Evidence that the alpha pattern is probably from a printer made in 2016, HQ Beta implicated too
I may have finally cracked the alpha code with a couple new printed samples I picked up today.
Reminder that HQ, LQ are low quality and high quality.
To bring us up to date:
-
Alpha prototypes are printed black & white so limited analysis can be done. Majority of the copies (LQ) sold seem to be photocopies of a single HQ set.
-
Roughly 8 Alpha playtest that have been graded have HQ backsides. All cards I tested with HQ back have dots that say 1996. Majority of alpha playtest (specifically the ones that first appeared on auction) have a unique dot pattern. Since then, there have been trainers and a few of the more expensive Pokemon that have appeared with dots that decode to 2024.
-
Majority of Beta playtests are LQ, all tested have 2024 dots. A single near-complete HQ set exists with no dots on the front. The backside contains traces of the alpha pattern, which suggests the back of the HQ beta are photocopies of the alpha playtest back - which means they shouldn’t be older than alpha playtest.
-
Delta playtest, gamma, alpha presentation have 2024 dots on all copies tested.
-
Beta presentation (corocoro magazine cards) don’t appear to have dots; no obvious issues with them that I’ve seen
TL;DR: certain alpha and beta playtests are inconclusive at this point. If alphas are printed recently, betas are implicated too.
Contrast between the “Xerox” dots seen on LQ beta and the “Konica Minolta” dots on alpha. All inconclusive alpha cards have the exact same dot matrix.
This Seadra also has similar Konica Minolta dots to alpha but it has a unique configuration. I’ve explained why I believe this to be a photocopy.
I apologize in advance: this will be pretty technical. You can skip to the end to see the main result which doesn’t require all the details.
I’ve collected 32 unique examples of Konica Minolta printed dots (plus the two above). Majority of them come from this dataset: https://madm.dfki.de/downloads-ds-mic - this is the data that was used to decode the “Xerox” pattern. All printers here are pre-2005. I found about 8 more examples that were published in research papers and printed 2 samples out myself today.
Basically the pattern is a grid with 30 boxes. Each box encodes one number from 0 to 5. Multiple boxes can be put together to form a single number in base 6. For example, the boxes used store the hour information (box 25 and box 24 below) encode “2” and “3”, which together is 23 (in base 6), which is 15 (in decimal) = 15:00 = 3pm. I put the time-based blocks in blue, I’ll note that the way to decode the time blocks does not seem to be 100% consistent across all printer models.
Unfortunately majority of the samples have the time information absent - including the alpha pattern (and Seadra pattern):
I suspect most of the black boxes have to do with the printer serial number. One thing that stood out to me is that the boxes I highlighted in orange are the only ones consistent across the same printer model. However, box 7 was the same across all my pre-2005 samples. When I finally collected data from newer printers, I noticed box 7 changes on newer models.
Basically, it seems box 7, 3, and 2 together encode a number that corresponds to the printer model. I checked the numbers of all the samples I have, here are the results:
Note: I did my best to find the manufactured year of each. It is actually quite difficult. Many of them can be found by looking at the manuals for each model. It also has very little to do with the numbering of the model.
| year | code_base6 | code | brand | model |
|---|---|---|---|---|
| 2001 | 001 | 1 | Minolta-QMS | Desklaser 2200 |
| 2001 | 001 | 1 | Minolta-QMS | Magicolor 2210 |
| 2001 | 001 | 1 | Minolta-QMS | Magicolor 2210 |
| 2001 | 001 | 1 | Epson | AcuLaser C2000 |
| 2001 | 003 | 3 | Minolta | CF1501 |
| 2001 | 003 | 3 | Minolta | DiALTA Color CF2001 |
| 2003 | 005 | 5 | Minolta-QMS | Magicolor 2300DL |
| 2003 | 005 | 5 | Minolta-QMS | Magicolor 2300DL |
| 2003 | 005 | 5 | Minolta-QMS | Magicolor 2300DL |
| 2003 | 005 | 5 | Minolta | Magicolor 2300DL |
| 2003 | 005 | 5 | Minolta | Magicolor 2300DL |
| 2003 | 005 | 5 | Konica Minolta | Magicolor 2300DL |
| 2003 | 005 | 5 | Konica Minolta | Magicolor 2300DL |
| 2003 | 005 | 5 | Epson | AcuLaser C900 |
| 2003 | 005 | 5 | Epson | AcuLaser C1900 |
| 2003 | 005 | 5 | Epson | AcuLaser C1900 |
| 2003 | 011 | 7 | Minolta-QMS | Magicolor 7300 |
| 2004 | 022 | 14 | Konica Minolta | Bizhub C350 |
| 2005 | 030 | 18 | Konica Minolta | Bizhub C252 |
| 2005 | 030 | 18 | Konica Minolta | Bizhub C252 |
| 2005 | 030 | 18 | Konica Minolta | Bizhub C252 |
| 2005 | 030 | 18 | Konica Minolta | Bizhub C252 |
| 2005 | 030 | 18 | Konica Minolta | Bizhub C252 |
| 2005 | 031 | 19 | Konica Minolta | Magicolor 2430DL |
| 2005 | 031 | 19 | Konica Minolta | Magicolor 2430DL |
| 2005 | 031 | 19 | Konica Minolta | Magicolor 2430DL |
| 2005 | 031 | 19 | Konica Minolta | Magicolor 2430DL |
| 2008 | 100 | 36 | Konica Minolta | Magicolor 8650 |
| 2012 | 221 | 79 | Konica Minolta | Bizhub C754 |
| 2012 | 221 | 79 | Konica Minolta | Bizhub C754 |
| Alpha pattern | 300 | 108 | Konica Minolta? | ??? |
| 2016 | 303 | 111 | Konica Minolta | Bizhub C658 |
| Seadra pattern | 333 | 129 | Konica Minolta? | ??? |
| 2021 | 353 | 141 | Konica Minolta | Bizhub AccurioPrint C4065 |
The code starts from 1 sometime around 2000-2001 and increments for newer models. You can see some models (and even Epson - a different brand) have the same code. I believe because they are using the same internal OEM parts. Konica and Minolta did not merge until 2003 and in those early days the printers were sold under a few different brand names. As a scatterplot, we can really see the relationship:
Primary result - comparing printer model code to approximate year produced
It would be extraordinarily unlikely to see this by pure chance (R^2= 0.983). Not to mention that all printers that share the same model also share the same code. If my interpretation is accurate, what it also implies is the following:
The alpha dot pattern was likely produced by a printer manufactured around 2016.
(2019 for Seadra)
Now, this is not 100% definitive. I’m making an inference based on the data; there could be weird exceptions. Notably though, I also think this pattern may not have existed before 2000. Basically, Minolta (which was mostly a camera company at the time) started mass producing their own inkjet printer heads around the year 2000. So it makes sense that this is when the dot pattern was established and why the model code would start from 1 around this point in time.
source
Let me know if you want any raw data so you can validate my work. I’m also happy to continue to check more prints from Konica Minolta machines to see if it conforms.
Summary:
Konica Minolta printers appear to encode the printer model number dot pattern matrix. The dots from the alpha pattern point to the “alpha printer” being built in 2016 (unclear when the actual printing of alpha happened, we can only date the printer)
I am convinced to my own personal satisfaction that the alpha prototypes were printed from a modern printer. If true, this also implicates the HQ beta prototypes (ie. every beta copy has now been brought into question - are they even variants that existed in the 90s?)
In other words, if a card has a backside and it’s not high quality, I would assume it’s been printed recently.