> Anything involving trust cements the incumbents or at least creates a force to an outcome of few players.

I don't think that's even true, unless you're using "trust" as a synonym for centralization.

Suppose you had actual competing app stores. Google doesn't control which ones you use; you can use Google Play or F-Droid or Amazon or all three at once and anyone can make a new one. You could get Android apps through Apple's store and vice versa. And then you choose who you trust; maybe you only trust F-Droid and Apple and you think Google and Amazon stink. Maybe you install 90% of your apps through F-Droid but are willing to install your bank app on GrapheneOS from Google Play because you trust your bank and you also trust Google enough to at least verify that the bank app is actually from your bank.

This is the thing that doesn't help the incumbents, right? The bank and the customer both trust Google to distribute the bank app but Google isn't allowed to prevent the user from trusting F-Droid for other apps as a condition for getting the bank app from Google Play. You can have trust without centralization.

▲

danielmarkbruce 8 hours ago | parent [-]

You have given a situation where there are a 3 players - a very concentrated market. Give an example with 30 players and think through all the implications for all the actors. You'll quickly realize it's a total disaster. Building broad trust requires scale on some dimension.

▲

AnthonyMouse 7 hours ago | parent [-]

How is it in any way a disaster?

Consider how Linux distributions work. Every distribution is distributing variants on the same kernel and utilities, but there are hundreds of distributions and dozens of popular ones each with their own repositories. You can choose whichever you like, and make a different choice than someone else.

Coming in at #31 on DistroWatch is a lightweight distribution called Alpine Linux. It's popular on things like firewalls and VoIP servers but is rarely recommended to ordinary users because that isn't its niche. It doesn't matter that most people haven't heard of it because the people relevant to it have. It's fine for things to have a niche, and the people in that niche are the only ones who need to be familiar with it.

Meanwhile around half of Linux users use Debian derivatives. Debian and Ubuntu are very similar, but their repositories are maintained by different organizations, so even when choosing between two things that are nearly the same, you have different options.

And the distribution is not the only place to get software. Maybe you like a stable distribution in general but you want the bleeding edge drivers for your GPU. You can add the repository for the hardware vendor and still get everything else from the distribution. The vendor doesn't even need to maintain their own full distribution to have enough of a reputation for people to make an informed choice about where they want to get their drivers.

> Building broad trust requires scale on some dimension.

The flaw is in assuming that broad trust is a requirement. Narrow trust is good.

	▲	danielmarkbruce 5 hours ago \| parent [-]
		The long tail of linux distributions work precisely because they need very little trust and are consumed by highly technical users who can verify all manner of things themselves. They especially don't require multi-party verification. Broad trust is required in lots of situations. Hardware attestation, financial clearing networks, or even physical supply chains. Ie, you have multiple independent parties who need mutual, verifiable trust to operate. Establishing that requires transaction costs like audits, SLAs, legal liability, and cryptographic integration. The economics don't work for 30 different players to cross-verify each other. So, we have oligopolies...