How would you decide what to block? Wikipedia or Stack Overflow can be extremely useful but you can waste time there as easily as on Facebook, Twitter or Hacker News.
Any system designed to solve the attention problem will have to be opinionated, blocking specific websites, or at least parts of it, I guess it's an interesting case study for machine learning.
Another helpful one was removing the "next page" links on Reddit (I use the classic version). That way I can check on some things I'm interested in, but not get carried away.
I'm considering doing the same for Hacker News.
For example, if you need StackOverflow it can replace it's UI with only a search box and results list, preventing you from clicking related links.
When you go on reddit, it could asks you which subreddit you need and why before letting you in, reinforcing the reason you need to visit the site. I think it can be powerful - requiring active input from the user as to what his intentions are. If you need to write "I'm really bored" as a reason you go on Facebook, I think there's a pretty good chance you'll end up closing it.
I believe there are hidden patterns that you might be able to break when the product helps you see and understand these patterns.
The main problem I see with the concept is that it has a poor intellectual foundation. I'd like to see a better explanation of why learning is so dangerous.
Overeating, overfeeding.... They share a lot of similarities.
Maybe I'll read 2-5 particularly interesting articles in depth, but even from what I'd call great websites a lot of the content is just noise and not worth the time spent to read. That really becomes apparent when you have a pretty barebones, super focused and condensed list.