There are many reasons for why the "outcomes have not been very useful," so there's not much point in speculating. Suffice it to say if patents have been filed in the past (i.e. some number of people thought their research was worth protecting) and researchers at a given institution are still allowed to file patents on their work, it makes no sense to take away the supporting resources that the institution is supposed to provide to help in filing patents.
"Donald Knuth, a computer scientist, refuses to use e-mail on the ground that his job is to be “on the bottom of things” rather than “on top of things”.
If you read the article, the "complexity" refers to the lack of translation from bench to clinic. IP is the fundamental currency to attract investment to get scientific discoveries through the regulatory process. You can still have "open science" in the sense you are using this term and have patents; see CRISPR/Cas9 as a prime example.
One way to deal with the "complexity of the problem" is to actually keep patentability and license out the patents to start-ups or established companies that can actually do something with them. I guess this would require more effort to match up/recruit the institution's patents (assuming they have any useful ones) to companies that might be interested. But economically, in terms of getting their research translated to the bench more quickly, this is an established model that might be more effective in the long-term.
Indeed, "Participation is voluntary, and researchers can pursue patents on their own, but MNI will not pay the fees or help with the paperwork." This basically shifts responsibility from the institution to the researcher; not sure how this is a good idea.