If you read the article it seems that he was an early pioneer in the software industry and indie application developers like me probobaly owe him some thanks for that.
But yes, patenting software/algorithms/maths is a stupid idea. Glad that Pythagorus didn't patent a^2 = b^2 + c^2.
If Pythagoras had patented it, the patent would have expired 2600 years ago, meaning you would have had no effect from it.
And about Pythagoras and cults...
Holy non-sequitur, Batman.
How about the idea that people have ever used a publicly disclosed technique that has aged out of patent protection? (Including indirectly, such as by using a library incorporating the technique)?
https://news.ycombinator.com/item?id=37987214
You can't use information that ia kept secret from you.
Yes, keeping software in house and selling services around itbinstead of releasing software has been a popular option as long as software has existed, and changing the IP landscape (or any other part of the legal and business context) in a way which changes the relative incentives between those options changes the frequency at which one option is chosen over the other.
And that's even before considering the way that changing the IP landscape impacts the decision of whether it is worthwhile to develop software for a given market at all.
What goes back to the initial reply you got, nobody every read a software patent to understand an algorithm. People would rather decompile and deobfuscate your binary. What has the added benefit that your binary actually contains the algorithm, while most software patents do not really disclose the thing they are reserving.
Algorithm secrecy was historically never a problem in software. Currently with the cloud and SaaS it is starting to be a very narrow one here or there, but since software started being protected, patents never disclosed the things they protected, and there's no movement on making they start to do that.
This is ultimately true of any product. But if you distribute closed-source, compiled software, the only way to discover the secrets is the same as with any product: reverse engineering.
Furthermore, a high percentage of software is never distributed, but exists only as back-end code, or as proprietary software only ever run on-prem.
Nor can you use obvious techniques that some prick has patented either. :(
So the question becomes "is a software algorithm closer to math or an invention?". I would personally argue the former as you can mathematically model basically any algorithm with TLA+ or CSP or something, and as such it shouldn't be patentable, but I will acknowledge I'm not a patent lawyer and as such I'm kind of speaking out of my ass.
Well, yes, and that would hvae been a problem if Pythagoras lived when there was a patent system and wanted to patent it. Which one imagines he might have hacked around by avoiding the disclosure of a patent and keeping it as cult knowledge.
Because, you know, even without a patent system...