are you pro or con on computer grading?

[halfpint]

 

this is an interesting and long read on the subject.

 

http://www.coingrading.com/compgrade1.html

[PerryHall]

As long as eye appeal is a grading factor, computers will never be able to grade coins.

[MTHead-migration]

neutral to not caring

[RWB]

It’s bad enough that people spend huge amounts to have somebody grade their pocket change and put it into plastic slabs. Why would anyone want to have the same done for their computer? Mine’s all dusty, and there’re stains on the pop-out cup holder, and scratches where I have to insert the hand crank to get it started…..

[World Colonial]

No.

 

I was too lazy to read your link but my opinion is that as long as the coin hobby or any segment within it is dominated by collectors, this idea will go nowhere no matter how advanced technology becomes.

 

As a prior post alluded to, there is market grading and then there is technical grading. A computer program can in theory perform technical grading but I consider it impossible for it to perform market grading because that is a subjective outcome based upon what people prefer, and this cannot be done consistently because preferences differ between collectors and change over time, sometimes possibly quickly.

 

If it is tried, I only see it working if coins or coin segments are controlled by "investors" (like generics and conditional rarities, some of which bring absurd prices) who want a commodity they can buy sight unseen. But I do not see this happening generally because coins ultimately are only worth the prices they bring because a collector is the one who creates the market for them.

[World Colonial]

It’s bad enough that people spend huge amounts to have somebody grade their pocket change and put it into plastic slabs. Why would anyone want to have the same done for their computer? Mine’s all dusty, and there’re stains on the pop-out cup holder, and scratches where I have to insert the hand crank to get it started…..

 

[nwcs]

As long as eye appeal is a grading factor, computers will never be able to grade coins.

Pretty much this is how I see it. I think it'd be perfectly fine for technical grading at some point, though. Now if we went with a two grade system again, maybe it would be good. One on a state of preservation and one on subjective eye appeal. Send everyone back to get their stuff slabbed again.

[123cents]

I think computer grading could be useful if used in conjunction with 1 live grader.

[just-john]

What do we know who's to say if some type of computer grading not being done now.When you see some of the coins that get into slabs that are the wrong date with a very questionable grade wrong type and then add in the ten's thousands of upgrade submissions who knows.Would you still pay your 20 bucks if you knew a machine was the only thing that saw your coin.But still when you figure the 10's of thousand's or more each month graded by the TPG'S divided by the number of graders employed don't think a machine or compute would be much faster maybe cheaper.BUT what do we know.

Like in the wizard of OZ don't look behind the curtain

[prudden]

It's been tried before.

 

http://www.sampleslabs.com/compugrade.html

[James_OldeTowne]

I don't think a computer can grade coins, but I think computer-aided grading is a distinct possibility.

[PerryHall]

I think computer grading could be useful if used in conjunction with 1 live grader.

 

If they disagree, who is right? Do they get another computer to break the tie?

[123cents]

I think computer grading could be useful if used in conjunction with 1 live grader.

 

If they disagree, who is right? Do they get another computer to break the tie?

 

 

No you have the computer flip a coin.

[rrantique]

It’s bad enough that people spend huge amounts to have somebody grade their pocket change and put it into plastic slabs. Why would anyone want to have the same done for their computer? Mine’s all dusty, and there’re stains on the pop-out cup holder, and scratches where I have to insert the hand crank to get it started…..

 

(thumbs u

[BankNote1]

Computer grading was used years ago on just Morgan Dollars I believe. I can't remember why it failed but fail it did.

I guess if that is the way coin grading is going then I hope they have gotten all of the bugs out of it!

[Oldtrader3]

PCGS, had an "Expert", Morgan dollar computer grading system back in the late 1980's but the technology was not there at that time, plus PCGS does not have the engineering talent to do something that complex and the system failed.

 

I managed the design and building of a needle and catheter visual inspection system starting at about that same time and spent about 3 years, probably $1 million in engineering costs and 1/2 million dollars in hardware costs to get it to work. It was an Allen Bradley based system, but we had our engineers redesign and reconfigure the system extensively. PCGS probably used an Allen Bradley system also. Hence the name "Expert" which was an A-B trademark.

 

Grading is somewhat subjective and I don't believe that Artificial Intelligence is to the point yet where computers can grade coins.

[coinman1794]

I would be willing to give computer grading a try if the result were used only as one of the consensus opinions, in addition to the 3 opinions that go into assigning the coin's grade currently. A 4th set of eyes, if you will...

[zrlevin]

The TPGs need to be careful, though. If they make technology that is cheap enough and easy enough to use, they will destroy their own market.

[MAULEMALL]

I think computer grading could be useful if used in conjunction with 1 live grader.

 

If they disagree, who is right? Do they get another computer to break the tie?

C A C

[okbustchaser]

Since it is no secret around here that I disagree with the entire idea of third party grading/pricing I suspect that most of you will be suprised to learn that I heartily endorse the idea of computer grading...for the very reason that many of you say it wouldn't work--ie, it takes "eye appeal" out of the equation. Instead, it creates a repeatable technical grade for a coin which can be used as a starting point between buyer and seller--the only two people who should care what the coin should sell for.

 

[MikeInFL]

I am in favor of computer grading, however I believe that it can't be done yet. We simply don't have the technology to assess eye appeal, and like voice recognition -- it's not very useful unless it automates the process entirely, so a solution that is 90% good enough really isn't good enough.

[jesbroken]

It will not happen, IMHO, simply because if the software/hardware were available to realistically do this, then why would we need the TPG's. All could own the equipment and software to do it ourselves(even if too expensive for the average collector, as the market became available-the price would come down-as has happened already to so many programs when groups or many people produced a market). The TPG's would not wish this to occur if it only took 2% of their market it would suck for them. Not gonna' happen.

Jim

[ant]

the whole "eye appeal" thing gets me. how is a coin graded by eye appeal? i like original looking coins and some like bright dipped clean looking coins "white" if you will. it seems as though there job is about condition, not brilliance.

 

 

[Conder101]

The problem is the services try to tweak the programming to put "eye appeal" back INTO the equation. So the final result is one that fails at market grading because the computer can't judge eye appeal, and it fails at technical grading as well because of the additional messing around wih th results. The only good thing you can say about it is that it is repeatable. Give it the same coin multiple times and you will always get the same result back. At least until they attempt the adjust the program again to fix the "eye appeal" failure.

[World Colonial]

the whole "eye appeal" thing gets me. how is a coin graded by eye appeal? i like original looking coins and some like bright dipped clean looking coins "white" if you will. it seems as though there job is about condition, not brilliance.

 

 

I believe the asnwer to your question is because the primary purpose of the TPG is to make the coins more marketable from a financial standpoint, If it were possible to remove financial considerations from the equation, there would be little or no business case for it other than for relatively expensive coins to determine authenticity.

 

The financial side is why I have my better coins in holders. In my case, I can buy many of them raw, get them graded and sell them for a substantial markup though I keep most of them. The collectors of these coins prefer graded coins and will pay for it. But if this were not the case, I would not bother with it. I'm not worried about buying fakes and I can certainly get a decent looking holder for less than the $19 or $32 I pay to NGC.

[coinman1794]

the whole "eye appeal" thing gets me. how is a coin graded by eye appeal? i like original looking coins and some like bright dipped clean looking coins "white" if you will. it seems as though there job is about condition, not brilliance.

 

 

Eye appeal has been one of the key factors of coin grading since its inception, and is inseperable from the process. Where the grading services go horribly wrong is in market grading coins, i.e., taking eye appeal to the next level by assigning a higher grade based on an eye appealing coin's extra value, rather than on a consideration of its various merrits.

 

[lehigh96]

the whole "eye appeal" thing gets me. how is a coin graded by eye appeal? i like original looking coins and some like bright dipped clean looking coins "white" if you will. it seems as though there job is about condition, not brilliance.

 

 

Eye appeal has been one of the key factors of coin grading since its inception, and is inseperable from the process. Where the grading services go horribly wrong is in market grading coins, i.e., taking eye appeal to the next level by assigning a higher grade based on an eye appealing coin's extra value, rather than its merrits.

 

I completely disagree with that statement. If it is OK to lower a grade by market grading for weak strike or poor luster or ugly toning, then it should be OK to raise a grade for strong strike, superb luster, and fantastic eye appeal. That is the whole purpose of market grading and the TPG's state very clearly that they practice market grading and not technical grading. Just my opinion!

[World Colonial]

I agree with you if I understand you correctly. Market grading makes sense for valuing coins and since this is the purpose of it, it would be pointless to use only technical grading because no matter what the technical grade, a coin which most or many collectors find inferior is not going to sell for more than a supposedly lower grade coin that looks better. That makes perfect sense to me.

[MikeInFL]

Market grading will be a huge challenge. Technical grading (or details grading in EAC parlance) is far, far easier. It's the subjective nature of eye appeal that makes it a huge challenge for computers, IMO...Mike

[DC Dawg]

Gee... somewhere in the past, we discussed this and I was very for the concept of computer grading. I think it is a great idea that would standardize grading. But first, let's look at a few of the things that have been said:

 

"PGCS tried this..." and "Compugrade failed..." is past history. Computer imaging technology and analysis has improved significantly even in the last five years. Look at your 10 and 12 megapixel cameras and think back to 2002 when a 2.5 megapixel camera was king.

 

Nowadays, most medical imaging systems are so much better than film that new systems do not automatically print these images unless necessary. Five years ago, imaging technology had less resolution than film. Today, the computer generated images are not only produced at a higher resolution than film, but many systems can create a virtual reality look into the body from the images collected!

 

"Man v. machine competition..." is a common fallacy that those of us in the computer industry has been dealing with since the rumors of computers going to replace people in the 1970s. Professionally, I have worked on many projects that tried to replace people with the machines. All but one failed. The one that was a success only displaced 130 workers and many of those were shifted to new jobs. Otherwise, if the projects are implemented correctly, the computer can help productivity allowing the same number of people to do more.

 

Can you image? Instead of waiting three weeks for the TPG to grade your coins, you can get them back in a week? Computers are the tools that work better with people to enhance their work. Computers should replace people in very limited circumstances and allow people to do other jobs.

 

"Technical grading v. market grading..." I know this is a touchy subject because I've stepped in it before. However, could it be that market grading has caused some of the problems with the TPGs? Could their attempt at market grading be the cause of the tensions that lead to the creation of a fourth party grading service (FPG)? Maybe the TPGs should re-consider market grading?

 

Technical grading has its merits. First, it is easier to be consistent. Technical grading can consider the number of bag marks, abrasions, tell if the luster is wrong, and many other physical characteristics of the coin. You can look at two MS64 coins and know that technically, they are really the same grade. It will force the buyer to look past the grade and into the plastic and determine whether the coin is something they want in their collection.

 

Yes, I know that this does not take into consideration "eye appeal." But like aspects of market grading, eye appeal is subjective. It disturbs me that the TPGs are deciding what is appealing and what is not and grading coins accordingly. Why not just grade the coin and let the collector decide if the coin has the proper eye appeal.

 

HOWEVER, I would not replace the graders with the computer. I would make the graders "finalizers" and let the computer do the grading. Computers should not be the end-all-be-all of any business (although we have seen too many people abrogate their brains to computers to the detriment of society). In this case, if the finalizer disagrees with the computer, the coin could be put aside and examined the "traditional way." If the coin is human graded, I think the label should note that the coin went through a "human review."

 

"Reality v. fantasy..." I think imaging and image analysis technologies would be excellent tools for numismatic grading. It would standardize grading and potentially reduce the need for the FPG. However, I am realistic enough to think that it will not happen any time soon. The problem is that the R&D resources that would be required to customize this technology for coin grading may be too expensive for the TPGs. We can look at the academic work for image analysis of gemstones for reflective properties (for applications other than cosmetic) and the resources that was required to perform that work, I don't think that the cost-benefit would help the TPGs.

 

Gemstone classification is used for other industrial purposes. The costs justifies the means. However, coin grading has a limited market. Plainly, there's no money to be made. Our only hope for computer-based grading would be for a graduate student to do this research as part of a dissertation that would be funded as part of the student's education. Otherwise, I doubt computer-based grading will happen.

 

I thought about doing a Ph.D. Maybe... instead of studying the economic impact of information security public policy changes, I could go into image analysis and do my dissertation on coin grading...

 

 

Scott