Markush groups, as GP mentioned, are there specifically for chemical patents. So that's what's shaped the patent law, not software or cars.
Perhaps, but chemistry is chemistry. It existed before software did, and you need to understand it on its own terms.
And FWIW, chemistry is only a little bit older than software. Like most other fields - there was a qualitative jump in the last ~100 years; arguably we didn't know shit about anything before the 20th century, before the theoretical and practical tools converged to allow some kind of formalized, precise knowledge and experimental work.
> The company which applies for a patent makes a general claim for the usage of the molecule without revealing to their competitors the exact molecule for which they are declaring a useful application
Isn't that against the very point of a patent? That you reveal (make "patent") what you're doing to the public so that the art isn't lost, in exchange for some form of commercial protection? How in the world is this legal?
Early hits are step 0.1 in bringing a drug to market. You patent a large space that gives you room to optimize the structure in terms of safety, efficacy, Pk, metabolites, etc.
Very rarely would you ever get 1 hit in a huge combinatorial screen. You'd likely get a few dozen. But you have no idea which ones would the best.
The company will usually nominate 5 or 6 hits (across the scaffold space) for further screening, then slowly whittle it down from there. By the time it hits humans, it's like 1 with 1 or 2 backups.
So you patent them all.
If you could only patent one, one of two things would happen: 1) the company would just file thousands of individual patents to accomplish the same thing or 2) not both developing the drug further unless they were 100% certain it was the right one (which I've never seen).
Technology and legal practices have far outpaced the USPTO. We have accidentally incentivized sheer volume and intentionally poor record-keeping as a moat. This is a fundamental _inefficiency_, because it de-emphasizes property based engineering and pushes ADMET down the road. Why do so many candidates fail for ADMET reasons a few years down the line? Because people aren't thinking about the risks sooner! Why not? Because surely the patent will cover everything.
Add to everything how siloed big pharma is and it becomes impossible to coordinate across the pipeline. IANAL, but certainly there must be some middle ground here -- the last time the USPTO took a stab at redefining this mess was 2007, but they failed.
https://www.uspto.gov/sites/default/files/web/offices/com/so...
They more or less proposed what you said: file a claim for each different invention, putting the onus of proof on the filer. In the amendment, Markush is appropriate if you have evidence backing that the chemical space has a true shared utility via structure i.e. test it or forget it OR the chemical space is 'obvious'.
When you read the original intent of the Markush decision i.e. "members of [a] Markush group are alternatively usable for the purposes of the invention," it becomes clear that we have strayed waaay from this definition. A Markush can easily contain structures that are impossible to synthesize at all!
You're not supposed to patent anything you haven't actually made, however, there is no validation of that and as a result patent applicants are incentivized to patent as broad a space as possible.
However, if we did move to a "single molecule, single patent" approach, the workload on the PTO would skyrocket. For pharmaceutical companies the cost of a patent and its preparation is infinitesimal compared to the cost of developing a drug - pharma companies wouldn't bat an eye at submitting 1000's of patents for each discovery program.
But, as I stated above, I don't disagree the process could be made better.
So let's remove software from the list of patentable subject matter. Then Pharma won't care about us anymore. One thing at a time.
And considering the pharma industry never really took off until the mid-century I’m not sure your claim of “saw the biggest industry develop” before the parents we have today holds any weight.
not true. That's Actual Reduction to Practice. There is also Constructive Reduction to Practice.
I wrote about this in depth, for software:
https://albertcory50.substack.com/p/no-source-code-no-patent
This is a good place to repeat, "let's just remove software from the patent system."
If you don't, then any changes you propose will being the Big Pharma lobby down on you. Once we get our own IP protection laws (or no IP laws), the pharma companies won't care about us anymore.
I don’t really understand this argument. Patent law is just another moat. If it doesn’t exist for all competitors, it’s just different baseline for calculations. The market is still big and attractive, so all those who give up will open space for those who figured out how to survive without patents. Likely this will lead to a bigger concentration of capital, because in the absence of patent protections you may need to spend more on security. But it is happening anyway for other reasons, so we have to deal with monopolies anyway.
Financially it makes zero sense.
And if the logic is - well the government can fund the research (which is a huge stretch since Pharma R&D is several multiple of gov't funding today), ok, but then you end up paying for the medicine through taxes instead.
>Financially it makes zero sense.
Innovations in funding, protection of secrets etc do happen time to time, so even if I cannot give you satisfying answer now, it does not mean such answer does not exist. It took thousands of years to invent modern corporation and stock exchange. Cryptocurrencies are recent invention. Imagine some sort of pooled research, where all competitors/partners agree to compete only after the discovery, i.e. in production efficiency and costs, marketing etc, with some . Imagine some clever chemical engineering, where a pill is such a mixture of acting substance and harmless components that reverse engineering is too hard to complete without comparable budget. Maybe it will be something completely different that we will find obvious only after it is discovered.
There's lot of promising science that can't be done because incumbents are so adversarial about patent space for structures they don't even care about. To make matters worse, patent data and notation is:
* Poorly structured and defined,
* Difficult to parse,
* Uses combinatorial/wildcard notation (Markush structures [0], which is what modern drug patent law is based on).
So it's a total mess. The best analogy I could give for software is: imagine if you could patent closed source code (the literal code) and also attach a wildcard to every branch in logic within that closed source code, but all you publish are the filenames (as screenshots of directories, not text). You don't have to run due diligence that the permutations are run to spec or even compile, but now nobody can write logic infringing on your hypothetical code (which you never wrote).
If your reaction is 'what the fuck?', yes I agree.
[0] https://en.wikipedia.org/wiki/Markush_structure