Hm, this both potentially solves a sloppiness quality problem and creates a (second) gatekeeper problem. I guess since effort-to-token is far less than effort-to-paper, the relative degree of gatekeeping will probably be less. Nonetheless, the potential for it to provide leverage for an editor to encourage "quality" that really is just justifying an editor's perspective shouldn't be entirely discounted. Editors do already have a large degree of control over what is published, but at least it's transparent that they do, and there's no expectation that the editor is tilting the review process except by who they choose to review. Every move towards less transparency should be made thoughtfully, and if reviews are going to remain private (to e.g. prevent recrimination), then any reasoning behind awarding or withholding of tokens would have to be also.