Emphasis added because I feel like I addressed this in the paragraph immediately after the one you quoted:
> Plus, as me and others keep saying, Redis (and other proper queues) isn't a complex addition to your infra
I'm speaking out of experience and, as I already pointed out in another subthread, Postgres is far more complex than Redis. Consider the presence of "DB Admins/Specialists" and the lack of counterpart thereof for Redis and other queuing solutions.
Of course, if queues are not central to how your platform operates, you might be able to get away with Postgres. I still advise using Redis as a reasonable hedge against someone famous tweeting organically about your service because in this case, you don't want your DB to go down because some queue table had a surplus of transactions (or vice versa).
Not to mention, at an early stage, your tech decisions set precedents for the team. Maybe you have 10K users with a low qps but soon you are sending marketing emails to them and your system has periodic bursts of queue activity for all 10K users at once. When discussing this marketing "feature" rarely anyone thinks, "Hey we can't do that with our Postgres queue", rather "Yeah I saw functions in our codebase for queuing---this is doable". This is a small effort but a huge technical investment for later on.
Early stage startups die because of lack of PMF. Diverting focus and resources away from finding PMF kills companies. Most companies should focus on the product, tech debt be damned.