Just Suppose? A color grading POLL

[valeria101]

I am actually surprised why automated color grading is not more popular already. What could compete with GIA's ?

- some way to tell higher D color ?
- some label for an extended range of "colorless" ?

At this point, I would vote for the second.

Let's assume a new color scale comes with equal steps between color grades, equal in width with what is now done for GIA's colorless range. Let's also assume there is a reasonable way to prove this new way is more consistent, without making the grades finer (serious problem, I would say).

Out of curiosity: what decision is facing now someone who would want to get a (GIA) H color diamond ?

The fifth grade from top down on the new system is closer to cololess than the old fifth grade (the desired H). It could actually look as a bargain even with some premium on the new grading services. Or the same person may well think that the new grades give more choice since the same range of phisical color would be split into more boxes now...

Lots of ways to talk up an even-paced scale even without stretching the eye further into the invisible.


Still, IMO, as usual

 

[valeria101]

Can a machine grade color more precisely than the eye ? Sure.

Technical precision can make the grading activity faster, more consistent, endlessly repeatable... But precission does not have to lead to narrower grades. This is a different choice entirely.

So, what would folks rather improve on the current scales ?

 

[nicknomo]

Date: 11/4/2004 7:21:43 PM
Author: valeria101
How can a grading system that cannot be controlled with anything but the one proprietary technology that produces it (because of high precision) increase confidence ? It does create exclusivity for the producers of the system... at least for a while. Could be a long while, if done properly.

Splitting even finer hairs exactly where technically needed lest (top of GIA scale) would only position this new system on top of the market.

Would it work ? Why not

It would not be the first time. Every new quality scale tempts with the question ''hey, those guys must know better, since their grades are more accurate'' Sounds as good as ever.

DO I like the idea? No way.... But again, that''s just me. I am still keeping a printout of this article on a poster

and the main idea is by no means in line with a finer diamond color scale.



Most likely this sounds bad. It shouldn''t though. I can understand why any new system must be in some blatant way ''more'' than the old to stand any chance: more precission, steeper ''quality'' scale, more lip service to ''rarity'' for an already large audience who has already internalized the concept.

Well the whole idea is that it is done by a machine, which has a much smaller margin of era. The way it is done now, the margin of era is +- 1 grade. That is far too high, considering close calls may cost or save the end buyer several thousand dollars.

As far as it being proprietary, I don''t think this is something that has to be subjective. light it light. It can be measured quantitatively by frequency/wavelength and amplitude. It''s just whether or not we have the technology capable of implementing this in a way where the data is consistent.

Even if you take out the "extra precision" on a grading report, machine grading would still be the best idea (if plausible)... Because even if it grades it a G, you will know it''s a "G". At this point though, why not go for the extra precision on a report itself?

 

[Nicrez]

Great article Ana and Hysterical!

 

[valeria101]

Date: 11/4/2004 7:53:22 PM
Author: nicknomo
At this point though, why not go for the extra precision on a report itself?

Nothing against precision... but against narrower grades.

When you say "G be a G" - this is about the accuracy and repeatability of grading - something that a machine can probably do better than people can.

Better technical precision makes narrower grades possible to implement. But this is an extra decision beyond that of automatingthe grading process. This is what I was arguing against.

As far as I understand, the very narow-ness of the grades is what makes grading difficult and inconsistent (that +_1 grade). It is almost too much even for trained viewers in a controlled environment, let alone buyers who never get either the training or the unrealistic environment. Why go even further ? Just 'cause a machine can do it ? At least I would not agree

 

[Nicrez]

Try buying a car in silver, and noticing there are 20 shades of each color available. Great, and each one has it''s own price, reflecting the most desirable being more expensive, and so forth...

People are scared to buy diamonds AS IS, imagine having a salesperson take the time to explain not just 26 colors, but 100+!!!

Again, this may fly at Pricescope, but the avarage diamond buyer is happy to buy a stone they like, think is attractive and consider it''s technical aspects really only when discussing price to fit their budget. Don''t you think most people would buy a D IF, if it was the same price as a same sized J SI2?

I think it''s a scary concept to further detail and categorize something that is categorized ENOUGH...Again, we are just basically creating a market that will be more specified, and just imagine that poor couple who can''t seem to find that perfect G1, eye clean SI1, 1.02ct, 34.5 crown angled, 57.5% tabled, 61% depthed, Hearts and Arrows stone because someone in Japan just bought it.

Honestly, the market has only so many stones of particular specification, so what makes you think that a FURTHER specification will make more of these stones appear? Then you''ll get a SUPER high premium on stones that are harder to find (maybe it could be G1''s), and it may just be easier to buy a damned D6!!!!

Can''t we just FINALLY creat a machine that can accurately give proportions for fancy shaped stones, instead of using the Sarin Brilliant Eye and this CREATED for rounds...

 

[Rhino]

Here''s an example of what Dave it talking about (but perhaps he may be using a different technology... don''t know).

Both of these examples are of D colors (why do I get the feeling most people wouldn''t care if we were talking about K''s.

)

This first stone, a GIA D tests numerically at a 1.2 (with 1.0 being the highest D).

 

[Rhino]

This one testing as a 1.8. On the colorimeter there is an option to see either *normal* results (like the pix in these posts) or *numerical* results. I kinda like seeing the normal results personally. The numerical results also list a temperature (to which the color is grading too as well).

 

[aljdewey]

Date: 11/4/2004 6:11

3 PM
Author: Greentree
Wow! Look at the poll results. People really do want an even finer gradation than what already exists. Gotta be that the consumer is more interested in measurement and specification. You know, ''My car has more cubic inches than your car''. Never mind how it looks or how it drives.

That''s kind of unfortunate. A diamond should be bought because you really like how it looks and how it handles light. Its specifications should be of mostly academic interest.

Wow, are you ever missing the point! You are grossly exaggerating to suggest that folks who want good data are of the "never mind how it looks or drives" mentality.

Ultimately, people DO select diamonds because they like how it looks or how it handles light. Unortunately, those things don''t tell them anything about what a fair price is for that stone, do they? Of course not.

A diamond''s specifications are WAAAAAAAAAAAAY more than academic.....they directly dictate how much I should pay for it. If an F costs more than an H and you''re telling me that you''re selling me an F, I want to KNOW that it''s an F. I don''t want to buy a lovely diamond at an F price and find out that I got an H stone. It won''t make it any less beautiful to me, but it will certainly affect my satisfaction with the price I paid.

 

[Rhino]

Date: 11/4/2004 8:36

9 PM
Author: aljdewey

Date: 11/4/2004 6:11

3 PM
Author: Greentree
Wow! Look at the poll results. People really do want an even finer gradation than what already exists. Gotta be that the consumer is more interested in measurement and specification. You know, ''My car has more cubic inches than your car''. Never mind how it looks or how it drives.

That''s kind of unfortunate. A diamond should be bought because you really like how it looks and how it handles light. Its specifications should be of mostly academic interest.

Wow, are you ever missing the point! You are grossly exaggerating to suggest that folks who want good data are of the ''never mind how it looks or drives'' mentality.

Ultimately, people DO select diamonds because they like how it looks or how it handles light. Unortunately, those things don''t tell them anything about what a fair price is for that stone, do they? Of course not.

A diamond''s specifications are WAAAAAAAAAAAAY more than academic.....they directly dictate how much I should pay for it. If an F costs more than an H and you''re telling me that you''re selling me an F, I want to KNOW that it''s an F. I don''t want to buy a lovely diamond at an F price and find out that I got an H stone. It won''t make it any less beautiful to me, but it will certainly affect my satisfaction with the price I paid.

Al, you have a talent for hitting the nail squarely on the head.

 

[Diermint]

I realise the accuracy depends a lot as to who has done the grading, but can any experts out there hazard a guess as to the percentage of wrongly-graded stones? Are we talking <1%, 5%, 10%???
I for one always want to get what I''m paying for...

From what has been said already, by using a colorimeter-based approach the accuracy should increase and the number of errors decrease. That is a good thing from my perspective regardless of the number subdivisions in the grading system.

With any measuring device there will always be some error with the very border-line stones due to precision of the instrument but by having a larger number of grades the difference it makes becomes negligible. ie. a H that is labelled an G will make a big difference in price, but a G6 that is labelled a H1 would have less of an impact.

Also, consumers are increasingly buying diamonds over the internet where they must narrow down the search for good stones without actually seeing them. I reckon the more accurate the data, the better.

 

[Rhino]

Hi Diermint,

The percentage of off graded stones, I would say would have to be determined by the lab doing the grading. Ie, I expect a much lower percentage of misgraded stones coming out of the GIA and AGS labs with more coming out of IGI, EGL, etc.

Gary & Leonid''s latest study on the labs however suggest EGL is getting better. Is that right guys?

 

[glitterata]

I think there are two questions here:

1. Would you want a machine that grades colors precisely and repeatably, eliminating human error?
2. Would you want a grading scale with finer distinctions than the ones we have now, with 6 numerical steps replacing one alphabetical step?

If the machine really IS precise and repeatable, eliminating human error, I say absolutely, yes.
As for the second question, I''m not sure. Would these steps be visible to the human eye? If not--if we''re in the realm of invisible differences--then I don''t know. I wouldn''t want to pay more for something I couldn''t see. However, many diamond buyers do want to pay more for something they can''t see--hence buyers of VVS and IF stones.

This brings up a final point. What would those guys do who currently want D IF stones for their perfect brides? Would they be unwilling to settle for 2, 3, 4, 5, and 6 stones, which would currently grade as D but in the new system would be less than perfectly colorless?

 

[kfds]

Why take an unintuitive system (why does it start at D? Why is the color difference between some letters greater than that of others? How do I quatify the difference bwteen a G and an I) and make that system even more granular?

Maybe this is simplistic but...

Why not just assign a percentage value?

For example:
1.5% yellow (or conversely 98.5% white or pure or clear or what ever marketing euphamism that works)

The scale could run from hypotheitcal 0% to to a hypotetical 100%. The trick would be what to base the 100% on. (The most yellow a diamond could theoritically be or a "impossible" solid yellow) Using solid yellow would result in lower overall scores since the high end of the scale really wouldn''t be attainable. (say 80% yellow is the wost actual score since a diamond couldn''t be completely solid yellow).

 

[valeria101]

Date: 11/4/2004 9:14:25 PM
Author: glitterata
I think there are two questions here [...]

[and] many diamond buyers do want to pay more for something they can''t see...

I was desperately trying to split the same to issues one from the other Gliterata...

Anyway, whoever what whiter than D can definitely get it. Just look at Jonathan''s colorimeter screen

AT least a few sellers already offer to select stones based on it beyond GIA''s grid. Ask Fred - he does it. Others do to. What cheaper branding can there be than just making some grid finer ?

 

[oldminer]

People often pay for things they cannot see. Diamonds are an excellent example of paying for qualitythat one cannot possibly appreciate with the fine tuning that already exists in the GIA system. However, people strive for accuracy and clearly distinct categorization. I don''t think finer grades would result in more diamond sales, but possibly finer, more accurate grades would result in happier consumers. If you had a distinct assurance that you were getting what you paid for, are you not more likely to be pleased?

The various methods suggested of providing this more refined system of reporting color grades will be thoroughly considered. Thak you for your participation.

If you have not entered the POLL, please place your vote and make any comments you''d like to add.

 

[canuk-gal]

HI:

It is not my intent to hijack this poll--but I need clarification as to who in the industry would have access to this "proprietory technology" that provides a wider or tigher or repeatable grading system. Everyone who "flogs trickets" (Holloway), only the GIA labs and compliant independant appraisers? Given that there are many uncerted diamonds on the market, I might believe the former--but could this be feasible or even reasonable? Also, will the other labs be expected to adopt new standards (will it be requistite to practice) and how will that be handled? And in the event of technology failure, who will be responsible for the education of the users of the new grading system?

Gosh, let me know if I am missing the point...................

I guess my queries tend toward the huge and very costly issue of program implementation, evaluation and support. I could go on for hours, but it is Friday I really just want to say that many "good" ideas have gone into developing needed programs, without much thought for ongoing education, evaluation, and most of all support to the people affected by change. Upon this subject the research is clear: lack of personnel support=program failure (or at the very least systemic sickness).

cheers

Sharon

 

[Hest88]

I voted yesterday, but didn't comment.

Obviously, I wouldn't mind gradations as finely as we can make them. For consumer purposes, though, 1/6 is too much. I think something closer to 1/3---"High D, Med D, Low D" is more useful. Perhaps the finer evaluations can be reserved for people who request them, along the lines of requests for Sarin report, DiamCalc reports, etc.

 

[nicknomo]

Date: 11/4/2004 9:14:25 PM
Author: glitterata
I think there are two questions here:

1. Would you want a machine that grades colors precisely and repeatably, eliminating human error?
2. Would you want a grading scale with finer distinctions than the ones we have now, with 6 numerical steps replacing one alphabetical step?

If the machine really IS precise and repeatable, eliminating human error, I say absolutely, yes.
As for the second question, I''m not sure. Would these steps be visible to the human eye? If not--if we''re in the realm of invisible differences--then I don''t know. I wouldn''t want to pay more for something I couldn''t see. However, many diamond buyers do want to pay more for something they can''t see--hence buyers of VVS and IF stones.

This brings up a final point. What would those guys do who currently want D IF stones for their perfect brides? Would they be unwilling to settle for 2, 3, 4, 5, and 6 stones, which would currently grade as D but in the new system would be less than perfectly colorless?

I agree, there are obviously two different discussions here, that go hand in hand. A machine that is capable of determining within 1/6 of a grade will make more accurate grading.. It could also cause more precision in the grading system. They are related but separate.

I don''t think anyone is arguing with the precision. However, I think this alone is reason alone to use such a machine.

As far as the grading system, I think this is a more complicated issue. Should the grading system be updated to show more information? I think it could potentially complicate things if implemented poorly.

One poster suggested using a percentage, which I think would actually be more intuitive than the current grading system. When I started, I was confused with the grading letters, as it startes at D, and at face value says nothing about how white the diamond comparitively is. Is H good? Is it bad??? If you had a number, say yellow/color %, you could potentially make things more comprehensible.

Either way, I don''t think we could go wrong implementing machine ratings. We need the accuracy to ensure the consumer gets what he pays for..

 

[diamondlil]

Date: 11/4/2004 2:45

5 PM
Author: strmrdr
I still think were not comunicating :{

What im saying is ignore the current grading and just have a scale with each numder the exact same increase in color.
Then just as a reference create a chart where the gia sytem falls on the new scale.
That way from colorless to yellow 1 number means the exact same amount of color difference.
imho the current gia system with the wider range in color between some letter grades is flawed.
Splitting up a flawed system is still flawed.

I understand, Strmrdr, and completely agree. I never have understood why there are wider ranges in color for lower color grades. It would make more sense to have each grade (whether it be numbers or letters) be equal increases in color from beginning to end.

DiamondLil

 

[perry]

Great Question, and lots of good comments.

My $0.02 is that going to machine grading with a proprietary machine (or at least a proprietary light source) is probably the only way to go. This is not an unusual approach in the world of "Standards." For example. Gasoline "Octane" is determined in test in a a specially constructed single cylinder engine manufactured in Waukesha Wisconsin, where the design of this engine was fixed something like 60 years ago. Every oil refinery and test lab in the world - even during the cold war - and the "bitterest" enimies of the US uses the same test engine. Thus, Octane ratings of gasoline are the same worldwide. Quite honestly, the supplier dosn''t sell many new engines now - but lots of rebuild parts.

I know that there are other examples of that: where one specific technology is chosen and maintained forever - but the Octane engine pops to mind as I had a ASME group tour there a few years ago.

However, I do not think that breaking each existing grade into 6 groups is the best thing.

The pattern I know exist in some Non Destructive Examination standards (looking for flaws in metal of composit assemblies) would more detail where it counts, and less where it does not.

For example; you might break D-G into 3 segments, H-k into 2, and leave L-Z alone. This would give a totall of 35 differnt catagories assuming my mental arrithmatec is accurate (not worth my time to write it down and figure it out exactly). Different scenios exist, but I think that 50 total catagories would be too much.

I hope that you can get a bunch of people together to form the right committee and implement this. Be sure that there is an adequate and accurate definintion of the light source.

Perry

 

[aljdewey]

Date: 11/5/2004 12

7:45 PM
Author: Hest88
I voted yesterday, but didn''t comment.

I wouldn''t mind gradations as finely as we can make them. For consumer purposes, though, 1/6 is too much. I think something closer to 1/3---''High D, Med D, Low D'' is more useful.

I honestly don''t even see how breaking it into High D, Medium D, Low D is going to be helpful if such designations don''t reflect a difference in price.

I frankly think that it would be enough of an improvement if machine grading could remove subjectivity to the point that a G stone would be considered a G stone anywhere.....instead of "I call it an H, you say it''s a G". Right now, the variance can be one to two grades off. If we could improve that to a variance of no grades through machine grading, I see that as sufficient.

Anything more than that is overkill. The biggest problem with consumer confidence in grading isn''t because there aren''t enough "steps" of any given color (high D, medium D, low D).....rather, it''s that one cannot reasonably be assured that a stone is the color it''s represented to be.

I believe there is no point in breaking an individual color grade (say an H) into H1, H2, H3, H4, H5, H6 unless that designation comes with benefits/repercussions (meaning an adjustment to the price.)

If the industry intends to price all G stones the same, there is no point in the demarcation between a G1 and a G4. As Rhino pointed out, doing so with no price implication could lead vendors being stuck with the arbitrary lower tiers in each category (consumer rationale would say "why would I want to buy a G6 if I could get a G1 for the same money?).

 

[Nicrez]

I totally agree with Al...wow...

I like the idea of the precision of the technology, but I think the further classification is absolutely ridiculous. It''s actually cumbersome and unnecessary. If that''s the case, there are so many shades of EVERY color. WOuld hue, saturation and tone also be quantified. Imagine THAT daunting report.

I say if it''s possible, transition a machine to grade the color, instead of the human eye, OK. Sarin did it. Electronic scales did it. Quantifying color is OK, and may end up being the new way to grade with PRECISION, based IMO solely on the need to benefit the consumer on a PRICE level only. Buy a G for $3000, get a G, and not an H for $3000. Fair enough.

I suppose it''s like pedigrees for dogs. Each one has a personality, and just try finding a specific breed with specific characteristics, like one droopy ear, and a tail that twitches to the left, with one brown paw and the other white. But then again, in the human need to impose structure on chaos, many people shop for dogs on paper (breeding and lineage), instead of just asking, "How much is that puppy in the window" and just loving it for what it is...

 

[hoorray]

I actually agree with both Strm and Al!

I think if you are gong to introduce more granularity, it would be better to do it in a manner such as Strm suggested without trying to map it back to the existing system. (You''d probably have to map it for a while, until it bacame understood and well integrated into the selling/pricing process.) I like the linear granularity it could provide if the grades were defined appropriately. It is easy to understand a linear scale, and be able to decide a range that would be your target range when shopping, without having to understand the differences in how the scale works at different points, such as the current problem that there is more difference between an H and and I than a E and a F. This makes the current scale non-linear, and harder to make sense of for consumers who don''t have lots of viewing experience.


I don''t think it makes sense to try to differentiate the differences between a G3 and a G6 tho. The non-linear issues discussed above still exist, and I think it just makes a somewhat confusing system even more confusing. Maybe 3 categories per grade would be more understandable to cover the high, medium and low issues, but pricing would have to vary accordingly, and you really have to feel that there is some benefit to the consumer of differentiating between these grades.


I think the bigger benefit is the reliability of the color grading since it is one of the more subjective items today and has a big impact on pricing. I''m excited to hear more about it.

 

[aljdewey]

Date: 11/5/2004 6

9:26 PM
Author: Nicrez
I totally agree with Al...wow...

HAHHAHAHAH --- a sure sign that Armageddon is indeed upon us!


(kidding Nic.....I knew you''d come over to the dark side....

)

 

[noobie]

Date: 11/6/2004 11:20:52 AM
Author: aljdewey

Date: 11/5/2004 6

9:26 PM

Author: Nicrez

I totally agree with Al...wow...

HAHHAHAHAH --- a sure sign that Armageddon is indeed upon us!





(kidding Nic.....I knew you'd come over to the dark side....

)

After reading this thread, I ahve to sheepishly agree with Al too. Oh well, it's usually more fun to debate with her.

 

[Jennifer5973]

OK--I''ve got my bulletproof vest on....

I voted "no." I am all for protecting the consumer but at some point, all this minutia sucks the joy right out of it. If there is a way to protet the consumer better since color is tied to price, then great. But six shades with one color band? I have a headache already.


I think Nicrez nailed it ... and how about instead of just aksing for that puppy in the window, going to the local pound and getting a dog that is just as cute and cuddly for a $50 donation? In the end, you''ll have the same result and maybe even feel better about the whole thing anyway.

 

[valeria101]

This thread turned out a bit of a puzzle to me...

The original question does not mention multiplying the grades, but just making grading more scientific, acurate, repeatable and... machine-made. No wonder most votes went pro. How could one opt againt scientific precision for anything ?

And then further explanation by David opened the can on multiple grades, and the discussion ended up much more diverse than the results of the poll suggests.


Do you guys agree that there is need for another thread asking "would more grades be better for you" ? The two topics hardly sounds equivalent up to this point.

 

[Richard Sherwood]

ith the common machine colorimeter gradiing available to this point, that the human eye has proven to be more accurate. (ie, human eye more accurate than Gran colorimeter).

Is the machine grading you're proposing more accurate than the human eye, Dave? How does it compare to Marty Haske's SAS-2000 spectrophotometer grading?

 

[aljdewey]

Date: 11/6/2004 11

8:47 PM
Author: valeria101

This thread turned out a bit of a puzzle to me...

The original question does not mention multiplying the grades, but just making grading more scientific, acurate, repeatable and... machine-made.

It did mention it indirectly. The original query contained a statement that the machine grading would entail precision to within 1/6 of a stone's present grade .... which implies that the machine could actually split a grade into six different levels.


Consequently, no, I don't think it's necessary to generate a separate thread on breaking colors into more subsets. It seems germane directly within this thread to me.