Cozy Hypertext for the Dark Forest Web

We need to reinvent hyperlinks in a cozy idiom for the contemporary dark forest online environment before web2 platforms succeed in their decade long quest to kill them (though they don’t deserve all the blame for hyperlinks being increasingly suboptimal constructs). I trace the origins of this effort to Google’s Amp product which had “faster mobile user experience” as the casus belli for the disproportionately belligerent war on links, which is now starting to look like the Drug War in the US. Now Google routinely hijacks links for shitty and unnecessary eager highlighting and PDF handling. The view of hypertext culture I shared 15 years ago in The Rhetoric of the Hyperlink, which was quite popular at the time, now seems hopelessly idealistic. When I just googled for that link, ironically, I had to strip out some shitty highlighting crud. The war extends even to cozy internal linking within a blog to the extent you rely on external tools, such as by using Google as your internal search tool. Which is why cozy products like Roam and Notion wisely choose to build seamless internal search-and-link-while-writing author experiences (AX) that can’t be “improved” by Google.

The disingenuous philosophy in support of this war is the idea that URLs are somehow dangerous and ugly glimpses of a naked, bare-metal protocol that innocent users must be paternalistically protected from by benevolent and beautiful products. The truth is, when you hide or compromise the naked hyperlink, you expropriate power and agency from a thriving commons. Sure, aging grandpas may have some trouble with the concept but that’s true of everything, including the friendliest geriatric experiences (GXes). My grandfather handled phone numbers and zip codes fine. URLs aren’t much more demanding and vastly more empowering to be able to manipulate directly as a user. Similarly, accessibility considerations are a disingenuous excuse for a war on hyperlinks.

A useful way to think about this is the interaction of the Hypertext Experience (HX) with Josh Stark’s notion of a Trust Experience (TX), which needs to be extended beyond the high-financial-stakes blockchain context he focuses on, to low-stakes everyday browsing. We all agree that the TX of the web has broken and it’s now a Dark Forest. The median random link click now takes you to danger, not serendipitous discovery. This is not entirely the fault of platform corps. We all contributed. And there really is a world of scammers, trolls, phishers, spammers, spies, stalkers, and thieves out there. I’m not proposing to civilize the Dark Forest so we don’t need to protect ourselves from it. I merely don’t want the protection solution to be worse than the problem. Or worse, end up in a “you now have two problems” situation where the HX is degraded with no security benefits, or even degraded security.

Because of course it benefits certain parties to increase the dark-forestification (I find this to be a broader, more well-posed, and analytically more useful idea than Cory Doctorow’s polemically overloaded but similar enshittification frame) rather than actually address it, so you’ll cling tighter to their security theaters. While they continue to externalize actual security costs to you via “identity protection” products. But that’s a rant I’ll save for another post.

The question is what to do about it. The low-agency response is to retreat to a defeatured cozyweb of messengers flooded with screenshots and Shares From Nowhere at worst, and an limited archipelago of cozy products with improved local HXes at best. What Bruce Sterling called favela chic. The Web2 corporate platform responses of in-app browsers, link obfuscation, and inbound link redirection to apps over browsers, is basically an enclosure movement targeting the HX commons, actively kept more dangerous and untrustworthy than it needs to be, to detain you in-app through fear when you can’t be retained with actual utility. What Bruce Sterling called gothic high-tech.

There is also the retreat from pURLs (pretty URLs) to ugly URLs (uURLs) with enormous strings of gobbledygook attached to readable domain-name-stemmed base URLs, mostly meant for tracking, not HX enhancement (in fact uURLs are a dark HX pattern/feature if you’re Google or Twitter). Even when you can figure out how to copy and paste links (in 10 easy steps!), you’re forced to edit them for both aesthetics and character-length reasons. And this is of course even harder on mobile, which suits app-enclosure patterns just fine. In this arms race for control of the HX, we users have resorted to cutting and pasting text itself, creating patterns of useless redundancy, transcription errors, and canonicity loss (when transclusion is now a technically tractable canonicity-preserving alternative). Or worse, screenshots (and idiotic screenshot essays that need OCR or AI help to interact with) that horribly degrade accessibility and create the added overhead of creating alt text (which will no doubt add even more AI for a problem that shouldn’t exist to begin with).

Grim irony: One of the best reasons to kill the HX is if we get to viable content-addressing at scale (things like IPFS, where you find things by identifiers based on their content rather than location addresses). But indiscriminate copying and screenshotting is already creating a sort of ersatz content-addressing scheme. With AI and OCR, we may not need formal hash-based content addressing at all. This is a bit like saying why get on a Mars rocket when you can just go for a nice walk on Earth, but oh well. Speaking of content addressing, someone please invent a pretty version with meaningful-name aliasing that travels with the hash itself.

The only good reason to mess with traditional hyperlinks now, absent usable content addressing, is if they point to targets that require special handling that generic browsers can’t handle. But this is still weird. Substack and GitHub apps trap their inbound links by default for no good reason mostly. Google Docs mobile apps are worse than their browser versions. For most use cases, being in the app is worse than browser. This situation rhymes with the successful battle to kill Net Neutrality. The good reasons for moving past NN, like the different technical characteristics and demands of streaming video and VoIP packets relative to the default of text and image web packets, were used disingenuously as an excuse to try and kill the good features of the commons model as well, in pursuit of profitable enclosure drives. Instead of traffic-constraints-sensitive NN 2.0 we got enclosures, legislative battles, and seething progressives unable to see past narrowly construed holy cow affordances.

There is a general pattern here: Just like comparable privately owned products and services, public commons and protocols of course have their flaws and limitations, and need innovation and stewardship to improve and evolve. But if you’re fundamentally hostile to the very existence of commons goods and services, the slightest flaw becomes an attack surface and justification to destroy the whole thing. It’s not a tragedy of the commons problem created by participants in it; it’s a vulnerability of the commons to outsiders problem. A technical warfare problem rather than a socio-political problem.

Traditional progressives try to mount purely ideological defenses against enclosure attempts, bringing codes of conduct to innovation arms races. Treating the commons as an always-already beautiful thing that never needs any fixing or innovation beyond tiny incremental advances that can be agreed upon by toxic tyranny of structurelessness governance models. They do not even like people doing a spot of public-spirited BDFxing because it triggers fears of autocratic private actors looking for enclosure targets. Every improvement attempt stalls due to resistance rooted in political principles shaped by rigid understandings of the underlying media. They let their socialist fear of competition and capitalism (which are not actually necessary features of commons) turn them anti-innovation. Or worse sociopath leaders exploit the sincere fears of their peers, which they don’t actually share. This is the real tragedy of the commons — politically ambitious, tech-hostile stewards maximizing the personal political gain they can extract from the commons in the name of stewarding it, driving it into premature technological ossification. A stewardship theater on the defensive side to pair with security theater on the offensive side. I think the once-promising Fediverse has entirely succumbed to this dynamic. It’s a real tragedy. Premature ossification is the new root of all evil, and it’s not limited to Bitcoin.

In the case of the ongoing killing of hyperlinks, commons stewardship theater also contributes (for example, via over-aggressive cozyfication by using robots.txt settings to block search indexing). I’m sure there are more examples.

I think with the right tech-positive attitudes, and a co-option of capitalism within powerful public protocols, the commons can actually outcompete and out-innovate the private sector. The first growth spurt of the internet is my existence proof. The commons need not remain fearfully on the defensive, reliant on threadbare progressivism tactics like Codes of Conduct invented and enforced by petty communist Karens. It can go on the offensive and win on better technical ideas, built better, and deployed faster. Sufficiently powerful open protocol drives can swamp petty enclosure attempts by platform satraps, so long as they’re not being “stewarded” into premature ossification before they actually prevail.

One of the major successful gaslighting campaigns of traditional neoliberalism-style private capital is that it embodies Darwinian competitive spirit better than any alternative. This is not actually true. Private capital is actually pretty bad at competing and innovating outside of certain narrow and fragile regimes. In many regimes, private capital is most likely to be the loser of unbridled Darwinian innovation competition. The task of critics of capitalism is to see past the gaslighting and move the competition to regimes where they can win on the merits of their innovation advantage. When traditional capitalism sees this starting to happen, it tends to start leaning monarchist (not merely monopolist), via enclosure tactics. When you need to control an overly creative crowd that’s starting to out-innovate the capitalists, you invent a king to strike the fear of gods into them.

I don’t know the right offensive strategy to save and improve the public hyperlink commons in this specific war, but I have some basic ideas:

• Don’t fight the Dark Forest but don’t fuel it either: Accept that the median link is dangerous
• Design TX before HX, HX before UX and AX
• Design the GX/Accessibility separately without compromising the default HX
• Build on the “external link” warning pattern (Discord) but tunable to be less nanny-like
• Don’t make in-app browsers. Help users use their default browsers better.
• Support pour-over links (finest bit of WordPress AX)
• Build a seamless internal-linking HX and AX (Roam, Notion)
• Work towards transclusion to the extent you can
• Support pURLs, stripped URLs, next-gen link-shorteners
• Support one-tap clean-URL copying on mobile
• Protocolize your product to generate URLs liberally, visibly and beautifully
• Make your TX commitments discoverable and verifiable, and expose them
• Work on decentralized, extended, user-configurable trust environment protocols beyond just authentication standards like oAuth.

There’s more at stake here than your personal convenience or annoyance. The vast value of the Web, on which its newest layers are being built with AI, were the product of the generativity of a strong, retail-user-friendly public HX. Yes it’s obsolete and badly in need of improvement and subordination to TX, which is more fundamental in a Dark Forest world. But a strong HX rooted in a commons is still the biggest engine of value creation online. Even if/when mainstream crypto becomes a reality, if the web remains valuable at all, it will be because the hyperlink has remained the basic unit of currency. Once you lose the hyperlink, you basically lose everything else.