New value feature needs some editing control

13 posts

This message aims at: suggesting an idea to improve Numista

[Catalogue]
Status: Opened
Upvotes: 0
Downvotes: 0

» Quick access to the last post

If we are not careful, the new value feature could make Numista the laughing stock of the numismatic community. I have just seen this Lithuania 5 Litai where the value shown is something like 10 times over what it should be. I would strongly recommend that the method in which these values are calculated should be reviewed and/or the country referee should have the authority to overwrite a given value if they know it to be exaggerated or just plain wrong.
Just because you can't see it ... doesn't mean it isn't there - Anon.

Coin catalogue referee for England, United Kingdom & pre-Union South Africa.
Banknote catalogue referee for England & United Kingdom.
Quote: "radrick007"​ I have just seen this Lithuania 5 Centai where the value shown is something like 10 times over what it should be.
​I live in Lithuania and I can confirm that the value shown is correct. This particular coin has face value of around 1,5 euro but it was removed from circulation quite soon, so the market value is explainable by scarcity. (It is 5 litai, not 5 centai.)
ūūūūū
Thank you for your comment, I have edited my mistake.
Just because you can't see it ... doesn't mean it isn't there - Anon.

Coin catalogue referee for England, United Kingdom & pre-Union South Africa.
Banknote catalogue referee for England & United Kingdom.
Even if in this particular instance the value is actually accurate, I agree with Rick that some control over the values system should be exerted.

I suggest maybe if a referee (or member) sees an unreasonable value, they should be able to report it and a ref/admin would come along and, if deemed appropriate, delete any outlandish contribution to the average, or reset that particular value completely (as necessary and applicable).
Mornin' Rick.

I was under the impression that this issue had been addressed by discounting the two extremes rather than just averaging the numbers. That was how it was originally proposed by Imreh and accepted by Xavier. Is this not the case? I'm still using the original system so I'm just guessing. I am however looking forward to seeing it in action once I have a bit more free time to try it out.
Non illegitimis carborundum est.  Excellent advice for all coins.
Make Numismatics Great Again!  
IIRC, the listed value is the median, not the average, so the extremes are discounted automatically.
(In the case of exactly four values, taking the median is mathematically equivalent to discarding the two extremes and taking the average of the rest.)

In this particular case... I guess the coin actually is that rare. Which is weird, because its Numista rarity is somewhat on the lower side. Maybe we just have that many collectors of European coins by type?
I should probably try to look for that one next time I visit a coin shop...
Quote: "January First-of-May"
​Ah, that is good to hear. I guess they already factored in potentially stupid entries when designing the system. I just tried testing it by adding an entry worth €5.000 to a coin with a value of €0,54 and it didn't register. :wiz:
Quote: "pnightingale"discounting the two extremes
Right, this is what is implemented: the two extreme values are not part of the value calculation.
Sapientiae plerumque stultitia est comes.
Si c'est un grand plaisir d'être reconnu par ses amis, c'est peut-être encore plus flatteur d'être reconnu par ses adversaires.
Be yourself; everyone else is already taken.
I'm more worried about this kind of case : https://en.numista.com/catalogue/pieces7931.html
I think the algorithm should indirectly take into consideration the values of the lower and higher grades.
Quote: "CassTaylor"... by adding an entry ... and it didn't register. :wiz:





​One is not enough X-D two are enough

0,45 0,45 5000 --> 0,45
0,45 0,45 5000 5000 --> 2500,22
0,45 0,45 0,50 5000 5000 --> 0,50
0,20 0,45 0,45 0,50 5000 5000 --> 0,47 ( 0,20 UNLIKELY )
BOINC
Quote: "Cycnos"​I'm more worried about this kind of case : https://en.numista.com/catalogue/pieces7931.html
​I think the algorithm should indirectly take into consideration the values of the lower and higher grades.
​Not implemented yet but it's been discussed with related topic of precious metal coins where, to be consistent, a gold coin could not have a value below its melt value.
Sapientiae plerumque stultitia est comes.
Si c'est un grand plaisir d'être reconnu par ses amis, c'est peut-être encore plus flatteur d'être reconnu par ses adversaires.
Be yourself; everyone else is already taken.
I think some sort of value-control may be beneficial, but if referees have total control, the values members put in might turn out to be useless. I am not sure if this would be possible to implement (or if it is even a good idea), but it was just something I was thinking of:

What if referees were allowed to add, at most, four different values for a specific date/grade? That way, coins that no one (or only one or two or three) people own would be able to have a value displayed, the higher-values-for-lower-grades situations could potentially be avoided (for at least some coins), and if other members start to add their own values, the values chosen by the referees would still be calculated in the median, meaning the values they chose would not necessarily be the ones displayed.
IIRC there was already talk about also adding values from (local) catalogues or sales. These could indeed be added by the referee, but will then also be backed up by some real world value (not just made up).
Just call me Bram

Oh! And do read my profile page before you open a swap ...

» Forum policy

Used time zone is UTC+1:00.
Current time is 10:23.