Letter to the Editor of the Australian Gemmologist

[strmrdr]

The GG is the MSCE of the diamond world.
You have the background for learning and maybe the tools but the education has just begun.
And sadly a high percentage have the attitude that they know everything too so they dont learn much.


padparashah, would you give a person who just got their GG yesterday with no experience 1 million dollars and tell them to go buy diamonds for you in say Israel (as an example)?

 

[michaelgem]

Date: 3/6/2007 2:44:25 AM
Author: Serg

Re: GIA :” ..What is not adequately mentioned by the authors is that there is much overlap in the cut grades from the two systems…”

I saw big difference between AGS0 2005 chart and GIA Excellent.
I do not see principal difference between ASG0 PGS 2006 and GIA Excellent now.
AGS0 and GIA Ex are too similar now.

There is huge difference between AGS 2005 charts and AGS PGS 2006 grade.

Serg,

Your observations should be addressed by AGS, and I second them with the hope that they will be better noticed and receive the attention they deserve.

Folks on Pricescope believe AGS is more open and forthcoming about its grading system than GIA. Lately, this no longer appears to be the case. AGS should return to its policy of openness.

The easiest and sufficient way is to update the AGS 2005 charts with the AGS PGS 2006 grades.

Michael Cowing

 

[Serg]

Date: 3/6/2007 1:40:39 PM
Author: michaelgem

Date: 3/6/2007 2:44:25 AM
Author: Serg

Re: GIA :” ..What is not adequately mentioned by the authors is that there is much overlap in the cut grades from the two systems…”

I saw big difference between AGS0 2005 chart and GIA Excellent.
I do not see principal difference between ASG0 PGS 2006 and GIA Excellent now.
AGS0 and GIA Ex are too similar now.


There is huge difference between AGS 2005 charts and AGS PGS 2006 grade.

Serg,

Your observations should be addressed by AGS, and I second them with the hope that they will be better noticed and receive the attention they deserve.

Folks on Pricescope believe AGS is more open and forthcoming about its grading system than GIA. Lately, this no longer appears to be the case. AGS should return to its policy of openness.

The easiest and sufficient way is to update the AGS 2005 charts with the AGS PGS 2006 grades.

Michael Cowing

re:The easiest and sufficient way is to update the AGS 2005 charts with the AGS PGS 2006 grades.

Hi Michael,

See http://oct.gemology.ru/oct/mss/

Second chart with AGS 2005 guideline
Third chart with AGS 2006 guideline
Fourth chart with AGS-PGS data.
Results and reasons are more a less clear

 

[Serg]

on AGS cutting guideline chart 2006 Pav40 Cr39 has grade even better than in PGS software.
How many times did AGS change new cut grade software?

 

[Serg]

How have PGS, please check point P40Cr39T57 . I hope it was missprint

Date: 3/6/2007 3:13:05 PM
Author: Serg
on AGS cutting guideline chart 2006 Pav40 Cr39 has grade even better than in PGS software.
How many times did AGS change new cut grade software?

 

[adamasgem]

Date: 3/6/2007 3:13:05 PM
Author: Serg
on AGS cutting guideline chart 2006 Pav40 Cr39 has grade even better than in PGS software.
How many times did AGS change new cut grade software?

Sergey..
1) The guideline charts summed the Ceiling function of each of the four parameters considered in the performance analysis, as a conservative guidline, assuming perfect symmetry, to get the GUIDELINE i.e., if there were a deduction of 0.1, the guidilines would indicate

2) The PGS software ROUNDS, to the nearest integer, the sum of the four parameter deductions, to get the GRADE, so a total deduction of 0.49 would still get a 0 grade, if there were no assymetries that increased the deduction, which is HIGHLY unlikely

3) GIA averages the parameters, and then ROUNDS to get their input point, to establish the GRADE

4) PGS uses the components of each parameter, i.e. the actual scan, to establish a final GRADE

In the GIA system, a stone with only VG symmetry, could still get a EX cut grade

In the AGS system, the same VG symmetry stone, would probably get substantial light performance deductions, AND would be limited in GRADE by the VG symmetry call, not to mention less than EX Polish

You are trying to fit a square peg in a round hole..

I did a database study of close to 70,000 round brilliants with GIA paper that listed symmetry and polish, from Polygon, which is probably a better overall representation than the better stones typically seen on PriceScope.

One only has to look at the relative verbal definitions for polish and symmetry of GIA''s five grades for polish and symmetry, versus AGS''s 0 to 10, to realize that there are going to be substantial ADDITIONAL deductions applied., and if the same 70000 stones were graded by AGS, that PERHAPS the stones with both GIA EX for polish and symmetry would get an AGS 0, while we KNOW that those stones pGSwould get an GIA EX for cut if they were inside the EX cut paramater boundaries.

What you are trying to do, is admirable, but doesn''t work.

You need to run the actual stone on PGS..

What did the stones you actually cut get basedd on the scans, properly graded for polish and symmetry..

GIA EX cut grade contains AGS 0 to 4''s at the least, in the real world.

 

[Serg]

re:You are trying to fit a square peg in a round hole..

Marty,

Where is hole? What is peg?
Your post has not ANY connections to my posts.
Any explanation of problems is absent in your post.
If you do not want to see problem it does not mean what I use square peg for round hole.
What are you doing?

 

[Serg]

re:1) The guideline charts summed the Ceiling function of each of the four parameters considered in the performance analysis, as a conservative guidline, assuming perfect symmetry, to get the GUIDELINE i.e., if there were a deduction of 0.1, the guidilines would indicate

Marty,
please receive AGS guideline grade 2 from PGS AGS grade 3 using this AGS rule for T 57Cr39P40
See full PGS report.
Again , Could anybody check same diamond in PGS software?

 

[Serg]

Other issue, Very strange brightness column( first column) for Cr 39.
I did not see correlation with DC Light return.

For example what is reason for very big jump between
Pavilion 40.2( Brightness=1.56) and
Pavilion 40( Brightness=0.68)
?

 

[padparashah]

Date: 3/6/2007 6:38:28 AM
Author: strmrdr
The GG is the MSCE of the diamond world.
You have the background for learning and maybe the tools but the education has just begun.
And sadly a high percentage have the attitude that they know everything too so they dont learn much.


padparashah, would you give a person who just got their GG yesterday with no experience 1 million dollars and tell them to go buy diamonds for you in say Israel (as an example)?

Dear Storm ,

Thank you for the question .
In regards to your question , I find it difficult to answer .

The Diploma is a Foundation for learning the correct way .

But the learning needs to continue. My Israeli lecturers, Abba & Brian keep telling us from day 1.
I do not know where the idea that a GG is the end of all learning came from .
I do not think GIA preaches any sort of such belief .

Ultimately its the same for any professionals . eg: Doctors .
Doctors have to keep up with science .
Doctors have to be familiar what are the common infections .
And doctors , as they grow older , are more expensive becos of their experience.
its the same principle.

As to your question of whether i wld give a fresh grad $1 million dollars to go to Israel to buy diamonds ?
I guess you must be speaking from this point of view because you have done that .
Alone . then that''s very good for you .

In any jewelry organisation , as what i know , team work is paramount .
Its not so much of whether u are a GG or not .
You need a team to synchronize and get things done.
So to ur question , i believe in a team not a single person.
Such a journey requires very advanced security measures.

So i would put somebody who is familiar with the buying process to partner my new staff . GG or not .
Your question brings out too many issues like whether i actually even trust him with my $1000,000.
Frankly i wont trust anyone with my money......its not about the education.
its about his integrity .
If your boss has ever given u that kinda money to go use ....u r great ....

warmest regards , dave

 

[strmrdr]

padparashah,
Thanks for taking the time to answer.
I''m not in the trade, computers are my thing.
While Iv done $100000 deals from the sales side cant say iv done a million.

As a consumer Iv ran into a lot of GG''s and non-GG''s in the trade that know far less about diamonds than I do and a lot of the GG''s have had an attitude that Im a GG I know everything.
Which reminds me a lot of msce''s in the computer field.

 

[adamasgem]

Date: 3/7/2007 4:03:39 AM
Author: Serg



re:1) The guideline charts summed the Ceiling function of each of the four parameters considered in the performance analysis, as a conservative guidline, assuming perfect symmetry, to get the GUIDELINE i.e., if there were a deduction of 0.1, the guidilines would indicate

Marty,

please receive AGS guideline grade 2 from PGS AGS grade 3 using this AGS rule for T 57Cr39P40
See full PGS report.
Again , Could anybody check same diamond in PGS software?

Sergey.. Please send the STL file you used to Jason and I

1) The guideline charts you show (the one without the AGS copyright on them ) were generated prior to the generation of the PGS software using the Ceiling Function on the first four elements, I believe.. As I understand, they were generated using DiamondCalc STL''s at that time, and were unofficial as a courtesy, I believe, prior to the issuance of the official charts withe the AGS copyright on them..

2) The guideline charts were only for light performance, using the Ceiling function and would treat an individual deduction of 0.9, 0.1, 0.01, 0.004 as the same... i.e. a deduction of 1

3) The output you show is from the PGS software, issued probably 2 years later.. It sums all the deductions and then rounds to get the light performance GRADE or 1.. Its individual light performance metrics are rounded to two decimal places.. and would show a 0.004 deduction as 0.00..

NOW PLEASE NOTE, BECAUSE I ONLY WANT TO SAY THIS ONCE, and my letter sises will get bigger
l
The Guideline charts, TO MY Knowledge, DID NOT INCLUDE ANY PROPORTION DEDUCTIONS, and WE KNOW, and can SHOW, that certain DC STL files, have had problems, because of unanticipated tiny facets, generated by STL generation software, that software maybe using double precision arithmatic internally, and output single precision STL''s. SRN files also had the same type of tiny facet problems.

It might just be a generic type of problem in trying to make a closed form single precision vertex set. When do you treat two vertices as the same when you either generate it or read it into another set of software, and we see it in scans all the time, when we examine scanner output that apparently doesn''t show meet point faceting, or conversion from one form to another. If you read in an ascii format STL into DiamondCalc and then export a binary STL file, YOU WILL GET TWO DIFFERENT VERTEX SETS, PERIOD. WE WENT OVER THIS IN ANOTHER THREAD. You are not alone with this problem.

So there may have been a light performance component that had a small 0.004 light performance deduction of an additional Ceiling function componet of 1.

The PGS output you show, based on STL''s , may or may not have "tiny facet" problems, however it shows a proportion deduction of 2 giving a GRADE of 3.

I do not have the insight into what is going on with the STL file unless you send the one that you used.

You are trying to use the Guideline charts for a purpose they were not intended for...

They were warning cutters away from proportion sets that were starting to loose light performance and were NOT grading charts...

 

[strmrdr]

Date: 3/7/2007 12:11:01 PM
Author: adamasgem
They were warning cutters away from proportion sets that were starting to loose light performance and were NOT grading charts...

That should be Marty sized.

 

[adamasgem]

Date: 3/7/2007 4:18:33 AM
Author: Serg
Other issue, Very strange brightness column( first column) for Cr 39.
I did not see correlation with DC Light return.

For example what is reason for very big jump between
Pavilion 40.2( Brightness=1.56) and
Pavilion 40( Brightness=0.68)
?

I don't know why you don't see a correlation with DC results and PGS, but perhaps it is because PGS analyzes in both the face up position and at a 15 degree tilt, and DiamondCalc does something different. In fact isn't there an additional "tilt" deduction also.

DC obviously uses a different metric and environment..

GIA uses only the faceup for brilliance.

So there are going to be differences, but my guess is the tilted component of the metric may "jump" for that proportion set, and the stone suddenly get darker, just as we discussed at quite some length two years ago when I presented the azimuth and tilt study I did here on Pricescope along with our sharing our chromatic flare comparisons.

No simple answer....

You could look at the 15 degree tilt DiamondCalc photoreal representation, just as Gary showed years ago to support my data hypotheses of stones going dark at certain tilt angles.

Your software is, and was, the best tool around for showing this effect..

 

[Serg]

Date: 3/7/2007 12:11:01 PM
Author: adamasgem

Date: 3/7/2007 4:03:39 AM
Author: Serg




re:1) The guideline charts summed the Ceiling function of each of the four parameters considered in the performance analysis, as a conservative guidline, assuming perfect symmetry, to get the GUIDELINE i.e., if there were a deduction of 0.1, the guidilines would indicate

Marty,


please receive AGS guideline grade 2 from PGS AGS grade 3 using this AGS rule for T 57Cr39P40
See full PGS report.
Again , Could anybody check same diamond in PGS software?

Sergey.. Please send the STL file you used to Jason and I

1) The guideline charts you show (the one without the AGS copyright on them ) were generated prior to the generation of the PGS software using the Ceiling Function on the first four elements, I believe.. As I understand, they were generated using DiamondCalc STL''s at that time, and were unofficial as a courtesy, I believe, prior to the issuance of the official charts withe the AGS copyright on them..


Strange answer

2) The guideline charts were only for light performance, using the Ceiling function and would treat an individual deduction of 0.9, 0.1, 0.01, 0.004 as the same... i.e. a deduction of 1

3) The output you show is from the PGS software, issued probably 2 years later.. It sums all the deductions and then rounds to get the light performance GRADE or 1.. Its individual light performance metrics are rounded to two decimal places.. and would show a 0.004 deduction as 0.00..

NOW PLEASE NOTE, BECAUSE I ONLY WANT TO SAY THIS ONCE, and my letter sises will get bigger
l
The Guideline charts, TO MY Knowledge, DID NOT INCLUDE ANY PROPORTION DEDUCTIONS,

very interesting. Are you sure? What is value Such GUIDELINE CHARTS .



and WE KNOW, and can SHOW, that certain DC STL files, have had problems, because of unanticipated tiny facets, generated by STL generation software, that software maybe using double precision arithmatic internally, and output single precision STL''s. SRN files also had the same type of tiny facet problems.

Does PGS have any problem with DC binary STL ?

It might just be a generic type of problem in trying to make a closed form single precision vertex set. When do you treat two vertices as the same when you either generate it or read it into another set of software, and we see it in scans all the time, when we examine scanner output that apparently doesn''t show meet point faceting, or conversion from one form to another. If you read in an ascii format STL into DiamondCalc and then export a binary STL file, YOU WILL GET TWO DIFFERENT VERTEX SETS, PERIOD. WE WENT OVER THIS IN ANOTHER THREAD. You are not alone with this problem.

We did not use ASCII STL for work with PGS. You do not need use it too. Do you have better explanations

So there may have been a light performance component that had a small 0.004 light performance deduction of an additional Ceiling function componet of 1.

Did you realy understand problem?

The PGS output you show, based on STL''s , may or may not have ''tiny facet'' problems, however it shows a proportion deduction of 2 giving a GRADE of 3.

Did you realy understand problem? PGS has grade 3, Guideline has grade 2.

I do not have the insight into what is going on with the STL file unless you send the one that you used.

You are trying to use the Guideline charts for a purpose they were not intended for...

They were warning cutters away from proportion sets that were starting to loose light performance and were NOT grading charts...

re:Sergey.. Please send the STL file you used to Jason and I

Marty,

For full checking It should be done independently from Cut Group work.
You and Jason can do it easy. . Do you remember your answer to Michael?
I am not work in AGS

 

[adamasgem]

Date: 3/7/2007 3:08:49 PM
Author: Serg

Date: 3/7/2007 12:11:01 PM
Author: adamasgem


Sergey.. Please send the STL file you used to Jason and I

1) The guideline charts you show (the one without the AGS copyright on them ) were generated prior to the generation of the PGS software using the Ceiling Function on the first four elements, I believe.. As I understand, they were generated using DiamondCalc STL''s at that time, and were unofficial as a courtesy, I believe, prior to the issuance of the official charts withe the AGS copyright on them..


Strange answer Not so strange, the guideline charts also may have included a weight deduction and a tilt deduction.. They evidently, at first used DiamondCal to put together the STL files, and then wrote their own STL generator to automate the process for the later charts.

2) The guideline charts were only for light performance, using the Ceiling function and would treat an individual deduction of 0.9, 0.1, 0.01, 0.004 as the same... i.e. a deduction of 1

Correction, weight and tilt were evidently factored in.

3) The output you show is from the PGS software, issued probably 2 years later.. It sums all the deductions and then rounds to get the light performance GRADE or 1.. Its individual light performance metrics are rounded to two decimal places.. and would show a 0.004 deduction as 0.00..

NOW PLEASE NOTE, BECAUSE I ONLY WANT TO SAY THIS ONCE, and my letter sises will get bigger
l
The Guideline charts, TO MY Knowledge, DID NOT INCLUDE ANY PROPORTION DEDUCTIONS,

very interesting. Are you sure? What is value Such GUIDELINE CHARTS . Sergey, what don''t you unerstand about having cutters avoid the edges of cliffs. Some cutters try to hit averages, and the averages are right in the center of the 0, and can''t get a zero because of asymmetries in the cutting. So they learn how to cut with better symmetry.



and WE KNOW, and can SHOW, that certain DC STL files, have had problems, because of unanticipated tiny facets, generated by STL generation software, that software maybe using double precision arithmatic internally, and output single precision STL''s. SRN files also had the same type of tiny facet problems.

Does PGS have any problem with DC binary STL ? Sergey,I''ve said it before, PGS, at the current time, only uses binary STL''s, we discovered the conversion problem in DC in the other thread when we got different results from the converted (ascii to binary) files, and got tiny facets whose normals are all over the place. You would probably get that difference problem with any conversion, like SRN to STL or ASC to STL.

It might just be a generic type of problem in trying to make a closed form single precision vertex set. When do you treat two vertices as the same when you either generate it or read it into another set of software, and we see it in scans all the time, when we examine scanner output that apparently doesn''t show meet point faceting, or conversion from one form to another. If you read in an ascii format STL into DiamondCalc and then export a binary STL file, YOU WILL GET TWO DIFFERENT VERTEX SETS, PERIOD. WE WENT OVER THIS IN ANOTHER THREAD. You are not alone with this problem.

We did not use ASCII STL for work with PGS. You do not need use it too. Do you have better explanations

Sergey, for the 10th time, when you asked for 4 PGS runs, I generated the ascii STL''s and sent it to RockDoc who found out that PGS doesn''t use ascii files, and then converted the ascii STL''s to binary STL''s, using DiamondCalc, so that PGS could read them, becuse it can only read binary files. And PGS had problems because DiamondCalc ADDED tiny facets in the conversion. WHAT DON''T you understand? ANY conversion, from ANY software, might have similar problems because of reading in single precision numbers, processing them with double precision arithmatic, and then exporting them as a single precision file. It is a fact of life.

So there may have been a light performance component that had a small 0.004 light performance deduction of an additional Ceiling function componet of 1.

Did you realy understand problem? I understand what you are trying to do with published data tha was never meant to do the comparisons. I''ve made the same misinterpretation at times.

The PGS output you show, based on STL''s , may or may not have ''tiny facet'' problems, however it shows a proportion deduction of 2 giving a GRADE of 3.

Did you realy understand problem? PGS has grade 3, Guideline has grade 2. Yes, I understand your "concern", but you may have a stone on a borderline with a deduction not included in the guidelines, and arithmatic accuracy may factor in.

I do not have the insight into what is going on with the STL file unless you send the one that you used.

You are trying to use the Guideline charts for a purpose they were not intended for...

They were warning cutters away from proportion sets that were starting to loose light performance and were NOT grading charts...

re:Sergey.. Please send the STL file you used to Jason and I

Marty,

For full checking It should be done independently from Cut Group work.
You and Jason can do it easy. . Do you remember your answer to Michael?
I am not work in AGS Understood, and AGS does not work for Octonus, to my lnowledge. I asked for the STL file that you used because there may be differences in math or output between versions of DiamondCalc code that I or AGS do not have. My request eliminates a variable to try to understand and analyse your problem. There are borderline thresholds which may be numerically sensitive, just like uncertainties in scans on borderline stones may give different results. Remember the differences in Helium/Sarin/Ogi scans on the same stone.

You might be interested that I have been lobbying for the addition of confidence bounds on all types grades issued by all labs, and more explicit "grades", without the Rounding.

It is the same type of issue that arises when GIA uses the International weight standard and rounds up to two decimal places only if the weight in the third decimal place is a 9 , where AGS puts the weight to three decimal places.

In the US a 0.995ct stone can be legitimately and legally called a 1.00 carat stone, while the bourse (and GIA) requires a stone to be 0.999ct before it can be called a 1.00ct stone.

I''m not saying that I disagree with GIA on that, but I do agree with AGS in putting it out to 3 decimal places.

I do disagree with GIA''s half-assed rounding scheme based on averages though, a rounding scheme unlike normal arithmetic everyone is taught.

 

[Serg]

re:Sergey, what don''t you unerstand about having cutters avoid the edges of cliffs. Some cutters try to hit averages, and the averages are right in the center of the 0, and can''t get a zero because of asymmetries in the cutting. So they learn how to cut with better symmetry.

Marty,

Let''s me say what I am much better know what cutters need and what they can do.
You and AGS are much more far from cutters technologies than I .

If you do not like see my statements, It is fine for me. Be happy.
I will continue this my work in any case .

 

[padparashah]

Date: 3/7/2007 11:58:54 AM
Author: strmrdr
padparashah,
Thanks for taking the time to answer.
I''m not in the trade, computers are my thing.
While Iv done $100000 deals from the sales side cant say iv done a million.

As a consumer Iv ran into a lot of GG''s and non-GG''s in the trade that know far less about diamonds than I do and a lot of the GG''s have had an attitude that Im a GG I know everything.
Which reminds me a lot of msce''s in the computer field.

Dear Storm,

Thank you for the reply .
At least now i am aware of your amount of knowledge and experience.
You are truly great.
Pardon me for my ignorance.

warmest regards , dave

 

[adamasgem]

Date: 3/7/2007 4:58:59 PM
Author: Serg


re:Sergey, what don''t you unerstand about having cutters avoid the edges of cliffs. Some cutters try to hit averages, and the averages are right in the center of the 0, and can''t get a zero because of asymmetries in the cutting. So they learn how to cut with better symmetry.

Marty,

Let''s me say what I am much better know what cutters need and what they can do.
You and AGS are much more far from cutters technologies than I .

Well Sergey, the statistics that I posted showing that only about 14% of GIA graded RBC''s got an EX/EX for polish and symmetry tells me, as far away from the cutters you may think I am, that they, the cutters, have a lot to learn. Hope you can help them with the basics..

If you do not like see my statements, It is fine for me. Be happy.
I will continue this my work in any case .

 

[Serg]

Date: 3/8/2007 2:19:54 AM
Author: adamasgem

Date: 3/7/2007 4:58:59 PM
Author: Serg


re:Sergey, what don''t you unerstand about having cutters avoid the edges of cliffs. Some cutters try to hit averages, and the averages are right in the center of the 0, and can''t get a zero because of asymmetries in the cutting. So they learn how to cut with better symmetry.

Marty,

Let''s me say what I am much better know what cutters need and what they can do.
You and AGS are much more far from cutters technologies than I .

Well Sergey, the statistics that I posted showing that only about 14% of GIA graded RBC''s got an EX/EX for polish and symmetry tells me, as far away from the cutters you may think I am, that they, the cutters, have a lot to learn. Hope you can help them with the basics..

If you do not like see my statements, It is fine for me. Be happy.
I will continue this my work in any case .

1) 14% is not bad if you account what not all Eightstar diamonds received EX/EX
2) 5-20% market niche is good target for Excellent products. It is silly do one quality product( Do you realy think what market need only AGS0 and GIA EX?)
3) GIGO

 

[michaelgem]

This post explains why it is reasonable to compare the AGS and GIA grading systems using parametric grading charts. It supports the case for updating the cut grade guideline/estimation charts so that the AGS-PGS Ideal 0 range is correctly represented for the cutters and consumers.

Some feel that you can’t reasonably compare GIA’s top grade to AGS’s because one system is parametric and the other non-parametric. I called this assertion a red herring. They can be fairly compared. Here is how and why:

The AGS guideline charts are a parametric representation of their grading system for the mathematically symmetric case. So both AGS’s and GIA’s charts are parametric representations of their grading systems. Though GIA’s system is represented by parametric grading charts, it was developed from a statistical analysis of nonparametric data. So in that sense both are nonparametric systems represented by these parametric grading charts.

For this symmetric parametric case, both GIA and AGS methodologies reduce to be quite similar.

1. Both measure the diamond''s angles and proportions with a Sarin type device.

2. Since the diamond is perfectly symmetrical, GIA''s averaging is not a difference, since the parameters averaged are all the same. (Rounding vs. no rounding is a difference that can be side stepped by judicious choice of, for instance, 80% half-length)

3. Both then model or have modeled the diamond from this similar Sarin data.

4. Both then apply their software metrics (or look up the stored application of their software metrics) to the diamond model to arrive at a grade.

Using these parametric charts of idealized diamonds to compare the two systems is instructive and reasonable, because it is comparing apples to apples.

Michael Cowing

 

[Garry H (Cut Nut)]

Marty Sergey is right. if you or I were cutters who were given a De Beers type allocation there would be rough that needs to not be perfect in order for the bank and the other stake holders to live.

People like Paul buy only the rough they hope to make magic #''s with, and they only have a market for these goods.

People like the rest of us leave the not so goods on the table as rejections.

And they need to find a home.

What is amazing is that GIA sym and polish standards were set soooo long ago and technology has improved soo much that there is an imbalalnce in the # of poor, fair and good''s.

Maybe the bar needs to be set higher so only 3% could be ex ex or AGS0? Maybe that is about right for fine wine, hotel beds, cars etc?

And on all these fronts Helium Polish is being used by manufacturers to improve standards in ways not previously possible.

 

[Garry H (Cut Nut)]

Date: 3/8/2007 10:14:44 AM
Author: michaelgem

This post explains why it is reasonable to compare the AGS and GIA grading systems using parametric grading charts. It supports the case for updating the cut grade guideline/estimation charts so that the AGS-PGS Ideal 0 range is correctly represented for the cutters and consumers.

Some feel that you can’t reasonably compare GIA’s top grade to AGS’s because one system is parametric and the other non-parametric. I called this assertion a red herring. They can be fairly compared. Here is how and why:

The AGS guideline charts are a parametric representation of their grading system for the mathematically symmetric case. So both AGS’s and GIA’s charts are parametric representations of their grading systems. Though GIA’s system is represented by parametric grading charts, it was developed from a statistical analysis of nonparametric data. So in that sense both are nonparametric systems represented by these parametric grading charts.

For this symmetric parametric case, both GIA and AGS methodologies reduce to be quite similar.

1. Both measure the diamond''s angles and proportions with a Sarin type device.

2. Since the diamond is perfectly symmetrical, GIA''s averaging is not a difference, since the parameters averaged are all the same. (Rounding vs. no rounding is a difference that can be side stepped by judicious choice of, for instance, 80% half-length)

3. Both then model or have modeled the diamond from this similar Sarin data.

4. Both then apply their software metrics (or look up the stored application of their software metrics) to the diamond model to arrive at a grade.

Using these parametric charts of idealized diamonds to compare the two systems is instructive and reasonable, because it is comparing apples to apples.

Michael Cowing

You are joking Michael?

Have you read our exchange of letters?

A grading system based on 70,000 judgements in a lighting environment where if a badly cut princess cut was included, the princess cut would have got the best brightness score.

Someone asked who was representing GIA on Pricescope?

 

[michaelgem]

re:The easiest and sufficient way is to update the AGS 2005 charts with the AGS PGS 2006 grades.

Hi Michael,

See http://oct.gemology.ru/oct/mss/

Second chart with AGS 2005 guideline
Third chart with AGS 2006 guideline
Fourth chart with AGS-PGS data.
Results and reasons are more or less clear
How many times did AGS change new cut grade software?
Sergey Sivovolenko

Hi Sergey,

AGS has stated that their goal is “to refine the AGS Ideal 0 using the most advanced diamond cutting knowledge and science of the day.” This refinement has been on going until it has now resulted in the AGS-PGS software. AGS appears quite satisfied with this software and the current state of the range of Ideal 0.

Until the AGS issues new charts reflecting the current PGS range of Ideal 0, your cut groups chart work is the best reference as to how that range has grown from its original extent shown in the guideline charts. I enclose a superposition of the cut groups charts to examine how that range has grown.

The yellow is the Ideal 0 range of the current guideline charts, and the blue is the Ideal 0 range the cut group obtained from the PGS software. The first thing I notice, besides the large increase in Ideal 0, is that the growth is not symmetrical.

The current guideline charts don’t seem to be the best aid to the cutters. Notice that on the shallow/shallow side the guidelines have the cutter on the edge of falling out of the zero, while there is greater room for error in the steep/deep direction. The cutter is also not seeing in the guideline charts the large area of zero’s in the shallow/deep direction.

If any of this analysis is wrong or misleading I anticipate that AGS will provide clarification and correction.

Michael Cowing

 

[Serg]

Date: 3/8/2007 10:22:45 AM
Author: Garry H (Cut Nut)

Date: 3/8/2007 10:14:44 AM
Author: michaelgem

This post explains why it is reasonable to compare the AGS and GIA grading systems using parametric grading charts. It supports the case for updating the cut grade guideline/estimation charts so that the AGS-PGS Ideal 0 range is correctly represented for the cutters and consumers.

Some feel that you can’t reasonably compare GIA’s top grade to AGS’s because one system is parametric and the other non-parametric. I called this assertion a red herring. They can be fairly compared. Here is how and why:

The AGS guideline charts are a parametric representation of their grading system for the mathematically symmetric case. So both AGS’s and GIA’s charts are parametric representations of their grading systems. Though GIA’s system is represented by parametric grading charts, it was developed from a statistical analysis of nonparametric data. So in that sense both are nonparametric systems represented by these parametric grading charts.

For this symmetric parametric case, both GIA and AGS methodologies reduce to be quite similar.

1. Both measure the diamond's angles and proportions with a Sarin type device.

2. Since the diamond is perfectly symmetrical, GIA's averaging is not a difference, since the parameters averaged are all the same. (Rounding vs. no rounding is a difference that can be side stepped by judicious choice of, for instance, 80% half-length)

3. Both then model or have modeled the diamond from this similar Sarin data.

4. Both then apply their software metrics (or look up the stored application of their software metrics) to the diamond model to arrive at a grade.

Using these parametric charts of idealized diamonds to compare the two systems is instructive and reasonable, because it is comparing apples to apples.

Michael Cowing

You are joking Michael?

Have you read our exchange of letters?

A grading system based on 70,000 judgements in a lighting environment where if a badly cut princess cut was included, the princess cut would have got the best brightness score.

Someone asked who was representing GIA on Pricescope?

Garry,
Michael did not say what GIA system better or worse than AGS.
He said what we can use charts and parametrical( ideal models) for compare RESULTS GIA and AGS cut grading systems.
He said it because Marty wrote what such method is not correct.
If course we can do it, We are doing it. And this method is correct.

 

[Yuri]

Date: 3/8/2007 2:19:54 AM
Author: adamasgem

Well Sergey, the statistics that I posted showing that only about 14% of GIA graded RBC''s got an EX/EX for polish and symmetry tells me, as far away from the cutters you may think I am, that they, the cutters, have a lot to learn. Hope you can help them with the basics..

Marty, cutters work with various types of rough diamonds. Rough classification has more than 18 000 positions and is far more complicated than the current classification of polished diamonds.
1. Out of all incoming rough diamonds cutters preselect stones that intended to be polished to Ideal, Ex/Ex/Ex, AGS000. it can be 15% and depends on quality of rough. The similar idea works for fancy cuts: stones are preselected because of their shape and inclusions. At the last GIA symposium we saw a 100 ct D IF emerald cut that was so flat that every observer could see the backgroung through it. I guess that if a curret choose to polish an excellent rbc, he will probably get 30 - 40ct but not 100. Do you think this stone needs to be recut?
2. Is seems that both GIA and AGS new grading systems helps to sell 14% of round diamonds and makes sales of all other rounds and all fancy cuts more problematic. But it does not mean that all other diamonds are worse. The problem that there are many variations of rough and cutters know this very well. They do their good work to provide many types of diamond to many types of consumers who, in turn, have variuos tastes, opinions and preferences.
And now what? The foremost authorties who never cut diamonds came to the scene because they are going to have more money by issuing papers and say that they have the best cut standard in the world now and all diamonds sould be cut equal like balls for a ball-bearing.

I will be glad to hear your comment on this.

 

[Serg]

.

 

[adamasgem]

Date: 3/8/2007 11:02:40 AM
Author: Serg

Date: 3/8/2007 10:22:45 AM
Author: Garry H (Cut Nut)


You are joking Michael?

Have you read our exchange of letters?

A grading system based on 70,000 judgements in a lighting environment where if a badly cut princess cut was included, the princess cut would have got the best brightness score.

Someone asked who was representing GIA on Pricescope?

Garry,
Michael did not say what GIA system better or worse than AGS.

He said what we can use charts and parametrical( ideal models) for compare RESULTS GIA and AGS cut grading systems. Let''s see now.. 65% of stones would qualify for GIA EX cut while 14% would qulaify for AGS, if both labs graded the same.

The AGS GUIDELINE charts were not intended as GRADING charts.

He said it because Marty wrote what such method is not correct.

If course we can do it, We are doing it. And this method is correct.

If you want more "accurate" Guideline charts, here is what you should do..
1) Get the PGS Software, and run every combination of symmetrical stones to get the PGS grade
2) Now when you want to make the comparisons..
Factor in the differences for
a) Polish and Symmetry
b) Weight reduction criteria GIA vs AGS (I think GIA may use 8% and AGS 5% before a ding, but I''m not sure)
c) averaged, half assed, rounded parameter sets versus actual raytrace
d) you might want to subtract out the "taste test" component from the GIA paradigm if you can figure it out..

OR... Take xx,000 or more ACTUAL Sarin or Helium scans of GIA graded stones and run them through the PGS software to get the ACTUAL grade, assuming that AGS and GIA grade polish and symmetry (finish) similarly (which may be wrong, as it appears that AGS is tighter)
[since this is a multi dimentional non linear problem, I don''t know how many samples you really need to get a stated confidence]

Then take the same xx,000 scans, use the averages to compute a symmetrical stone with Normal rounding, and then take the averages with GIA half assed rounding and run those symmetrical stones through PGS, so you have three sets of data. Actual PGS, averaged PGS symmetrical, and half assed rounded symmetrical PGS

Now look at the statistics of the differences between actual PGS grades and "idealized" grades that you want to parameterize..

BTW, since there are non linearities and gradients, the statistics at each "averaged" point will most probably be different due to the different sensitivities at that point for the same level of assymetries..

At each point, you could even compute a likelyhood function, or probability of a cutter getting the "Grade" you try to represent on the symmetrical stone PGS chart

Now that would be an interesting study...

I think you could even fund AGS to do that for you, or perhaps GIA could do it, since they have infinately more financial and manpower resources than AGS, and may even have the scans they based their grade on..

I think what they would find is that 4% or less of GIA EX cut grade stones would get AGS 0''s, and that the GIA EX range probably includes AGS 0''s through 4''s or worse

You would get different results depending on what database you selected, I suppose if you looked at ONLY Pricescope vendors, their stones may have tighter statistics than the average lot of GIA graded stones, represented by the database study I did from CertNet. Rapnet statistics may be slightly different.
Certainly I would expect that AGS would get a higher % of better cut stone to grade than GIA.

 

[strmrdr]

Date: 3/8/2007 11:57:59 AM
Author: Yuri

Date: 3/8/2007 2:19:54 AM
Author: adamasgem

Well Sergey, the statistics that I posted showing that only about 14% of GIA graded RBC''s got an EX/EX for polish and symmetry tells me, as far away from the cutters you may think I am, that they, the cutters, have a lot to learn. Hope you can help them with the basics..

Marty, cutters work with various types of rough diamonds. Rough classification has more than 18 000 positions and is far more complicated than the current classification of polished diamonds.
1. Out of all incoming rough diamonds cutters preselect stones that intended to be polished to Ideal, Ex/Ex/Ex, AGS000. it can be 15% and depends on quality of rough. The similar idea works for fancy cuts: stones are preselected because of their shape and inclusions. At the last GIA symposium we saw a 100 ct D IF emerald cut that was so flat that every observer could see the backgroung through it. I guess that if a curret choose to polish an excellent rbc, he will probably get 30 - 40ct but not 100. Do you think this stone needs to be recut?
2. Is seems that both GIA and AGS new grading systems helps to sell 14% of round diamonds and makes sales of all other rounds and all fancy cuts more problematic. But it does not mean that all other diamonds are worse. The problem that there are many variations of rough and cutters know this very well. They do their good work to provide many types of diamond to many types of consumers who, in turn, have variuos tastes, opinions and preferences.
And now what? The foremost authorties who never cut diamonds came to the scene because they are going to have more money by issuing papers and say that they have the best cut standard in the world now and all diamonds sould be cut equal like balls for a ball-bearing.

I will be glad to hear your comment on this.

1: I feel there is variation with in the new AGS0 grade and GIA EX grade.
the main problem with the GIA grade is that there is too much variation.

2: AGS didn''t pick their stair step shape on the charts out of their hats.
If a stone meets the criteria for brilliance/contrast/dispersion it gets the grade.

3: a few averaged and rounded numbers can not be used to set the AGS grade for a diamond. Every facet is taken into consideration.

Now why are the original cutting charts off?
My thinking on it is that once AGS started running actual combos thru their final software more combos made the grade than originally thought.

Will new charts be accurate? Nope because the charts cant be used to set the grade only the AGS-PGS software can be used to set the grade.

 

[adamasgem]

Date: 3/8/2007 7:47:36 AM
Author: Serg
1) 14% is not bad if you account what not all Eightstar diamonds received EX/EX Yup, GIA dings them for painting, and AGS may ding them for polish, as they are not going to reduce symmetry for less than perfect polish on on tiny facet, they certainly are optically symmetric.

2) 5-20% market niche is good target for Excellent products. It is silly do one quality product( Do you realy think what market need only AGS0 and GIA EX?)
I agree, but I question people trying to compare the fairly wide GIA paradigm with AGS''s apparently tighter standards. Remember on those 4 borderline stones that we got into a flap about, the DiamondCalc paradigm only rated them a VG for cut..

3) GIGO Yup, Helium apparently produces less Garbage In than other scanners, from what I understand.