[Aloisius]

There is a rather long history of the workings of cryptography products kept secret, so yes, it is entirely possible that the underlying math would have been kept as a trade secret.

It is also possible that it would have never been created in the first place because resources were allocated to other patentable inventions.

Of course, in the case of RSA, a similar algorithm was developed separately by the British government and kept secret for 24 years.

[Dylan16807]

Kept secret by the government is very different from kept secret in a product anyone can buy and encrypt arbitrary messages with.

What are some message-sending cryptography products where people didn't know how the encryption worked for a long time?

> It is also possible that it would have never been created in the first place because resources were allocated to other patentable inventions.

Given the history of RSA in particular, I'm extremely skeptical of that.

[Aloisius]

> Given the history of RSA in particular, I'm extremely skeptical of that.

Well then you might want to read about RC4 which only became public after it was leaked. Prior to being leaked, it was RSA's cash cow and one of the most popular encryption algorithms worldwide due to it's speed and the fact that it was exportable (with a 40 bit key).

Indeed, RSA was rather notorious for keeping crypto algorithms as trade secrets (RC2, SecuriID OTP, etc.)

[Dylan16807]

You quoted the wrong part of my post. A trade secret algorithm supports the idea that it would have been made anyway.

Looking at RC4, how widespread was it before that leak? How many users did it have? Wikipedia lists it being added to a bunch of protocols but all after the leak.

Also more recent cryptography has lots of extremely public competition between nonpatented algorithm proposals, which largely undermines this entire realm of study as a reason to continue to have software patents.

[Aloisius]

> A trade secret algorithm supports the idea that it would have been made anyway

Made and kept secret. If the leak never happened (like it hasn't for other RSA trade secrets), we may not know the algorithm to this day. No one could have built upon it. We may have spent years trying to reinvent the wheel rather than trying to improve upon it.

> how widespread was it before that leak

It was one of the most popular stream ciphers in the world, due to it's speed and the fact it could be exported, and it helped launch RSA as a company.

[Dylan16807]

How sure are we that the "leak" wasn't someone decompiling it? Because if it was decompiled, then even if that post didn't happen someone else doing it was pretty inevitable.

As for any intact RSA trade secrets, I doubt any of them were all that special by the 20 year mark. A trade secret slows down innovation but it has to get pretty extreme before a software trade secret slows down innovation more than a patent. (And yes, sometimes you can build on someone else's patent without waiting for it to expire, but on average the delay to the progress of the arts is pretty big.)

[Aloisius]

Would it matter if it was? Even if it could be reverse engineered, so can physical inventions.

Just like with physical inventions, the issue of trade secrets isn't just that it can slow down innovation by wasting resources reinventing an existing invention, but also that inventions can be lost altogether just because the inventor failed to popularize or commercialize it.

I really doubt the average delay for improving upon a patent into something novel is very long. I rarely see software patents that don't cite more than a few recently issued patents.

Nor have I seen much evidence of software patents actually stiffing invention, except for overly broad idea patents (thankfully neutered by Alice). Most of the issues with software patents instead seem to be around wanting to use the specific invention rather than improving upon it - which is rather the opposite of innovation. The LZW patent, for example, was an issue because it was used by GIFs, not because no one could invent a novel derivative of LZW - those took less than a year to appear.

That's not to say software patents don't have issues of course. We'd be better off if patent terms were shorter or required compulsory licensing, if applications were detailed enough to actually reproduce the invention rather than vague descriptions (the LZW patent, as annoying as it was, contained actual source code) and if the standards for what was considered novel were based on more than just abstract descriptions.