What makes it different is that as a profession, we have decided that Github is nice, good, and ubiquitous. Unfortunately, the portals that he's describing are crappy, bad, and balkanized.
As for the other half, they're asking him to comment, so fully anonymous access simply doesn't work. The fact that Github is publicly viewable would also be part of what's nice about it, thought I obviously wasn't explicit about that.
How so? Just give a link with the equivalent of Google Analytic's utm_source query parameter. Or a link that creates/uses an account that they internally tag as "this is Scott Aaronson".
They have his name, his email address, and can convince him to click a link. That should be enough.
Then you find that GitHub forbids multiple identity contexts for free accounts in their terms of service ("One person or legal entity may maintain no more than one free account"), so if multiple people want you to do things on GitHub but you don't want to mix the activities together, your options are to pay for a monthly subscription (possibly forever) or break the rules and hope your ability to continue participating doesn't suddenly get taken away if they catch you.
Centralization of interaction by type creates social paradoxes like this.
It would be as if part of your job required code reviewing open source projects in your area of expertise, even where you hadn't previously contributed.
I think Github/Gitlab/Bitbucket have shown us that you can do better than emailing patches. I think the academic world could do better than just emailing crap (the arXiv has already done this, but not for full on peer-review). It's just that the current tools are not good.
I think Aaronson is mostly right to tell off these journals. However, I think the idea of tools that improve peer review is a good one.