Concerning adoption for web2 developers, the main problem is that it’s trivial for a single canister architecture, but when one want to learn how to do the next facebook, it’s a whole new world.
The actor model is totally alien for the majority of web 2 dev. I have 10 YEO in pure web dev (with some projects with complicated infra and hundred of millions of request/day) and I never encountered it in real life, and with the way stable memory works it quickly became difficult to know what is the best way to architecture it for the problem at hand. The documentation can give cues on how to handle it, but there is no demonstration from start to finish, with an highlighting of the different caveats developers can meet. I play with a pet project at night for fun because I love the tech, but I had to rewrite my pet project several time to finally understand how to scale.
Web 2 is full of people coming from bootcamp (myself included), an when they will search how to do something, they will have the choice between a tech with hundreds of complete exemples, or doing something on icp, with the only exemples being either complete projects on github (unreadable when you just start) or read everything in the official documentation and pray to assemble the pieces of data in the right order.
I think concrete exemples of complete projects (multi canister, multi indexes, dynamically created buckets, maintenance canister), with step by step guide could go a long way to attract web2 developers, especially if it’s done with a familiar tech stack (js/python)
For my specific problem, I created a thread some days ago without any response (it’s my only missing piece of the puzzle):
How to handle filtering/pagination at scale - Developers - Internet Computer Developer Forum