They do improve the design of the architecture, but the question is "Will they improve the design of the architecture more than shipping the product, letting it contact reality, and learning from experience?" There's an opportunity cost - debates cost time, time delays shipping, and shipping is what enables customer feedback. In my experience, customer feedback and actual experience is much more important for improving the design of the architecture than debating with other people. You can try to make your design slightly less wrong by consulting with other team members who are equally as uninformed as you, or you can get it right by going out and getting hard data.