This article, and LoperOS in general, are schizophrenia applied to computing systems. As near as I can figure out, the author's argument is thus:

  [1] The concept of an operating system is fundamentally flawed.
  Properly designed hardware could safely run arbitrary code from
  untrusted sources without exposing the user to any security or
  performance problems.

  [2] The concept of a file format is fundamentally flawed. Formats
  are like special-purpose languages, each with their own virtual
  machine. For example, HTML is a language interpreted by a web
  browser, PNG is a language interpreted by an image gallery, and
  H.264 is a language interpreted by its appropriate decoder.

  [3] With proper hardware, there would be no need for formats,
  because every file (whether a document, image, or video) could
  be published as a single structure containing both data and the
  rendering code required to interpret that data. Users would simply
  download the file and execute it; the bundled code would read the
  bundled data and render it appropriately.

Even leaving aside the practical problems with the author's approach (developing a new hardware architecture, replacing extant computers / applications, and user education), there are several objections I have.

The first is that it assumes content providers are capable and willing to maintain their bundled code. I believe this assumption is obviously unfounded. Think of all those sites which have only grudgingly started testing against modern browsers. Now, imagine that instead of being coerced into upgrading their sites, they can simply bundle IE3 in perpetuity.

A second objection is that it gives content providers complete control over the user experience. How would you like it if websites could decide to show you minute-long, fullscreen, unskippable advertisements? How do you confirm that you're really logging into your bank, and not into a middleman in Nigeria? How do you prevent applications from uploading your tax data, but still allow them to upload homework documents or cat pictures?

Third, why does the author assume all files of a given format are rendered the same way? HTML can be rendered to screen, or printed, or spoken, or indexed, or analyzed. Even assuming a graphical output mode, users might opt for large text, or plain CSS, or with additional information added. Sorry grandma, you don't get to use the internet any more, file formats are too inelegant.

Finally, though I know the FSF isn't well-regarded on HN, this model is the polar opposite of Free software. Instead of users having absolute control over what runs on their system, they have practically *no* control at all. They surrender everything to the first webpage they open, and to whatever agent is controlling that page at a time. Software is now pushed to users in a closed, proprietary blob, beyond any reasonable modification or adaptation. Short of running tcpdump on another computer and disassembling the output, you might as well be at an internet cafe.