* Mongo is only easy to learn. Beyond simple demos, it gets harder and harder to use as projects evolve i.e. you have to do a lot of work yourself imo this is a common problem with nosql datastores that isn't exclusive to Mongo
* "Document databases can be great and flexible solutions for some kinds of projects": Postgresql has been able to work directly with JSON for some time now. There are also other document datastores that are more reliable than Mongo
* Scaling with Mongo is difficult, specifically the crazy setup. Even if you set it up properly, the results don't tend to match the marketing https://aphyr.com/posts/322-jepsen-mongodb-stale-reads
* "arguably quicker to get a project off of the ground with than traditions RDBMs" unless you're using Meteor, I'm also going to disagree here. Most frameworks target a relational database by default. Developing by convention tends to get you off the ground much faster than using something more specialized and niche
I doubt they meet in the middle, but both have great use cases. scaffolding out a quick isomorphic js app is great for mongodb for example. pg is faster and more robust.
its just tech, depends what you are optimizing for
Not sure what that means, but scalability is the worst thing about mongo (though my experience with mongo is all from ~2.5 ish years ago). As soon as your working set of indices gets bigger than memory, performance falls off a cliff and your entire app grinds to a halt. In my experience, mysql and postgres have a more gradual decline in performance so you have some time with only mildly degraded performance to figure out a solution (plus they have more options for tuning which can buy you more time).
There are some things MongoDB does fairly well:
* MongoDB is really easy to use
* Document databases can be great and flexible solutions for some kinds of projects
* Documentation is fairly good so learning the basics isn't too hard even if you know nothing about it
* scales fairly well at the initial stages
* arguably quicker to get a project off of the ground with than traditions RDBMs, which might be the most important consideration for any startup even if a complete rewrite would eventually need to take place
That being said, I've used MongoDB significantly before and it wouldn't be my first choice for most types of new project: PostgreSQL probably would be