And unlike physical patents it costs nothing to produce the software once the innovation has been made, so there is no need for a small guy to have something to sell to a big company to ensure the innovator is getting something for his efforts.
And unlike physical patents it costs nothing to produce the software once the innovation has been made, so there is no need for a small guy to have something to sell to a big company to ensure the innovator is getting something for his efforts.
PG has said in other essays that it often makes sense for a startup to not worry about whether what they are trying to do until they are sued. If they are still tiny it is unlikely they will be sued even if they are infinging, and if they are large then they will have the resources to either defend against the suit or make a reasonable settlement.
But that is a practical matter. None of that changes the fact that it borders on logically inconsistent to argue for patents of machines while against the patenting of algorithms (and a program is nothing but an algorithm that has been properly formatted).
If you (quite sensibly) throw out the obfuscating language about morality and property rights, however, it's certainly possible to assert "temporary, government-enforced monopolies on ideas of this type will encourage innovation, but government-granted monopolies on ideas of that other type will hinder progress and encourage misuse". This is a statement of factual nature, that may be correct or incorrect, but is not logically inconsistent. It also makes it more obvious that patents, despite being called "IP", are not actually about property at all--they're a regulatory hack, the government deliberately distorting the market with the goal of subsidizing particular behaviors, namely innovation and invention.