[dathinab]

> "HTML on a computer with foo and bar"

But here it's where the patent system is broken. Doing conceptually trivial combinations of on itself non patentable thinks should _never_ be patentable, at lest for software but probably for anything. Sure patenting a specific complex mechanism for combining two thinks might be patentable but in case of software this is hardly ever the case, most times its more or less trivial glue code.

So e.g. Face recognition + smartphone + unlock screen might all be patentable by itself. Especially a faceregocnition algorithm which works especially good for smartphone usecase might be patentable. Bu using existing face recognition on a smartphone to unlock the screen shouldn't be patentable. They are all existing components and combining them is conceptually trivial. In the end patents where meant to protect research investments, not random ideas.

[adrianN]

Every invention is just a combination of existing things with maybe a new twist somewhere. James Watt wasn't the first person to build a steam engine, he just introduced a separate condenser to an existing design.

It's really hard to find a good line of what should constitute a patentable invention. Personally I think that no software should be patentable, ultimately because I don't think that Mathematics should belong to anybody. But I understand that other people see this differently.

[bdowling]

> existing things with maybe a new twist somewhere.

To be valid, a claim must be novel (new) and non-obvious over the prior art. So there absolutely must be a new twist somewhere.

[adrianN]

I can still build a machine entirely from off-the shelf parts, connecting them exactly in the ways they're built to connect to each other, and claim to have invented it.

[bdowling]

If the resulting combination is a machine that is useful, new, and non-obvious, then I'd call that an invention.