Options
Computer assisted grading and coin ranking -- a two part question.
mercurydimeguy
Posts: 4,625 ✭✭✭✭
in Q & A Forum
Hi David,
A bit long-winded, two-part question:
Part I
Have you ever thought about computer assisted grading, which is different than computer grading (discussed thus far)?
Human beings can process contextually relevant information far better than computers. However, computers are able to mine this contextually relevant information more accurately and rapidly than human beings. So if computers can be used to present human beings contextually relevant information, it would seem that human beings would be able to more accurately make consistent decisions, and more rapidly (increasing consistency, turnaround times, reducing operating costs, etc.)
In essence a computer can be leveraged to quickly present contextually relevant information to a grader, who would then use this information to reach a (hopefully more consistently accurate) decision on a grade.
Part II
Have there been discussions around not changing the grading scale but rather adopting a different model altogether?
Using the above computer assisted model, for example, a coin can be "ranked" amongst its peers and have it's ranking become the numerical attribute of the coin. This is kind of like they do in scholastics and in childhood development (e.g. your child is in the XX% percentile). For instance, and an example of a common coin, if a computer had data about 1,000's of 81-S Morgans, it can say that this Morgan is in the 97%'ile of all the 81-S Morgans and this other Morgan is in the 82%’ile of all the 81-S Morgans.
It would seem that in the above "ranking" model a collector/dealer/investor could be better equipped with more contextual information about a specific coin (e.g. how does this 81-S Morgan compare to all the other 81-S Morgans). An MS65 81-S might translate in to 68.9% coin while an MS65 92-CC might translate in to a 94.5% coin. This ranking model, coupled with computer assisted grading, might also be able to mitigate any ambiguities around designations altogether because DMPL, for instance, would simply translate in to a coin in a higher percentile.
Thus, a ranking model would still move to a decimal based 100 system (albeit 100%), which has been discussed as potentially being easier to understand for newcomers, and at the same time it could also mitigate any ambiguities that sometimes shroud designations on coins (PL, CAM, FB, FH, FS, etc.).
I wanted to bring up some of these thoughts for discussion while at the FUN PCGS luncheon, but unfortunately time didn't permit for that much open Q&A.
The above are obviously real abstract/high level thoughts. Yet, I wanted to ask if you had ever considered anything along the above lines of thinking and/or what your thoughts were in any event.
Thanks in advance!
MDG.
A bit long-winded, two-part question:
Part I
Have you ever thought about computer assisted grading, which is different than computer grading (discussed thus far)?
Human beings can process contextually relevant information far better than computers. However, computers are able to mine this contextually relevant information more accurately and rapidly than human beings. So if computers can be used to present human beings contextually relevant information, it would seem that human beings would be able to more accurately make consistent decisions, and more rapidly (increasing consistency, turnaround times, reducing operating costs, etc.)
In essence a computer can be leveraged to quickly present contextually relevant information to a grader, who would then use this information to reach a (hopefully more consistently accurate) decision on a grade.
Part II
Have there been discussions around not changing the grading scale but rather adopting a different model altogether?
Using the above computer assisted model, for example, a coin can be "ranked" amongst its peers and have it's ranking become the numerical attribute of the coin. This is kind of like they do in scholastics and in childhood development (e.g. your child is in the XX% percentile). For instance, and an example of a common coin, if a computer had data about 1,000's of 81-S Morgans, it can say that this Morgan is in the 97%'ile of all the 81-S Morgans and this other Morgan is in the 82%’ile of all the 81-S Morgans.
It would seem that in the above "ranking" model a collector/dealer/investor could be better equipped with more contextual information about a specific coin (e.g. how does this 81-S Morgan compare to all the other 81-S Morgans). An MS65 81-S might translate in to 68.9% coin while an MS65 92-CC might translate in to a 94.5% coin. This ranking model, coupled with computer assisted grading, might also be able to mitigate any ambiguities around designations altogether because DMPL, for instance, would simply translate in to a coin in a higher percentile.
Thus, a ranking model would still move to a decimal based 100 system (albeit 100%), which has been discussed as potentially being easier to understand for newcomers, and at the same time it could also mitigate any ambiguities that sometimes shroud designations on coins (PL, CAM, FB, FH, FS, etc.).
I wanted to bring up some of these thoughts for discussion while at the FUN PCGS luncheon, but unfortunately time didn't permit for that much open Q&A.
The above are obviously real abstract/high level thoughts. Yet, I wanted to ask if you had ever considered anything along the above lines of thinking and/or what your thoughts were in any event.
Thanks in advance!
MDG.
0
Comments
I just don't thik the market is ready for more complex grading methods or scales. it's pretty precise now compared with other collectibles and people seem to enjoy the market...and the market's really good now.
Thanks,
David