>Am I going to stumble upon Steensgard's pointer analysis by accident?

Yes! Yes, you could stumble on Steensgard's algorithm. Like convergent evolution in nature, it's easy for requirements to shape implementation, and for multiple implementations to naturally overlap. There are so many software patents out there- it is a minefield.

Additionally, engineers at large companies are discouraged from doing their own patent research, because of the risk of losing a "sorry, we created this organically in good faith" defense in litigation.

Publication of patents does not offset the chilling effect and legal overhead on our industry.