I completely advocate building whatever you want: competition, evolution, test-bed, self-fulfillment, et cetera. That said—and there is some excellent commentary already—something no one has mentioned yet is the potential loss, to all parties, involved.
Let’s pretend I’m a reasonably seasoned web dev—suspend disbelief if you would be so kind!—and I’m dissatisfied with Template Toolkit. I know all the alternatives, from TT3 in Alloy to Mason and Mojo::Template to Text::MicroTemplate and all points between. I’m informed. I’m competent. I can write tests and documentation and manage a civil mailing list. (It could happen!)
The only major concern about doing a new templating system at this point is, why should I divide the effort? Isn’t even one of the current ones close enough to what I want for me to sink 5% of the time and effort it would take to start from scratch into improving and contributing to it instead? Ten years ago the answer probably would have been no, there’s plenty of unexplored territory. Today the answer is yes, my time and effort would repay me and everyone else better by joining a going concern as a contributor rather than starting over.