Okay, "decentralized hypertext", not distributed. Same problem still exists: servers that run an HTTP service would need to accept and track backlink pings. This was done a decade and change ago with Movable Type's trackbacks feature and that was, as you guessed, a massive spam vector.
You can, of course, construct a backlink system by indexing pages. That's basically how Google works and is probably the only use case I can think of where backlinks help more than they hurt. In fact, your distributed link database is basically half of what you'd need to build a search engine. The other half being, of course, an oppressive antispam system. The only reason Google works at all is because they frequently adjust their ranking algorithms or shove them into piles of linear algebra nobody can comprehend. No clue how you do that in a distributed manner, though.
> That's basically how Google works and is probably the only use case I can think of where backlinks help
Which also opened up for a massive spam problem, whic is obvious in retrospect. The only reason Google sort-of works is because it is not a protocol but a centralized service, outspending bad actors with a massive amount of warm bodies involved.
It is a common theme of well thought out hypertext system, both Xanadu and the Semantic Web which both have had a massive amount of research poured over them, that they aren't designed to operate in an adversarial environment.
Trustless systems such as the bare bones hypertext that is the original web didn't have this problem, in part because it wasn't ambitious enough. This can be seen to validate worse-is-better, or how vast permissionless systems can scale, or simply a stroke of luck. No matter in which perspective, it was a massive success.
> No clue how you do that in a distributed manner, though.
Distributed blockchains (Bitcoin, Ethereum, etc) solved this with Proof-of-Work, which is derived from the concept of an "e-mail postage stamp" to fight spam by forcing SMTP senders to compute an expensive hash. Similarly, if each backlink "cost" a PoW unit to earn Googlejuice it would probably be a useful metric again.
...or not: because that means indirectly placing a dollar-amount on a backlink, and well-funded marketing companies are more-than-willing to shell-out for that. Just an idea though...
Anyone want to start a "Xanadu 2: Electric Boogaloo" with a PoW algorithm?
I have no faith in "blockchain" as a component of a distributed system. Specifically: any system where mutations are guarded behind some thing of value as a consensus mechanism will trend towards centralization. (Hash chains are perfectly fine otherwise.)
PoW is not a distributed system in the slightest. The moment someone creates an accelerator for it, it's done. The set of people who can meaningfully interact with the system as a distributed system shrinks to the very small set of people willing to spend loads of money on ASICs; and the far larger set of people with CPUs is entirely crowded out. "One CPU, one vote" becomes "one ASIC, one million votes".
PoS is at least less of an environmental catastrophe, but it now explicitly involves capital. And here's the problem: capital is inherently centralizing. The idea that we can decentralize or distribute capitalism is absurd, because that's not how economies of scale work. Hell, the original distributed system, free markets, have been routinely subverted time and time again to the interests of those who already have capital in those markets.
I know of no other proof-of system other than work and stake; and both are hilariously inadequate for their intended use cases.
> This was done a decade and change ago with Movable Type's trackbacks feature and that was, as you guessed, a massive spam vector.
This is not an insurmountable problem. You can choose to accept trackbacks only from sites that can point to endorsement by some mutually trusted 3rd party. This is ultimately how the Fediverse works - bad actors will get booted from the most popular networks and find themselves unable to 'push' content to the network. And modern federated-web standards include the "trackback" case, e.g. via WebMention.
> You can choose to accept trackbacks only from sites that can point to endorsement by some mutually trusted 3rd party. This is ultimately how the Fediverse works - bad actors will get booted from the most popular networks and find themselves unable to 'push' content to the network.
I suppose this would have worked in the 2005-ish era when
MovableType, Pingbacks/Linkbacks, and Web 2.0 was all the rage, but in today's internet culture and its very partisan politics (...of a distinctly American flavour...) permeating everywhere, the concept of "endorsement", even in a technical sense, can be framed as a political gesture, even if the basis of an endorsement can be defended on purely technical grounds. We saw this happening with the "Twitter Verified" badges controversy (as the identity-verification of an account is a purely technical and factual thing, but because Twitter chose to use the imagery-of-endorsement to indicate verified accounts it led to... well, that's another discussion). My point being that even if a pingback/linkback/trackback system can accurately filter-out (commercial) spam, maintainers of those websites would likely still want to curate pingbacks anyway - which means still having to actively moderate/audit that, and who wants to spend their time doing that?
You can, of course, construct a backlink system by indexing pages. That's basically how Google works and is probably the only use case I can think of where backlinks help more than they hurt. In fact, your distributed link database is basically half of what you'd need to build a search engine. The other half being, of course, an oppressive antispam system. The only reason Google works at all is because they frequently adjust their ranking algorithms or shove them into piles of linear algebra nobody can comprehend. No clue how you do that in a distributed manner, though.