Oh, I don't know about that. With a good flow chart, I think a decent programmer can come up with an implementation of the algorithm.
The inventive step here is the how, not the what. If Microsoft was granted a patent on one of the implementations, I would be fine with that. Instead, Microsoft was granted a patent on a very general, obvious, process flow, with all the inventive steps represented as black boxes that they can fill with anything they choose.
Well, yeah, but there's also the specification discussing those steps. Additionally, if I give you, a carpenter, an instruction to fasten two boards together, that's very general and provides no real details... But you can probably come up with a dozen different ways to do so, including nails, screws, glue, dovetail joints, etc. If my overall furniture design for a combination couch/toilet is novel and nonobvious, why shouldn't I be able to claim any trivial variation of the design? Or conversely, why should you get to claim you don't infringe because you used Phillips screws rather than flathead screws?
Are you suggesting that the idea of fastening together two boards is patentable subject matter? If someone invents a novel and non-obvious way of fastening boards together, I think they deserve a patent, but only on their implementation.
In this case, there are non-trivial variations. The hard work, the inventive steps, are left as black boxes in this patent. The difficulty in implementing this idea is not in the process flow, but in the "analysis" and "models." In your couch/toilet idea, should you also get to own all non-trivial variations? In this patent, Microsoft does.
And if their idea is novel and nonobvious, then they should get a claim to cover nearly every implementation. Mind you, this doesn't mean the idea is novel or nonobvious, and a lot of time, those claims get significantly narrowed before the patent is ever issued. But, if, for example, you come up with a working method for faster than light communication, why shouldn't you get a claim that covers such a method, whether it's done in hardware or software, runs Windows or Linux, written in C or COBOL, etc.?
Nobody should be able to patent the concept of "faster than light communication." The implementation can be patented, in any form, but that has nothing to do with this patent. Microsoft has not patented any implementation, it has patented a very reasonable process flow for acheiving a wide variety of implementations. This patent would apply whether it extracts keywords from the page or whether it uses complex machine learning algorithms to analyze the text. They are trying to patent the idea of turning print media into images in as broad a stroke as possible.
I am sure this patent will be scaled back, but only after someone has paid millions of dollars to fight it. This patent does nothing to "promote the progress of Science", it is a roadblock.