So instead of wondering how an existing application of edge detection can benefit from this algorithm, HN is discussing the politics of profit in academia?
Let's focus HN-ers. How can this new algorithm be used to build a better "mouse trap"?
If it's currently (or soon to be) patented, then this algorithm can't even be used as a starting point for discussion. Anything that comes from such a discussion would be legally tainted by the patented algorithm. Patenting this renders it utterly useless to anyone but a patent troll, and that is not hyperbole.
Put another way: if someone was going to be developing an algorithm that does anything remotely like what this algorithm does, the legal advice that that person's institution would give them would be to not read anything about this particular algorithm. Like, don't even visit this web page or participate in this thread on HN, because doing so exposes a legal attack surface for future litigation. That's how effed up this is, it's inexcusable.
This new method has zero value, regardless of it's patent status, if it cannot be used to build a better mouse trap. So, let's first see if one can ever use this to enable a superior product first.
Let's focus HN-ers. How can this new algorithm be used to build a better "mouse trap"?