On the scope of the web, particularly web browsers.
While I don't agree completely, the points are valid, especially on the ever-expanding scope. Even implementing a complete HTML/CSS rendering engine, not even mentioning interactivity via ECMAScript and DOM, is a herculean task; when you couple that with a JS engine, it becomes near-impossible.1
On the other hand, web browsers might as well be the most generic and application-level-developer-customizable software conceived until now, with a solid security model and compatibility. When you take any single application in isolation, few can be proud to be as flexible as web browsers.
Though Drew tends to paint this in a more of a negative light, though I'm not claiming this is the entirety of his opinion, I vouch for the flipside. I'm not particularly proud of being primarily a web developer, if such a title even exists, but I can attest that I can do a lot with the tools that the web provides, at least within certain contexts and sets of limitations.
Also, it will never cease to amuse me that a significant complaint people had about this article was that the word counts were off. Of course some people need to have everything be factually precise, regardless the intended sentiment.
1 Though how this came to be is also an artifact of the system web browsers exist within; there has always been a pressure to add features to web engines, and the standard essentially is just a description of what engines do currently. Implementing a new engine essentially requires effort that is equivalent to what has been cumulatively put into the existing engines since 1995 (!). No wonder nobody else tries—because without engineering resources that only a few companies in the world even posess you'll be stuck playing catch-up with a 25 year gap, and which will probably not grow smaller by the day, quite the opposite.