It would definitely be cool if they open-sourced the platform, but, in the meantime, you might be interested in this project:
http://code.google.com/p/openmobster/wiki/learnmore
Personally, though, I don't mind paying for a service like Parse. If someone is willing to not only build the features I need but also to wear a pager and get up at 3am just to keep my users happy, I'm more than happy to give them some of my money.
My only wish was that I could place an easy API layer between Parse and my app, because I would want more fine grained control over how data is mutated (even with ACLs, someone with write permission could mutate data and corrupt it if desired).
There's a REST API, but I don't think I would want to use that as much.
Here's some data I exported from one of my Parse accounts: https://gist.github.com/2475461.
Deploying APIs in the cloud for primary use in mobile apps is an oft-repeated/able task, and it doesn't seem like there's anywhere near enough value in the Parse API on top of what you get from the native SDKs and a solid combination of open source stack software supported by an enterprise track record more promising than a Series A round a few months ago.
Also, as the comments here suggest, there are a lot of slight modifications to the default use case that reward having solidly built a stack not built on yet high-level PaaS abstraction. Sorry, but a JSON representation of your data is the low bar, not some instant proof that the product is magic.
In terms of value ... I'm really not sure how you lose - if your app is a flop you're going to coast along on the free plan, if it's a success you can focus on your app rather than your stack, and you can migrate away from them when you feel like it.