Or different manufacturer at least... this way if a very rare bug appears the other drive is probably still safe.

I always use many manufacturers, because no single manufacturer is totally failure proof.

You're close. Yes, I could only copy the files over to Windows from a linux partition, because linux did not complain about a corrupt file system. I think it's a strange, rare, Windows bug, or the GOG team that put together 1.63 screwed up somehow. Either way, Windows complains of corrupted files. From the 1.63 zip file, AND the individual files that I stored on bluray. That is the only explanation. My only solution to this is to buy an external HDD and copy them over from linux, and just run the installer from the external HDD.

It is strange. I think it's a weird Windows bug. I also bought Mafia Definitive Edition, copied over to bluray, and Windows does not complain about (the same sized files mind you) being copied over to Windows. I dunno man. Is it Gog, or a freak Windows thing? I honestly don't know, but it's one of the two.
And I will tell you. After all these decades, Windows is STILL not multitasking. Yes, it can make use of multiple cores. But every single time I have an external bluray (even dvd) drive connected, and I just try to open a folder on my HDD, while it is waiting to access the external drive -- windows totally freezes on me. It opens finally after the optical drive finally reads the disc. Windows is NOT multitasking. At least the file system. That they couldn't fix that over 20 years is ridiculous. I have no such problems in linux.

OK, what i did for testing:

System is the newest Windows 10, file systems used are all NTFS. The ZIP file is from my archive, no fresh download.

1. I was moving the 100+ GB CP_163 ZIP to a external M2 SSD over USB-C now.

2. I was unzipping the whole content using 7Zip.

3. I was moving the whole content [100+ GB CP_163 ZIP] together with 28 Setup files to a internal archive HDD.

4. The files arrived properly without any error message.

So, i am not able to detect a issue.

Besides, the data retention story is indeed a difficult matter because lets face the truth: No exact data is known, not even by those usually considered a "know it all".

The main issue is simply, we do not know at which point the data may fade toward a critical level, which means by how much a charge on a flash module or by how much the magnetic property of a disc is fading over time; a lot of guessing here... even of it could be tested by science... the interest is rather low for those capable of (which needs complicated equipment).

As well unknown how much of "loss of signal strength" can be considered critical for properly reading it. It surely can be different by manufacturer and their technology used, so there is no general answer. A improved "reading head" of a HDD may be able to read way weaker magnetic strength than a old reading head. However... the data on new platters got a way higher density which as well is more challenging for the reading head. I think a old air filled drive is probably more secure in long term as it got lesser complicated stuff (such as Helium able to leak over time) inside, able to become faulty over time.

Such "old technology HDDs" has only been built up to the size of 10 TB, and one of the best of them was the WD Ultrastar. Ultrastar is their best brand.... as well the Toshiba Enterprise MG10. There are still WD Gold or Toshiba X300 series around at a astounding price, yet they are worse than the cheaper "equivalent". It is simply a marketing-demand, as the people buying it "believe into it", as with everything holy to a human (reliable drives for their valuable data) the rational mind is not of much use anymore.... they simply have to "believe into it". If WD Gold or Toshiba X300 is considered better than anything.... then this is a product which is offered at a price some are willing to pay for.

Truth is... the usually cheaper datacenter versions or enterprise versions are the best drives those manufacturers got to offer, and they can handle everything.

As far as the fading of data goes (charge or magnetization strength) my guess is that a HDD may lose between 1-5% of strength each year, which may vary a lot due to many circumstances and the material used... so i would say a average loss of about 3%, For a flash module it could be anywhere between 10-50%, with a average loss of around 30% a year, so approximately 10 times higher vs. a HDD. As well, depends a lot on the flash module used and many factors involved. In general M2 SSDs offer way higher quality NAND which are more likely in the "lower range", so 10-30% i assume, while portable NAND is more likely in the upper range, so 30-50% i assume.

Nintendo is probably using a more secure "portable NAND", on their cartridges yet i dare to say that it may still be not better than the average M2 SSD NAND, so, those cartridges are no secure way for storing games longterm... and actually... Nintendo never had the intention to do so (sorry gamers... i want to be honest with you). I guess, above 5 years those cartridge datas is probably at risk.. yet i lack any data so far, because Nintendo fans are known to be hardcore at believing something. Even if the data is gone... the may still believe "it is still there".

Although, maybe Nintendo is using some "refresh methods"... such as on modern M2 SSDs, which is rather unlikely... given "how cheap" they are usually acting on their hardware-technology.

Anyway, i have to believe... that a HDD may slowly start to make reading errors at about 30% loss of signal strength (can be tested but the companies got no interest it seems), so after around 10 years the average HDD may become a risk for data integrity, but it can theoretically vary between 6 and 30 years (based on 1-5% signal loss a year).

For NAND (flash modules) it es even more complicated... i assume, if it may become hard to read at 60% loss (i believe, a NAND can be read at lower strength left vs. a platter, due to the non mechanical nature) of strength (this is just average loss, it can as well vary from cell to cell, not just from NAND to NAND. In rare cases the cell may even totally "give up" while another cell inside same NAND is still at 50%... so indeed a difficult matter).

My guess simply is, if a NAND is not refreshed in some way... it may already fade at around 3 years (based on 20% loss a year) on a average M2 NAND... and after only 1.5 years (based on 40% loss a year) on a average portable flash NAND.

So i guess... people should be worried a little and know the "limits" of todays data technology...

M2 SSDs nowadays should have a mechanism able to counter this issue, but it will need a modern controller at least and sufficient "uptime":

Besides, yes, i already had a portable drive (rather good quality 1 TB drive from Sandisk) which was completely losing any data on a folder... after around 2-3 years. I simply forgot this folder and i was not refreshing this drive, as i only was using it for moving around data between computers.... so i already had experienced such a loss., The folder was not important because it was backed up, so a valuable "lesson" at least. This drive still works properly if i copy any fresh data on it, but if data is left there for extended period... it will kill those data for certain!!!


So, always remember people: Any portable drive need to be refreshed every single year... and in general, no safe storage at all. Copy the files over to a HDD... refresh it... but never leave any data there for extended period, general rule.

Regarding the topic main issue: I believe the file system has been corrupted but Linux was somehow still able to read it by using another "work around". It is rarely known (by common people), that not only the data itself can become corrupted... as well the file system. Reason why it is important to reformat... not only rewrite!

I filesystem corrupted, some advanced programs (or specialized OS... Linux is good for it) may still be able to read it but this is a lottery case, which should be avoided.

Another possibility is some "screw ups" with the Linux data. Windows may act cranky if Linux is somehow dual booting or even just installed somewhere.... as it may influence its own "mechanisms".

So. dual boot or comparable stuff is sometimes able to screw up stuff.

I mean, Wndows is NOT natively supporting Linux, so this stuff is always "outside regular specs".