I am not sure that you can get further from clickbait than the Semiconductor Museum website: but the interview with Jack Haenichen has a lot of interest concerning IP. Haenichen was a pioneer at Motorola in developing practical silicon transistors, and one of his babies, the 2N2222, sold billions and still in production after more than half a century: a very successful inventor by any measure.The romantic ideal about patents is that they are a temporary monopoly granted by the state over some concrete method or system to allow an inventor to recoup their development costs and make the profits necessary to warrant the gamble.  Patents encourage innovation.

Now the development of transistors is surely a classic situation where this should be so: lots of money on research, smart people trying idea after idea,  the early days of an industry, etc.   But Haenichen gives a completely different reason for getting patents:

The real value in patents is in cross-licensing. For example, everybody in the early days was paying to AT&T, because they had the basic transistor patent. What you tried to do was get a portfolio of your own that your competitors, including AT&T, had to use. You tried to get a portfolio that had a value and then you went and “horse-traded” with other companies. At the end of the day, depending on the strength of your portfolio compared to the other guy’s, either you’ll end up paying royalties to him or he’ll end up paying royalties to you. We had a very important set of patents. It was more than one patent, because we had methods patents and structure patents in Motorola’s portfolio.

In other words, you need to get patents because your competitors have them, and they have to be good patents that you competitors will need to progress their businesses. You need to research to reduce your royalties, and hopefully to come out in the positive side.  So what patents did, in this situation, is to discourage the companies from treating the technology as trade secrets.  But there is no sense in which the research would not have been done if there was no patent system: the companies were at a stage of their markets where innovation was king.

As an aside, Haenichen also makes an interesting comment on the development of the transistor itself, saying there was an accidental aspect to this:, but Bell Labs had the ducks lined up to make it workable:

…Walt Brattain was working on at Bell Labs. He was doing work on point contact diodes, and when he was tinkering around, he “accidentally” made a transistor – then Bill Shockley and John Bardeen, both theorists, came in and provided a scientific explanation of what he had done. So germanium was what people started with, but it was widely recognized that any Group IV element would work.

(Brattain, Shockley and Bardeen got the Nobel Prize for Physics for this in 1956.)

Fast forward 50 years, and a leading Microsoft lawyer on IP told me exactly the same thing: that high tech companies need patents for cross-licensing, but for defensive cross-licensing.  As I see it, the logic of the thing has meant that whereas in the days of the transistor patents were highly specific, now the value of a defensive patent is largely in how vague and complicated it is: how much it might mire the both sides in costly litigation they might not have the pockets for.  It is hard to consider this as encouraging innovation, because the logic requires you to get these semi-junk patents on anything that you might need: so you get patents on things you have no intention of actually bringing to market, just in case.  Or you get patents on things someone would have done anyway, eventually.  Or patents on things that you do not seriously think have commercial viability, defensively.

At the same time as the race to the top, for genuine and costly innovation, is a race to the bottom, where the cost is not of research but of the patent system itself.  As much as the patent system encourages innovation, it encourages more patents and so becomes a cost and a drag on innovation.

So it seems to me that a problem with patents (even apart from the quality issue and my general distaste for artificial monopolies), but that it encourages oligopolies and perhaps even cartels: the patent system becomes a government-brokered collusion agreement (“to increase individual members’ profits by reducing competition.“)  In order to play, you need to have something to cross-license.  (Of course, I am not saying that every large holder of dubious patents is engaged in cartel behaviour, any more than that every small holder of dubious patents is a patent troll. One would hope there are at least some babies amongst all that bathwater…)

What can be done?  We are stuck with a patent system, so eliminating patents entirely or suddenly is just a fantasy. Strictly not extending the patent system (i.e. to software or genes) seem reasonable. Reducing patents on systems rather than methods (or vice versa) might do something.  Adjusting the monopoly grant where an unexploited patent within some time (I was told the Japanese used to do this, after 7 years) must be FRAND licensed would be good.  And the biggest is clearly that patent offices must be extremely skeptical of claims, so that every claim needs to be markedly innovative: the patent offices are the gatekeeper society relies on to prevent junk.

Would those things reduce the dynamic of cartelization we now have?

The problem with the patent system I see is at its heart: the idea that companies would not innovate without IP.  That is ridiculous: they would just innovate differently, and to meet different corporate goals.  The direction of innovation might shift towards on efficiency and effectiveness,  and perhaps to immediately marketable products. Market disruptors might appear more in established markets (though existing large players could have less barriers to take over as well.)

If we want the patent system to reward innovation, we really need to it to reward concrete research and development.  So when I am (elected as) King of the World, I would reform the patent system so that you could be granted one year of patent monopoly for every $100,000 you directly spent (after write-offs) on R&D on a technology (excluding sham-R&D costs such as advertising, PR, legal fees, cross-license fees, costs allocated for other patents, commercial and administrative costs, etc.) including opportunity cost for solo inventors in working on the R&D instead of having other income.  Up to some limit.

So: I think of a cool idea in the bath tub?  No patent! I have an organization with a buzzword generator? No patent!  I do a good amount of R&D, but the costs are pretty much covered by grants and tax policy? No patent.  I am a sole inventor who has been working on a project for years instead of having my day job?  Patents might be possible in this case, depending on how much salary I had forgone.  I have a perhaps long-running R&D project on this idea with expensive scientists or trials?  Patents would be possible in this case in (in combination with some kind of utilization FRAND, not to encourage monopoly profits.)

It would still be prone to its own abuses, I expect, but change the dynamic so that the patent system was more targeted to promote concrete  R&D rather than Pollyanna innovation.