I think this a great idea!
scott
Why is it so expensive to federate Bluesky?
Mostly because it depends on certain centralized services. You can create your own apps and even host your own content, but the discovery and distribution system has a copy of every post so that it is easy to access by everyone. The positive side is that you don't have to worry about missing replies in the conversation since their centralized database has a copy of it. The downside is that hosting such a massive database is expensive.
A marketplace of apps is actually a good thing. Not everyone wants something based on Twitter (i.e. Mastodon, and similar). Not everyone wants Reddit (Lemmy, Mbin, etc.). Not everyone wants a traditional forum (NodeBB). Not everyone wants a blog (WordPress, Hubzilla, etc.). Not everyone wants Facebook (Friendica, Hubzilla, etc.).
One of the goals is to build increased compatibility between apps so that you can choose which experience you want, yet can still talk to anyone else on the fediverse. Some big players will certainly emerge, but I think that there will always be hundreds of compatible apps.
To create something like this, you would need to federate two components, and optionally a third.
You would also need to create a standard (a protocol) for the app store to talk to the websites hosting the applications.
Application Hosting: Basically, everyone can create their own website with their own apps. That part would be unmoderated, similar to how you can go to a software publisher's website today and download a Windows program. They publish their application and data about their apps in a machine readable way where an app store could take that data and create a listing.
The App Store: There would be open source app store code that allows people to run their own app store. The people operating an app store decide what gets listed in the app store. Some app stores will be for a particular niche while some will attempt to list everything. For example, you might have app stores that only have open source software. This would still make the app store operator the gatekeeper, but what is different here is that anyone could use the same software and set up their own app store.
Curation Groups: This allows people or groups to create their own curated list of approved apps. This provides the app stores a shortcut so they don't have to review every single app themselves. This would allow individuals, communities, associations, and even businesses to create moderated lists of apps they reviewed and believe should be listed in app stores. Mastodon could publish a curated list of Mastodon Apps they recommend. Open source organizations could create a curated list of apps they recommend. The app stores could consume such lists.
People can then choose the app store and the apps they trust. App stores can choose the curated lists they trust.
This is similar to how podcasting platforms work, where a podcaster publishes their audio files and an RSS feed with information about their Podcast, and various Podcast Directories list their podcast. Or similar to how platforms like Steam work, where they list games, many of which can be obtained on the game author's website as well. The key point being that the authors of the apps can get listed in multiple app stores.
Optionally, both the Application Hosting software and App Store can be integrated with protocols like ActivityPub, AT Protocol, or Nomad/Zot protocol for the purpose of sending out notifications to followers who may be interested in updates and news about the apps or the app store. At the very least, it should list an existing fediverse handle where people can follow them.
So, yes, it can be federated.
To be safe and secure, you would want multiple organizations with resources to run competing app stores using this software and protocol. These organizations can be non-profits, cooperatives, or even small businesses. The reason why is because an organization is more likely to have the resources to moderate the list of apps in their App Store, whereas an individual most likely would not, unless that was their full time job.
Whereas anyone who created an app could run their own website with information about their app, and then request to be listed in various app stores and curated lists.