While true it's not impossible for engineers to make mistakes, even experienced senior engineers.

I think the mistake here is perhaps the idea that what is proposed are actually problems that need solving. The idea of indexes being managed by the OS is sort of handwaved away in the article because: "they have to store their indexing information in a separate database. It is easy for the database to become out of synchronization with the file system. Also, to speed up the indexing process, users often only index a portion of the file system so using the index might not turn up the file(s) you were looking for."

A second mistake is thinking it's possible to solve the "what is this file" problem:

    The metadata record does not have a file classification system. To determine what is in a file, the file name or the data stream must be examined.

This will always be true. You can make a best effort based on some indicators but it will always be prone to problems. What ever system you create that you believe will address this problem will be prone to failures, almost by necessity, because a "type" of a file is largely a philosophical thing that everyone pretends is technical. Even if you solve for 90% of cases, you're still going to have enough problem cases that it won't be reliable. It may work for simpler cases like text files, videos, zip archives, and images. How do you classify a docx though? Is it a zip file? A "document" file? Is something Javascript just because it ends in .js?

Is adding this level of overhead to writes useful? You can see in the video his test framework sets up "hundreds of millions" in seconds and yet copying actual live data from his Program Files directory takes 1m10s for 24,431 files. This suggests to me that at the minimum the "hundreds of millions" of objects are not representative of real live data, since that takes approximately 2 seconds to create.

I'd also suggest that if this is truly innovative and these problematic things from the presentation are actually just artificial and that it is doing things under the hood, it should be using a true OS filesystem browser to do this same work. To convince me it could replace standard filesystems, demo me an implementation of the OS running on it, or at the very least using it as a storage drive.