What cracking slabs tells us
Resubmission is the hobby quietly admitting that grades vary. The part I find genuinely interesting is what scanning technology is about to do about it.
I assess measurement instruments for a living. This is that lens pointed at something I care about as a collector.
A grade is a measurement, and I do not think the hobby treats it as one.
Condition is a private judgment. Two people looking at the same card, both acting in good faith, will disagree about the corners, and there is no way to settle it between them. A grade turns that private judgment into a public number that strangers can transact on without trusting each other's eyes. That is a real and useful thing to do, and the market built on top of it could not exist otherwise.
So the plastic is not the product, and the number is not really the product either. The product is the credibility of the number.
Which raises the question you ask of any measurement before you ask anything else. Not whether it is accurate. Whether it is consistent. If the same card goes through twice, does it come out the same?
The tell
I do not have data on this and I am not going to pretend otherwise. What I have is a behavior that the whole hobby can see, and I think it says something clear.
People crack slabs and resubmit cards hoping for a better grade. Openly, routinely, as a strategy. There are guides written about when to do it and how to pick which cards are worth the attempt.
Work through what that requires in order to make sense. If a grade were a fixed function of the card, a 9 would come back a 9 every single time. The expected value of resubmitting would be zero, minus the fees, minus shipping, minus the risk of damaging the card getting it out of the holder. It would be a straightforwardly losing move, and after a few people tried it the practice would die.
Instead it is an industry.
I want to be careful about what this does and does not show. It does not tell us how big the inconsistency is. It could be small and confined to genuinely borderline cards, which would be entirely reasonable and close to unavoidable with human judgment. Some disagreement between expert judges is normal in every field where people score things, and grading is hard skilled work done at volume under time pressure.
But it does tell us the inconsistency is not zero, because nobody builds a business on exploiting a coin flip that does not exist.
Why it matters more than it used to
Grade variance used to be a cosmetic problem. At current prices it is a valuation problem.
On a lot of modern cards the step from one grade to the next is not a few percent, it is several times the price. When one point moves the value that much, any inconsistency in assigning that point is not a rounding error. It is most of the valuation.
And the people most exposed are the newest. Somebody who has been at this thirty years can look at a card, form a view, and treat the grade as confirmation. Somebody who arrived last year cannot. They are not buying a card, they are buying a number, and they have no independent way to check it.
Where it gets interesting
This is the part I actually want to write about, because the constraint that created the problem is the one that technology is now removing.
The reason grades vary is that a person assigns them. People are inconsistent in ways that have nothing to do with competence: fatigue, volume, the card they looked at immediately before, the drift that happens to any human standard over months. None of that is a character flaw. It is just what human judgment does, and it is why every field that scores things by eye eventually builds instruments.
A machine does not have that problem. A deterministic system applying the same rule to the same image returns the same answer every time. That is not a marketing claim, it is a property of deterministic systems, and it means near-perfect consistency is achievable in a way it simply is not for human graders.
Centering is the clearest case. Centering is arithmetic performed on an image. There is no interpretation in it, no judgment call, nothing that should ever vary between two examinations of the same card. That it varies at all is purely an artifact of a person doing the measuring.
TAG is the clearest example of where this goes. They describe grading with photometric stereoscopic imaging on a thousand-point scale that maps back to the familiar one to ten, and they publish a defect report with the card: high-resolution images of each identified flaw, scoring broken out by attribute, centering metrics, and a viewer that zooms to eight hundred percent. Consistency is the explicit positioning rather than an afterthought.
Those are the company's own claims about its own product and I have not tested them. But the direction is right, and the direction is what I am watching.
The part that ties it together
Here is the detail I did not expect to find, and it connects the two halves of this piece.
In 2021, PSA acquired a company called Genamint. Its technology does card fingerprinting: identifying an individual physical card so it can be tracked over time. In PSA's own announcement, the stated capabilities include tracking resubmissions and detecting alterations to a card's surface.
Sit with that for a moment. The same imaging technology that could make grading reproducible also makes each specific card recognizable. Which means the resubmission strategy and the reliability question are not two problems. They are one problem with one solution. A system that knows it has seen this exact card before can tell you it graded it an 8 last year, and the entire economics of cracking and resubmitting changes.
The capability has been owned by the largest grader in the market for five years. Meanwhile, in 2026, people are still writing guides on when to resubmit.
I do not know what to conclude from that and I am not going to invent a conclusion. Possibly it is used and the guides are selling a strategy that no longer works as well as advertised. Possibly integrating it at that volume is genuinely hard. Either way I find the gap between the capability and the behavior more interesting than anything else in the grading conversation right now.
Worth noting alongside it: PSA was explicit at the time that the goal was to assist graders rather than replace them. "We're not eliminating humans from the grading process," as they put it. That is a defensible position, and it is also the decision that keeps the variance in the loop.
What machines do not fix
This is where I would push back on the optimism, including my own, because consistency is not the same thing as correctness.
A measurement can be perfectly reproducible and still be measuring the wrong thing. A scale that reads three pounds heavy reads three pounds heavy every single time, with beautiful consistency. Reliability is necessary, and it is not sufficient.
So the question I would ask of any automated grader is not whether it is consistent. It will be. The question is what its scale is anchored to.
If a system is trained to reproduce historical human grades, then it has learned the average of a noisy standard. You get an extremely stable reproduction of the thing that was unstable, which is progress of a sort, but it is not the same as measuring the card. If instead the scale is anchored to physical properties, measured directly, then you have a genuinely new standard, and it will disagree with the legacy grades in ways that will be uncomfortable for everyone holding a slab.
That tension is the real story of the next few years, and it is a harder problem than the engineering.
What I am watching for
Not demands. Just the things I think will tell us which way this goes.
- Whether machine-graded cards survive resubmission. This one tests itself. If people start cracking and resubmitting automated grades and it works, we learn something immediately. If the practice never takes hold in that corner of the market, we learn something too.
- What the scale is anchored to. Physical measurement, or a reproduction of legacy grades. This determines whether we get a better ruler or a faster version of the old one.
- Whether fingerprinting becomes visible to buyers. If a card's grading history is something you can look up before purchasing, that changes the negotiation more than the grade itself does.
- Whether the defect report displaces the number. This is the outcome I would most like to see, and it is the one my day job makes me biased toward.
On that last point: in measurement, you would nearly always rather have the component scores than the composite. A single number is a summary, and summarising throws information away. If I can see the centering figures, a map of every flaw, and zoom in far enough to judge for myself, then the one-to-ten grade becomes a convenience rather than the whole truth, and a point of disagreement between two graders stops mattering very much.
The most interesting version of this future is not a more accurate number. It is a world where the number matters less because you can see everything underneath it.
Where I stand
My own grading is on hold. That is a personal decision and not advice to you.
It is not a boycott and I have nothing against anyone in the business. It is that I cannot work out what I am buying. If I send a card in and it comes back a 9, I do not know how much of that is the card and how much is the day, and I do not know how to price the difference. So I am waiting, and what I am waiting for is this technology to mature enough that the question stops being interesting.
If grading makes sense for your cards, run the numbers honestly and go ahead. Plenty of people are better served by a graded card than a raw one, and nothing here argues otherwise.
And if I have any of this wrong, particularly the technical side, I would rather be corrected than right. Tell me and I will update the piece and say where I was wrong.
Sources
- PSA Acquires Genamint to Introduce Next-Generation Technology to Grading Process, PSA
- TAG Grading, for their description of their imaging and reporting
- Cracking Slabs & Regrading: When to Resubmit Graded Cards, PreGradeCards
- Over 26 Million Cards Graded in 2025, Sports Illustrated, for scale