Coin Grading - Human Versus Laser/Human
Zsmartie
Posts: 135 ✭✭✭
Okay, I will first off state I am relatively new to coin grading, however, have long enjoyed it. I admit I am not very good at it, however, I enjoy trying to guess the grade. My question is related to the method used to grade a coin. This has bothered me for some time now. I am a scientist by education and profession, Master of Science Degree in Atmospheric Physics. In my work the assistance of computers and electronic gadgets was absolute. So for me to go from that to trying to fathom grading a coin for bag marks, wear, strike, and lastly eye appeal totally by humans unimaginable in this day and age. And especially when the value of that coin can vary considerably by one or even less grade point. I realize these graders are some of the best available and know their business. It is just to me the judgements on the variety of items that go into a grade as being still subject to random variations due to conditions of the graders. Do all graders grade identical? Do they get tired? Does their mood affect their work? It seems to me in this age of electronics, the somewhat fixed items in a grade (i.e. bag marks, wear, strike etc.) could be very accurately measured by a laser. Obviously the laser programming would have to be educated, however, it sure seems like that phase should be automatic. Then the human could enter the grading by verifying to a degree what the laser measured and then add in the eye appeal. Eye appeal to me also is very subjective as what defines good eye appeal. To me what appeals to some collectors does not appeal to me and I would not buy it. However, this is important in the grade. If so, leave it as is but why not use technology to augment the process and reduce random variation?
Comments
It's been talked about for 25 years. AI might help, but it is very hard for the laser to differentiate things like weak strike vs. Light wear. You also needto separates visible hits in focal areas versus an accumulation of small hits in non-focal areas. Cabinet function versus light scratch. What does it think of die polish lines? Can it recognize light cleaning?
The fact is that grading is as much an art as a science. Would you buy an ugly coin with a high technical laser grade?
All comments reflect the opinion of the author, even when irrefutably accurate.
@Zsmartie asked:
"Do all graders grade identical?" No, but generally very close. Most professional graders can guess the grade of a coin in hand when it is graded by another professional. Apparently PCGS did a study and found that the best uniformity comes with three graders and a finalizer.
"Do they get tired? Does their mood affect their work?" Probably, graders are human. I don't notice getting tired but things outside the office can cause distractions. The minute you are distracted you may miss an authentication. IMO, grading is much easier except for uncommon series that I may need to confirm my opinion in a guide of some kind. I think some graders are so knowledgeable that they could grade a coin "blindfolded," The main problem is the difference between the coins actual condition (technical grade) and its commercial grade (value). A commercial grader's opinion is the best for a TPGS.
"It seems to me in this age of electronics, the somewhat fixed items in a grade (i.e. bag marks, wear, strike etc.) could be very accurately measured by a laser. Obviously the laser programming would have to be educated, however, it sure seems like that phase should be automatic. Then the human could enter the grading by verifying to a degree what the laser measured and then add in the eye appeal. Eye appeal to me also is very subjective as what defines good eye appeal. To me what appeals to some collectors does not appeal to me and I would not buy it. However, this is important in the grade. If so, leave it as is but why not use technology to augment the process and reduce random variation?"
So far, different methods have been tried to remove the human from the end point after the machines are set up. ALL have failed.
Thanks. It is just something for me as a scientist it is hard to comprehend.
I like Insider's insider insight. I love your post. I'm a Civil/Structural Engineer by training, turned surgeon. Coin grading is an unholy mis-mash of objective scientific evaluation and "Oooooohhh, look how pretty!"
The problem is this:
The current dogma in grading (appraising) is that the most important factor is eye appeal. Eye appeal is subjective, period. I don't see a way around this.
Even with eye appeal factoring into grading, it isn't perfect. Pretty toning might bump the assigned grade a point, but the market value by three or four grades. It's just a continual approximation and a work in progress.
I think you are talking about the Ben diagram intersection between technical grading and market grading. CAC market grades and PCGS has drifted that direction for decades which is one of the reasons the market prefers them along with their superior holders.
Price is the true grade held back by technicalities
11.5$ Southern Dollars, The little “Big Easy” set
Thanks. It is just something for me as a scientist it is hard to comprehend.
I'm a scientist, too (PhD), but the blend of science and art used for coin grading has never bothered me. I guess because I was around prior to the TPGs and access to computers.
In honor of the memory of Cpl. Michael E. Thompson
The other point I will make is that coin grading for the grading companies is financial. PCGS has long stated they are more accurate and probably restrictive in their grades. However, for me as a collector, am I helping pay PCGS at the sake of the getting the most accurate grade (ignoring most restrictive) for coins I am submitting. After all it is the coin, not all the fluff and whatever. I respect the point that coin grading companies are in the business to make a profit. However, us the collector, want to get the most accurate grade of our coin, irrespective of the grading companies politics or profit. So would change make a difference. From what I am hearing, probably not. The humans on their best are the best way to go. However, the point often made that PCGS is the most accurate (and restrictive) grading service does bother me. I do not know what way to think on that comment. I do want the most accurate grade, however, I do not want to weigh that against the goal to remain the most restrictive.
Well, if you are writing about "accuracy" and TPGs and believe it is the same as standardized accuracy and precision then you don't understand all the nuances of TPG grading. There is no definitive, universal grading standard (such as from the ANA) that TPGs must follow, which means that each TPG has developed their own, proprietary grading standard. Obviously, there is significant overlap between most grade ranges and most TPGs, but the "liner" coins and some problem vs. problem-free coins will grade differently depending upon the TPG that they are submitted to. However, even though they grade differently, both TPGs could be correct as long as they grade according to their own, in-house standards.
Therefore, a good way of looking at the PCGS branding is to realize that PCGS is stating that their grading standards are (slightly) more conservative for the majority of coins and series and (slightly) more precise on a coin-by-coin data point, which may translate into greater market value and/or liquidity.
In honor of the memory of Cpl. Michael E. Thompson
@Zsmartie said: "However, the point often made that PCGS is the most accurate (and restrictive) grading service does bother me. I do not know what way to think on that comment. I do want the most accurate grade, however, I do not want to weigh that against the goal to remain the most restrictive."
I don't know what you mean by "restrictive."
First, consider that we are posting on a PCGS forum. What do think you'll read here? And anyone who tells you that to your face is, ah, well...
Two things that NO INFORMED person can argue with:
As long as a grading service is meeting your expectations, that's the one to stay with. One more thing, PCGS coins will generally bring more money when sold.
@Zsmartie...I am generally in agreement with what you have stated. I am not a scientist, but worked in industries that required hard, documented standards (aerospace electronics) in various areas (mechanical, dimensional, color, lighting to name a few). I have stated here many times that grading is subjective and should have standards and even become computerized. The task is monumental. That is why it has not been done successfully yet. I do predict that at some future date it will become reality - perhaps the next great step for the TPG business. As far as 'eye appeal', that should never have become part of grading in my opinion. Cheers, RickO
I think that humans should continue to grade classics.
However, I think there's room for modern proof/BU silver eagle and commemorative grading by machine.
I think software would do an adequate job discerning beteeen a 68/69/70.
When we have an AI program capable of rating the desirability of Tiffany lamps by condition of preservation and retention of color intensity, we'll be ready for the coin art appreciation module.
Modern PR70/MS70-oriented issues could be much more easily computerized.
IMO, AI could easily detect marks, HLNS, spots, etc. I'll bet it could even distinguish a strike thru from PMD. Thus, it could separate a 70 from the top PR and MS grades of moderns.
It would take a while for AI to "learn" error coins if it ever could.
Most big ones are unique.
The biggest problem I see is the computer will point out all the AU-62's and the industry really doesn't want that. It's OK to pay MS-64 money for an AU, but not if the label says it's an AU. Market grading requires you program the computer to lie, which is fine, but where do you stop with that?
Not at all. The AI runs the definition as input. The definition is what the TPGS says it is. The definition is what they are currently using. It's not a "lie", it's a standard.
All comments reflect the opinion of the author, even when irrefutably accurate.
No, Same as I don't buy a fugly coin with a technical grade ensconced in plastic. Eye appeal + correct grade is the winning combo.
I imagine that a person or group of people will have to program the computer. So will a group of computers do the first pass on the coin, followed up by the head computer grader? How is that going to work? Hmmm.