The point is, up until the moment the patent is "exerted", the signal does not matter. The process is just a means of recording that the submitter had a certain idea at a certain moment. Which, to me, sounds perfectly legitimate.

With this construct, an inventor could go to e.g. Google, and sell them this idea ("potential patent"). This way, small inventors wouldn't need to have a big pile of cash just to secure their ideas.

This sounds wrong to me.

The idea of the patent is (at leas in part) to protect inventor's investment. When someone spent significant effort researching and prototyping some useful device, it makes sense to pay a small sum for its legal protection from copycats.

If someone just stumbled upon an idea, without any plans to use it for profit and without sunk costs that one would like to protect, why grant a patent? Someone else could (and historically did) invent the same thing independently, and put it to actual use. Why impede the efforts of that latter person?

While we're at, it let me remind the idea of exponential patent fees.

A patent should require e.g. $100 to keep for the first year, $400 for the next year, $1600 for the third year, etc, quadrupling the cost each year. If you're on a gold mine like Singer's needle, or Edison's lamp, or a transistor, you have no trouble paying for that patent for 10 or maybe 20 years. If you patented something useful but minor, you can keep it patented as long as it makes financial sense, and then put it to public domain. If you're a NPE patent troll, you're toast in a really short term.

Not really sure with the current climate what we need is more patents. The one thing the fee does do is pay to have it examined which does act as a filter.