Say, the year it is granted the patent costs $1000. The next year, $2000. Ten years since granting, a bit more than million, etc.
If this patent fuels your hockey-stick growth, you have no trouble keeping it.
Keeping a patent defensively becomes much less affordable.
Sharing patent rights for a part of the profit becomes a reasonable way to keep the patent for a few more years.
Even megacorps will eventually face the cost of keeping a patent, and decide to abolish it before it expires, just for economic reasons.
Administration does not become any harder, or more expensive.
[Edited: spelling]
Here are two proposals, pick the one you like better:
1) Abolish patents completely.
2) Abolish patents, except for pharmaceutical industry.
If anything, proposal number 2 should be strictly better than the status quo.
So what are the boundaries we'd place? Only companies whose primary income source is parmaceuticals are allowed to own patents? Are all the non-pharma companies still required to avoid patent infringement?
Or is it that patents can only exist if they're relevant to pharmaceuticals? If so, where do you draw the border? If a pharmaceutical company comes up with a new process for synthesizing some compound Z which they use for their drug, is that patentable? What if it turns out that compound Z is also useful for weather-treating surfaces, and gains widespread use in construction? Now, what if we switch to an alternate timeline where compound Z was invented by a construction company first, and then pharma started to use it? What's the patent situation look like?
#2 seems rather arbitrary to me, even moreso than existing patent law. I'm not convinced it's any better than the status quo, though also it's not clear to me exactly what you mean with it.
To avoid any worries that the drug research will cease if drugs are no longer patentable, the practical solution is to leave patents in place for drugs. Drugs are regulated in million of ways already. Patent system already has special provisions applicable only to drugs. Just craft the new intellectual property category that will cover 95% of ways patents are now used for pharmaceutical research and call it a day.
It doesn't have to be justified from first principles or fully cover all the edge cases because: 1) patent system is already justified on purely consequentialist grounds (at least in the US) 2) existing patent system already fails basic sanity checks, let alone perfectly handling all the edge cases.
You make some interesting points, however generalizing one example from 3d printers and concluding "patents are clearly bad" is really oversimplifying the issue by a lot.
There are plenty of examples of patents stifling innovation (oftentimes purposefully, where a company buys a patent just so nobody else can use it). There are much more complicated examples though, such as the pharmaceutical industry. At least in the US, which funds most of the world's drug research, the entire business incentive for developing new drugs would evaporate without patents. Now there are lots of problems with the pharmaceutical industry, however developing new innovative drugs (with the intent of profiting from them, often viciously) is not one of them.
Everybody agrees the patent system is broken. The problem is nobody can propose a better alternative.