Email Reinvention: The Design Space Is Real but Network Effects Win

A post titled 'Modern email can be built from borrowed parts' proposed a design that borrows Signal's message request model, domain-based identity, and other primitives to build something better than SMTP without replacing it. The thread was genuinely enthusiastic but also realistic. The line that captured the tension: 'The network effects make email hard to replace. Virtually everyone online has an email address. If this could include a migration path with backwards compatibility with SMTP, I think it would have a chance.'

This thread surfaces every year in different forms, and the same structural problem always ends the conversation: the network is the product. A better email design that cannot interoperate with existing email is not a better email, it is a chat app with delusions.

The interesting thing about this particular proposal is that it explicitly tries to preserve SMTP compatibility as a fallback, which is the only design decision that actually matters for adoption. Whether the rest of the design is good is almost secondary.