at my Favorite point top 100 finds on project-gc the FP% mean nothing and range from 75% to 7%. Several I can't even remember others were amazing and some I question why they got a single FP. The middling ones to me were the best experiences. 

 

So what to do. I use FP as an indicator that maybe I want to find that cache, nothing more. I ignore the FP% and would not be offended if they removed it all together (other may disagree). 

 

[+TeamRabbitRun]

Having contributed up above, I suppose I should add that I'm a software engineer, too. It seems the 'in' thing on this thread.

[+barefootjeff]

6 hours ago, MNTA said:

Personally I am not a fan of FP% due to the variability of how folk use FP across the game. FP awarding a lot of times has nothing to do with the greatness of the cache but more on the experience of the day. I did not award GC12 a FP most do with a 53% but when I found it was a five gallon bucket full of smelly stinky water and half the swag needed to be thrown away as an example. Add variability in awarding between regions and tourist draws skew things a lot. Looking at my Favorite point top 100 finds on project-gc the FP% mean nothing and range from 75% to 7%. Several I can't even remember others were amazing and some I question why they got a single FP. The middling ones to me were the best experiences. 

 

So what to do. I use FP as an indicator that maybe I want to find that cache, nothing more. I ignore the FP% and would not be offended if they removed it all together (other may disagree). 

 

I guess it's a personal thing, but the sort of caches I enjoy the most tend to be those that involve a fair bit of hiking (or kayaking) and take me to a beautiful place I wouldn't have otherwise known about. Those caches, though, generally don't get many finds, often struggling to get beyond single digits, so they're never going to get a large FP count or an FP indicator in the app. In New South Wales, by and large the caches with the highest FP counts are old caches placed in tourist hotspots around Sydney harbour, which is great for tourists I suppose, but having grown up in Sydney I don't find those particularly attractive. Even in my own region (the Central Coast), the cache with far and away the highest FP count (105 from 353 finds) is a ten-year-old urban novelty container perhaps dubiously attached to the roof over a roadside visitor information sign.

 

For COs who specialise in more remote higher terrain caches, about the only thing they have going for them amongst the website stats is percentage FPs, although I find Project GC's Wilson score is a better indication of of the likely quality of the caching experience. But this isn't helped when the website's displayed %FP is wrong.

[+funkymunkyzone]

10 hours ago, MNTA said:

The algorithm might get complicated due to...

 

 

That's why I am saying keep it simple.  The query based on the stats as they are right now.  It might not be perfectly representing the right number of FPs due to changes in people's memberships, but that's just life.  It's only meant to be indicative of a good cache anyway.

 

10 hours ago, MNTA said:

Personally I am not a fan of FP% due to...

 

 

Yep, neither am I and I totally agree with you, and hence my above assertion to just keep it simple.

 

Something intended as an indicator and that is 99+% accurate 100% of the time is better than something that is 100% accurate most of the time and otherwise gives weird mathematically impossible results when encountering membership changes etc.

 

[+Goldenwattle]

12 hours ago, MNTA said:

I use FP as an indicator that maybe I want to find that cache, nothing more. I ignore the FP% and would not be offended if they removed it all together (other may disagree). 

I would disagree, as compared to %, FP numbers can be irrelevant. A cache in a tourist area with lots of FPs, only because it has finds (sometimes several) almost every day means nothing. However, a cache with say only ten FP seems to pale into nothing compared to that, until you look at the % and realise it also has 100% or close to %. That means a whole lot more. FPs are 'cheap' and can mean only that cache gets logged frequently. A high % means that it's likely to be a great cache. % is MUCH more an accurate guide to the cache.

[+CheekyBrit]

It sounds like we need a new statistic on the hiders page of project-gc: average favourite point percentage per hide. 

[+geodarts]

On 2/18/2025 at 9:18 AM, Deepdiggingmole said:

One of the cachers is now only a basic member. .  One reason the one member who was premium and is now basic is because they died and their membership lapsed only when the next renewal was due and it didn't occur. 

 

I don't pay much attention to Favorite Points in terms of caches that I decide to seek, but it takes a certain amoint of thought to award favorites and I always assumed it was part of my caching history. Apparently that part will disappear after I pass. 

 

I realize I won't be around to care -- as much as I believe my profile page is worth reading, the odds of that happening are already slim and will be even less in the likely event of my death.  Still, I am a little sad that my favorites will disappear.  

[+barefootjeff]

40 minutes ago, geodarts said:

 

I don't pay much attention to Favorite Points in terms of caches that I decide to seek, but it takes a certain amoint of thought to award favorites and I always assumed it was part of my caching history. Apparently that part will disappear after I pass. 

 

I realize I won't be around to care -- as much as I believe my profile page is worth reading, the odds of that happening are already slim and will be even less in the likely event of my death.  Still, I am a little sad that my favorites will disappear.  

 

Several of my caches have received FPs from someone who was a PM at the time but has now dropped back to Basic. Those FPs haven't disappeared.

[+funkymunkyzone]

21 hours ago, geodarts said:

 

I don't pay much attention to Favorite Points in terms of caches that I decide to seek, but it takes a certain amoint of thought to award favorites and I always assumed it was part of my caching history. Apparently that part will disappear after I pass. 

 

I realize I won't be around to care -- as much as I believe my profile page is worth reading, the odds of that happening are already slim and will be even less in the likely event of my death.  Still, I am a little sad that my favorites will disappear.  

 

Your FPs awarded to caches beoonging to a particular CO will still show up as FPs for that CO, so no, your FP legacy will indeed live forever!

[+Deepdiggingmole]

On 2/18/2025 at 8:45 PM, barefootjeff said:

The algorithm has been broken for the last two and a half years (see this thread), with it not properly counting the most recent log. One of my more recent hides (GCAHPD9), with 4 FPs from 5 PM finds, shows 75% FPs.

 

All its finders were PM when they found it and are still PM, so a change in membership status isn't the cause in this case.

OK - so in my example where there were 16 finders (all premium, but that wasn't by design as it isn't a PMO cache) and all 16 had given an FP - I had presumed and maybe wrongly based on your finding that due to the elapsed premium membership of one of those cachers. Consideration that it is the possibly exclusion of the most recent log - is there a time factor in that. The last time an FP was added was over a month ago, I'd have hoped that after a month any calculations would have caught up
So I am still leaning towards the lapsed membership issue 
Oddly, though correctly, in the top favourites % table, project-GC do show this as 100% 

[+Deepdiggingmole]

On 2/23/2025 at 1:14 AM, geodarts said:

 

I don't pay much attention to Favorite Points in terms of caches that I decide to seek, but it takes a certain amoint of thought to award favorites and I always assumed it was part of my caching history. Apparently that part will disappear after I pass. 

 

I realize I won't be around to care -- as much as I believe my profile page is worth reading, the odds of that happening are already slim and will be even less in the likely event of my death.  Still, I am a little sad that my favorites will disappear.  

It isnt the FP disappearing that is the issue as the FP didn't disappear - the stats still show (on the cache in question in the OP) 16 finds and 16 FPs awarded - neither of those figures have changed - what has changed is the %age figure you get when you click on the Favourites box on the cache page it went from 100% to 94% despite no further finds and no FPs removed 
So your stats will remain :-) 

[+Deepdiggingmole]

On 2/22/2025 at 8:31 AM, CheekyBrit said:

It sounds like we need a new statistic on the hiders page of project-gc: average favourite point percentage per hide. 

That is sort of covered in the 'Top Favourite Caches (%)' found in the statistics tab - and there, this particular cache still shows 100%

[+barefootjeff]

2 minutes ago, Deepdiggingmole said:

Consideration that it is the possibly exclusion of the most recent log - is there a time factor in that. The last time an FP was added was over a month ago, I'd have hoped that after a month any calculations would have caught up

 

There is no catch-up time with this bug. One of my caches (GCA4AG7) was last found just over a year ago. It has 5 FPs from a total of 6 PM finds (all the finders were and still are PMs), but it was the most recent finder who didn't give it an FP so it's still showing 100% FPs.

[+Deepdiggingmole]

On 2/18/2025 at 5:44 PM, TeamRabbitRun said:

Take it up a notch. What happens if everyone stays at PM level, but one of those PMs REMOVEs a FP. Should the stat go down to 90%?

Actually I would say yes 
I agree with the linked comment about more FPs than premium members (as is the case with my cache) 16 FP/ 15 PM should be 107% (rounded up) but it would appear they have turned this round to read 15PM / 16FP which gives the 94% - however the fact as in my example it shows more FP given than PM counted which is illogical 
but going back to the above quote - if a PM removes an FP (it happens quite a lot, particularly when caches get archived) then this would be logical
Taking my example - originally 16 PM awarding 16 FP - one gets removed so 16PM now giving 15FP - that would then correctly show the figure as 94% - it certainly does seem a glitch in their calculations 

[+barefootjeff]

4 hours ago, Deepdiggingmole said:

I agree with the linked comment about more FPs than premium members (as is the case with my cache) 16 FP/ 15 PM should be 107% (rounded up) but it would appear they have turned this round to read 15PM / 16FP which gives the 94% - however the fact as in my example it shows more FP given than PM counted which is illogical 
but going back to the above quote - if a PM removes an FP (it happens quite a lot, particularly when caches get archived) then this would be logical
Taking my example - originally 16 PM awarding 16 FP - one gets removed so 16PM now giving 15FP - that would then correctly show the figure as 94% - it certainly does seem a glitch in their calculations 

 

Looking at your cache page, I'd say the most likely explanation, given what I've seen on other cache pages since the bug was introduced two and a half years ago, is that it's counted the most recent find in January but not counted the FP they awarded, making it 15 FPs from 16 finds (94%). You can see a similar thing on this cache (GCB829FE) which has 2 FPs from 3 finds (all PMs) but is showing 33%. My find was the most recent but at the time I didn't have any spare FPs so added that about a week later. It's counted my find but not my FP.

[+Deepdiggingmole]

On 3/3/2025 at 11:24 PM, barefootjeff said:

 

Looking at your cache page, I'd say the most likely explanation, given what I've seen on other cache pages since the bug was introduced two and a half years ago, is that it's counted the most recent find in January but not counted the FP they awarded, making it 15 FPs from 16 finds (94%). You can see a similar thing on this cache (GCB829FE) which has 2 FPs from 3 finds (all PMs) but is showing 33%. My find was the most recent but at the time I didn't have any spare FPs so added that about a week later. It's counted my find but not my FP.

Possibly, and not wanting to say that that isn't one of the glitches.
So you say this issue has happened since around  2 1/2 years ago - however in the cache in question when the last FP was given (which was only 2 months ago)  - taking it up to 16 that %age did go up to 100%. If indeed there has been an issue for sometime why wouldn't this figure have been incorrect before - the page shows the correct number of FPs - it was only when we discovered that one of the finders member status had changed that we discovered the change in the %age - it didn't happen following the addition of the latest FP 
So, as said, not disagreeing with the addition of the last FP as being a cause - but it would seem the issue with mine is the member status change but with the calculations all cock-a-hoop