Reducing costs might help, but how would you go about reducing the costs?

As GP points out, the most of the cost isn't government fees. It's paying your own private lawyer or law firm, for their time and advice, assuming you choose to do that. And to the other party's lawyer if you lose in litigation, perhaps.

Your lawyer is a private arrangement which mostly doesn't involve the government. The government fees are already low.

Given that, how does your proposal differ from "patent lawyers should charge much less for their time, to people with less money"? It's hard to imagine them voluntarily reducing their income by a large factor to a large number of people, or agreeing to take on lots of low paid work when they have better offers.

But there are other models, e.g. no win no fee works in some fields.

This includes but is not limited to 1) the costs of the courts. 2) the costs of the attorneys to defend, 3) the costs of the technology and executives to mount a proper defense, 4) the opportunity cost to the small defending company which COULD OTHERWISE be focusing it's resources on something productive like a new product or support (vs defending a bogus lawsuit).

Again, the courts do not even begin to scale - the courts are the opposite of scalable.

The new class of enabled patent trolls may lose one case, but they'll win others, possibly because they've bankrupted their victim, and can continue to plague the rest of society.

Moreover, the courts are massively inconsistent. Again anti-scaleable. The inconsistencies get worked out through appeals to higher courts. This literally takes decades and hundreds-of-thousands to millions of dollars per case. They do not get worked out when it is state cases, which is why we have already have venue shopping (you know about the East Texas patent scam courts, right?).

And your so-called "solution" to charge big players orders of magnitudes more cannot be gamed? Simple, make a small company, pay the small fee, buyout later. Or syndicate the fees, or get rounds of investors to handle the fees, since the return is virtually guaranteed. Ya, then you change the fee structure (years later) and the game begins again.

The fees are only a rounding error in the costs, and if you think exponentially higher fees cannot be gamed, I'd like to talk to you about a fantastic deal on an oceanfront property in Kansas, because you are so friggin gullible.

No, automatically granting patents with some weird scaled fee structure and letting the courts sort it out is one of the worst ideas ever. You would literally lay waste to entire sectors of innovation. It would be only a few years before people would soon be screaming for proper centralized regulation and a patent office capable of judging obviousness and prior art; congratulations, you've just re-invented what the founders invented nearly 250 years ago.

Seriously, you are making a nice demonstration of how ignorance of how a real system works creates the illusion of finding wonderous solutions.

For every highly complex system problem there are a huge variety of simple solutions, all of them wrong. Congrats, you just found one.