We've seen lots of questions like "how to patent an idea from a paper", but let's do this the other way 'round.
Let's say there is a commercial software product that has a nice feature: a user can use mouse gestures to draw on a virtual canvas using a realistically looking brush. The company patented this feature. The patent states what the user can do and what the result is. It does not describe the supposedly extremely complex algorithm(s) used to achieve the output. The company chose not to publish a paper on the algorithm(s) although a tremendous amount of research must have been done.
So the algorithm exists but is unpublished. This is an ugly situation for science.
What could be the reasons the company did not publish a paper on the algorithm when it is most likely only usable in the patented use-case scenario?
And, more importantly, what is the scientific community's take on finding an algorithm that behaves the same (might as well be the same, the researcher cannot know that) and publishing it?