Classic Coins: How Big Can A Rim Ding Be Without A Details Grade?

How tall is a tree?

I agree with wisdom our friend the frog has ribbeted (he is typically wise anyway)!

You are trying to apply something with an actual measurement to an industry where there are no factual, verifiable, measurable standards employed.

Sceniario:
Ask 10 people to draw a picture of a ruler for you as accurately as possible. Then use those rulers to measure the length of a board.

Now ask yourself which is the verifiable length of the board?

THis is how the grading companies grade coins...subjectivity.

Now...suppose two of the people you asked to draw a picture of a ruler for you were people who spend their time at work painting numbers on rulers using an actual standard scale. Then, of course, you would expect their pictures of rulers to be more accurate than the others correct?

And that is how the TPGs reason that their system of coin grading can be legitimate. They train their graders so it is not someone randomly coming in and grading a coin.

Yet a vast majority of people fail to understand that when there is nothing verifiable, by any legitimate means, used, then you cannot expect a legitimately verifiable grade to be assigned. At best we have a "fuzzy cloud" of what a mass of graders, over time, have assigned to coins that are very similar to one another.

But as proved by the fact they make a lot of money from people re-slabbing coins, what you see on a holder is NOT the grade of that coin. That ink on the label is only a current subjective opinion of the graders who saw that specific coin the day it was slabbed. That ink can, and will change depending on the grader(s) who saw it.
Graders are only human. After 8 hours a day in a cubicle looking at coins, there are going to be fluctuations in their own grading abilities on any specific day. WHich Is why I say the grade can depend on how good the coffee in the office is on that day. Case in point (also not hard to find when you start looking):



For a REAL eye opener check out the essay in my signature showing at a minimum (being generous) a 30% error rate they have concerning a rookie level coin variety.

This is very true...which is why I put in about the "professional ruler makers," being some of the people.

We are on the cusp of AI grading which, I am hoping brings more reliability and verifiable accuracy to the grading process. I know at least one model said they would have the AI fed a database of already slabbed coins to "average" the coins that grading company humans, overall, have assigned to each MS etc. designation. I think keeping the human aspect in the system will make this more accurate grading more palatable for a lot of people who have a lot of money invested into slabbed coins. Those people stand to lose a lot if one of their prized registry set slabs is accurately graded to be a lower grade (which some will). However, those same people also stand to gain if one of their coins is actually a higher grade (I also think some will).
Meanwhile it has the potential to make a one-time grade. Less money will be sent on re-slabbing and go into getting more coins for collectors. But the TPGs won't be happy about AI grading. Tech grading, and bringing legitimacy through verification, will eliminate re-slabbing and the large profits the TPGs see from this.

Just making a neutral statement here b/c I would like to see legitimacy and accuracy brought to our hobby. I personally do not care about how well people will accept the system. They will eventually accept it.

The reason I say this is b/c the excuse the TPGs used when they abandoned their computer grading systems in the 90s, which, BTW, they marketed as being much better than human grading b/c it was not subject to human error, was that people did not trust computers. So the rest of the world computerized while the coin grading systems went back to what the TPGs had called the inferior and less accurate system.

I suspect (purely theoretical), that actually what happened is the shareholders did not like the great drop in profits that the tech-based system caused. Re-slabbing is very profitable and when a different grade will not be given, then there is no impetus to re-slab...hence profits fall. Businesses go where money leads.

And another theory I have is that when you keep things so you can apply anything you want to a label, then you can also not be held accountable and can manipulate the market in any way you want for your own benefit. If you want to talk about the PCGS guarantee we can go there as well b/c I have studied it and the language used makes it seem OK, but it turns out to only benefit THEM unless it is a situation where they slabbed a fake coin.

I claim no expertise in the area of modern programming such as you mention. However, there was already a Morgan dollar AI grading system a couple of years ago that was put online by Daniel Malone as a test . ANyone could go to the website, upload images, and try it. It was consistent and was designed, as I mentioned above, using already slabbed coins at each grade level as its database. The funds to keep developing though were not available.

At present there is another AI grading system started called Numi that was featured here at CoinWorld:
https://www.coinworld.com/coinworld-podcast

This one is free although it takes a subscription to ChatGP. It is in development and has, according to the author, been getting better at consistency as it goes...which is what an AI should do as it learns.

Another point I would make is that I have been following up on every technology based grading programs I have been able to find/encounter. I am spurred on b/c from the essay in my signature you can see proof linked to the PCGS website where there have been unsuspecting people taken for a lot of money b/c the don't buy the coin, but have 100% faith in the slab.

I have seen actual AI programmers respond at times that they believe AI is more than capable of analyzing and grading. To this I can only say, I do not know. I am not an AI programmer. But I have seen it posted a number of times that people into AI programming have said this concept is more than feasible.

But...until it is done, it is not done!


I know grading could be done analytically as long ago as '98. I was working in QC department of a plastics company that had a large machine with a stage onto which we put very tiny parts.
We manipulated the part onstage to train the system. We would save the program and put another of same part onto the stage. The system scannend and analyzed the part to give us any calculation we wanted to know about how the two parts varied with measurements down to the micrometer level. That is more than enough for grading coins. Who knows, it may have been what the grading companies were using before abandoning the better system?
Nowadays, that system would be very primitive compared to the computers we carry in our pockets.

Facial recognition tech easily (taking 30 thousand data points to use for analysis) no doubt could easily grade a coin to the level we are accustomed to. But again, give a coin one grade and the companies lose a lot of profits. The TPGs are not going to welcome any change. They stand to lose a lot of power in the market and a lot of money.

Coin grading has much less than 90% accuracy. NGC wants their graders around the 65% accurate mark. Too wide of a spread to even consider placing objective values on something like a rim ding.

I even have a problem with them trying to determine how accurate the grading is at all. Just how does one tell if they get it right 65% of the time if there is no actual, measurable standard to tell what "right" is?

For example, there is no universal ONE standard Morgan dollar that has been verifiably graded as MS65 by anything verifiable. So how can a company say their graders grading a Morgan at MS65% get it right 65% of the time?

There is a promo video out there for PCGS called "CINweek: We take the CGS Grading Challenge."

Mr Charles Morgan is a CoinWeek editor. To see how good he is at grading coins (or so he thinks), Mr. Morgan attends a show and takes the PCGS challenge which involves guessing the "right" grades for 10 PCGS slabbed coins.

Mr. Ron Guth, a former PCGS president, is administering the test.

After Mr. Morgan is done giving his answers, Mr. Guth tells him which grades he was "right" about and which grades he was "wrong" about.

And...that seems to be about where most people's reasoning takes them. :doh:


However...
Since any coin can be cracked out, resubmitted to PCGS and never be guaranteed the same grade, the "right" answers Mr. Morgan is checked against could change! They are not THE "right" answers because they can change!

How would you like it if all the tests you took in schooling had no verifiable correct answers?

Yet Mr. Morgan...and I am sure a host of people watching the video, never pick up on this fact. :think:

If the slabs used in the challenges were cracked, and the coins resubmitted, then how many answers Mr. Morgan got "right" would also change and alter his overall score (yet he did nothing)! :doh:

Yet in the video, Mr. Morgan actually believes the score he received...oh well...bow to the gods of marketing.

Mr. Guth later in the video even says that the best graders only get it "right" about 85% of the time. What does that statement legitimately mean when applying some common sense?

When there is no factually verifiable way of getting THE ONE right answer, then how in the world can you possibly say how far off of the mark someone is? You cannot! There is no way to tell 85% from 65% "right" when there is no verifiable way employed to know what "right" means.

Every time in the video where they say things like, "the right answer," as a former teacher I cringe. And then there are other videos of people happy with what they got when taking the challenge.

If one set of two graders did all the slabs used in the challenge, then all a "good score" means is the person taking the challenge was good at guessing what those specific graders put ont he slabs. If a different set of graders other graders re-graded the slabs, then there is no guarantee the achieved score would stand..and likely the math would indicate the score would not stand as it was.

I would hate to think the number of people who absentmindedly think the video is legit in what it appears to show.




Read the essay in my signature if you want to see a bonafide study of their accuracy with assigning a rookie level Kennedy half dollar variety. I am generous when I say they have a 30% error rate...it is higher.

How true!

They probably have hundreds of coins that a set of senior graders believe are solid for the grade that are used to test newcomers. For incoming graders, they try to have them set on 65% accuracy before they are able join the team. I do not work for NGC, but this information is coming from an acquaintance.



Couldn't help but think of science vs art classes. In any math class, there is either single or multiple right answers. You either get it wrong or right. English class, a class in which there is no right answer, I never got 100% because there is always something to improve on, from the perspective of the teacher. Coin grading is an art, in which there is no right answer.

This distinction becomes clear with damaged coins, examples with varying eye appeal, and weak strikes. With a blast white Morgan dollar, there is a decent chance there is a "right" answer. On the other hand, a weakly struck bust half dollar will have different opinions among graders.

In this business, there is in fact a "correct" standard, and that is of the TPG grading services. If you have a Morgan dollar that you strongly believe is a 65, and PCGS continues to return it as a 64, what does it profit to you to continue thinking you are correct? When I get passed a coin to grade and value, I first ask which grading service will be receiving the coin. The multitude of companies have differing standards that need to be learned.