Looks tempting!! No doubt. But there are many parsers and scrappers out there which will fail immediately due to this.

What about social media parsers like "Facebook Lint" and "Twitter Validator"? I think Google is using this syntax for their internal use which will later compile and use a proper html syntax.

If you are using their tools, doing html their way probably won't hurt. If you want to still keep your code compatible with multiple browsers and other tool sets, just keep doing it the old way

Actually, this style is an unofficial/WhatWG standard. You should only write the mentioned tags when you need to add attributes to them or when they are important for page flow. The official HTML standard allows for this kind of parsing and any parser which cannot deal with it is incomplete. So yes, I fully support and make use of this style guide. It allows me to safe some extra bytes in situations a minifier cannot decide because it might change the result.

This doesn't increase performance much, and it's a matter of so few bytes. If you want to optimize in this category, remove all line breaks from your output source and compile all ID and names down to obfuscated short ones.

IMO, Relevant only to Google projects, mostly used by Google companies and friends. APM is what Google wants us to use instead.