You're also absolutely right that litigation is expensive, and most early-stage companies will never file a lawsuit. But patents actually create a lot of value for start-ups, and in practice, leverage can come from fundraising and acquisitions, competitive deterrence, and, hopefully, preventing someone else from filing a patent that blocks yours. Very few patents are actually enforced, but they do add value in other ways.
Re the clean-room / LLM reverse engineering point: patents don't protect code, they protect mechanisms. If someone clean-rooms a different implementation of the same underlying technique, the patent still covers it. That's actually the whole point of a patent vs. copyright.
Just executing well and hoping the best person wins is philosophically correct and justifiable. Unfortunately, though, sometimes it's not an even playing field. A big player with significant resources can quickly squash an early-stage start with fantastic ideas. I also say a big part of inventing is just identifying the problem. I've worked with a lot of startups that have really identified a gap in the market or a particular need and worked hard to develop a solution. Just to have a quick follow slipstream on all of that domain experience and hard work can be incredibly frustrating.
Maybe not the most satisfying response, but hopefully of some value. Thanks for commenting.