Three Angles on the New Web
Three Things #104: January 28, 2024
I spent the last three weeks listing the biggest problems facing the modern web as I see it. I could keep going and come up with several more weeks’ worth of problems, but that would be annoying and most of you would probably stop reading if I did that. It would also be unnecessarily pessimistic and depressing. In any case by now I think I’ve covered the biggest, most important points and it’s time to turn our eye towards the future and towards a solution. As an intermediate step I’m going to attempt to paint a picture of where we might be heading in attempting to solve the problems I laid out.
Sometimes with the very biggest, hardest problems a direct approach won’t work. Sometimes it’s necessary to approach the problem from the side in order to get close enough to bash it on the head. Sometimes you have to do that from several sides in order to stand any chance of making progress. That’s what I’m going to attempt to do here. Rather than try to outline the once-and-for-all grand solution to the web, a task that would be impossible, I’m going to approach possible solutions from three distinct but related perspectives. Once this is done we can try to reconcile them all, put the pieces back together, and figure out how it all fits together.
Thing #1: Better Reading App 📒
There’s more interesting content out there than I can read (and listen to, and watch) in ten lifetimes let alone the one I’ve got. Like I’m sure everyone else reading this, I’m constantly receiving things my friends and colleagues send me to read, and I’m constantly stumbling upon interesting looking content on social media, in podcasts, books, and elsewhere. I read, watch, and listen to what I can, but for lack of time the vast majority of it gets filed in a single place.
Long ago that place was emails to myself. Later I got a smartphone and began saving notes and links on it. Later I moved to an app called Readability, then a few years ago to one called Pocket. Recently I’ve begun playing with a delightful paid service called Readwise that’s a slightly better version of Pocket.
In general these apps and services let me do some degree of tagging and categorization of the links and content I save. I like to separate personal stuff from work-related stuff, I like to tag things that are relevant to a particular project or pursuit (“exercise”, “diet”, “politics”, etc.), and I like to prioritize things (even if I literally never get past the “REALLY SERIOUSLY WTF YOU NEED TO READ THIS NOW!!” tag). In addition to tags and categories, must-haves for me include downloading content for reading offline (I spend a lot of time on airplanes), the ability to share things with others very easily, great search (for finding that article I know I saved six years ago even when I just can’t remember the author or title), and something I’ll call “snapshotting.” Snapshotting means not only recording an immutable copy of a document as it was when I first saved the link (as opposed to when I try to read it—things change and disappear with a depressing degree of frequency on the web), but also doing so without ads, menus, suggestions, or any of the other garbage that tends to clutter web pages.
This last feature is particularly important and particularly hard to find. Pocket and Readwise both do a reasonable job of snapshotting but neither can handle every page and both tend to get confused by even simple things like PDF files and Google Docs. Both apps often revert to “web view” which means you’re back to where you started: ads, menus, and suggestions aplenty.
Another issue is my very strong preference for open source software. Sadly both of these apps/services are closed source and both are centralized which means I have zero guarantees that they’ll keep my data safe, will continue to operate in the future and won’t arbitrarily kick me off without explaining why. I also haven’t found a way to easily export my data from either app. While they’re okay and kind of get the job done they leave something to be desired.
I’d also love for such an app to defeat paywalls. As I wrote earlier in this series it’s not because I’m averse to paying for good content. It’s because I simply cannot handle the cognitive burden of managing dozens or even hundreds of different subscriptions. If such an app found a better, simpler business model and could wrap payment for content with better UX, I’d be all over it.
With that in place I can explain what I want, which I think is really quite simple and not too much to ask for. As a content consumer I just want a better experience saving, organizing, reading, searching, and sharing articles and other types of content. That’s it. If apps like Pocket and Readwise did a better job along these lines, and if additionally were open source and were decentralized or I could run my own backend, I wouldn’t be wasting my time writing this or researching how to fix this. But I think this is actually a great bellwether example of the ways in which the web is fundamentally broken because this is such a simple thing to ask for and yet for frustrating, structural reasons it doesn’t seem possible today.
To reiterate:
As a content consumer, when I come across an article or other piece of content on the web that I want to save and/or share with others, I want to be able to do so with a single click. I want to know that I’ll be able to find that content later in the form I expect.
And as a content creator I’d love if my readers could have such an experience. I’d love if my content could be placed alongside similar, relevant, high quality content, free of ads and paywalls and possibly with something better like micropayments built in.
Is that really too much to ask for? In my opinion expecting the web to let me easily find, consume, save, share, and publish content is like expecting McDonald’s to serve juicy hamburgers. You had one job, bro.
Thing #2: Decentralized Newsroom 📰
Two themes I write about a lot here are the breakdown of trust in traditional, centralized institutions and the tension between the upsides and downsides of decentralization. I’m going to touch upon both of these here.
On the first point, I’ve written many times here about how I’ve totally lost faith in mainstream media. And others have covered this phenomenon much more thoroughly and convincingly than I could so I won’t rehash those arguments here. For the purposes of this issue let’s take it for granted that in my eyes and in the eyes of many people, and perhaps most young people, there are no longer any trusted legacy media outlets. To the extent that I and many of these people trust anything or anyone to help us understand what’s going on, it’s increasingly individual people and individual brands (Joe Rogan and Lex Fridman come to mind), not companies and not institutions.
And yet as I wrote here a few weeks ago there’s one very compelling aspect to the product offered by mainstream outlets like NYT: curation and a bundle of content. There are many fantastic independent content creators who publish their content on the web in one format or another, but—as I keep saying—I don’t have the time, energy, or wherewithal to discover, curate, and subscribe to such content, including the burden of managing those subscriptions, jumping around to many different sites and apps to consume content, etc. By contrast I can open the NYT app and have one single, integrated, comfortable experience and consume all sorts of content across a wide variety of subjects. This experience just isn’t possible on the open web and decentralization and fragmentation make this much harder.
How do we square this circle? How do we create a trusted channel or platform that doesn’t get captured by one radical ideology or another and doesn’t serve its political and corporate overlords (as has happened to mainstream media) but that nevertheless can deliver a delightful, integrated, bundled experience? I think the answer is threefold: an open protocol, decentralized curation, and better incentives.
The open protocol is how you avoid capture. The protocol and its infrastructure shouldn’t be owned or governed by any single corporation or organization. Bitcoin and Ethereum have proven the viability of this model.
Curation is how you achieve bundling and how you ensure trust. There should be many content publishers and curators. Users may choose to subscribe to any subset of publishers and curators. They may choose to curate their own content or they may rely on curators to do this work for them. Some curators may naturally specialize in fields they know a lot about or are especially passionate about: sports, arts, music, economics, technology, etc.
Sound familiar? It should. If done right, this would be something like a decentralized newsroom with different “desks” occupied by teams of curators. Eventually meta-curators may begin to curate the work of many such focused curators together into a single, bundled experience like a new NYT. And due to the openness of the protocol, if any curator loses your trust you can easily switch to another or choose to curate content yourself. This mechanism would keep curators honest and make sure they act in the best interest of their readers/subscribers.
Finally, better incentives is how you make the whole thing sustainable. The mainstream media have collapsed due to broken incentives: first old fashioned print publications began to fail because digital platforms and digital advertising killed their business model, then the new, digital outlets became shitty and have begun to fail because their business model, competing for eyeballs in an ever more competitive online marketplace, also fell apart. It’s time for something radically different.
I haven’t worked out all of the details—we’ll dig deeper into this soon—but at a high level this will only work if readers are prepared to pay for good content and curation, and if content creators and curators are downstream of this revenue. Would readers be willing to pay for such an experience? I’m not completely sure and there’s a lot more research to be done but I suspect the answer is yes, especially given the apocalyptic state of the alternative. If you can deliver a comfortable, robust, familiar experience, something that looks and feels like NYT but is powered by a totally different social and economic mechanism under the hood then I think you stand a pretty good chance of changing the world for the better. It’s also possible to give new users a free budget to initially spend on content and to allow them to earn “coins” by providing a service to the network (such as publishing their own content, or curating, or adding commentary, or sharing, etc.) which can subsequently be spent on content.
What about the other side of the market, content creators? It may sound contrarian but creators should be charged for publishing content, since this imposes a burden on the network and, indeed, on the world. It’s a disincentive for publishing garbage or low quality content. But they should be unafraid because even halfway decent content that appeals to, say, 100 other people would more than cover costs and be profitable. It’s important that this equilibrium be established. And content creators should be able to earn a huge amount from popular content, even without advertising. It’s no different than a shopkeeper needing to pay for a storefront in order to go into business.
The case for curators is a little more complicated. It’s not quite so straightforward how they should be compensated but one simple model is that they should receive a small cut, say 10%, of the revenue flowing from readers who accessed content through their “feed” to the creators of that content.
If we can get even these basic incentives in place I think we could bootstrap a really exciting new economic model and totally reimagine how things like newspapers and magazines are designed and run in the modern era, using the power of cryptoeconomics and better incentive schemes.
Thing #3: Proof of Human Work ⚒️
In the earliest days of Bitcoin it was possible to mine from home. In fact, this was the only way to mine and all miners were by definition home miners! Of course, at the time no one other than a few cypherpunks and cyber-weirdos had heard of Bitcoin and it had no value to speak of. Over the following several years Bitcoin mining went through several phases of professionalization: from home CPUs to GPUs, then to FPGAs, and finally to ASICs. At least since I joined the Bitcoin community around 2017, and probably even before that, it’s been unprofitable to mine from home and it’s only gotten harder and harder over time.
A similar process played out with Ethereum prior to its transition to proof of work in 2022. Ethereum proof of work mining was designed specifically for GPUs. In the beginning people mined from home using the GPUs they already had. Over time, and as the value of ETH increased, mining became more and more specialized so that eventually it was only profitable to mine as part of a pool or an industrial mining operation. Eventually ETH ASICs also began to emerge.
When proof of work was designed there was a belief, somewhat utopian and unrealistic in retrospect, that it would be egalitarian and enable ordinary people everywhere to participate in the economic upside of cryptocurrency by mining from home using existing hardware and bandwidth. Today this idea seems naive to the point of laughability. It seems obvious and inevitable that, where there’s money to be made through professionalization, special hardware, streamlining of operations, etc., professional miners will take advantage of those things and, eventually, crowd out smaller home miners.
We nevertheless launched the Spacemesh project a few years ago with a similarly hopeful, utopian vision: that any Internet user anywhere in the world with any device and an ordinary Internet connection could download the Spacemesh software, join the network, help provide security and validate transactions, and create economic value for themselves. I wasn’t around at the start of Bitcoin or Ethereum but I wouldn’t be surprised to learn that many in those communities had a similar vision at the time. We thought we could defeat the centralizing tendencies that put mining out of reach for home users in Bitcoin and Ethereum due to clever unit economics: in a nutshell, as long as a home miner is mining using hardware and an Internet connection that they already have (i.e., without investing additional money into mining), they will never be priced out by pools or industrial-scale miners who will always have higher costs. This is possible because of Spacemesh’s bespoke consensus mechanism: proof of spacetime mining is much easier and cheaper (in theory) than proof of work mining, which consumes a lot of electricity on an ongoing basis.
While we’ve come a long way towards realizing that vision—the network is live and there’s thousands of home miners on it—it’s proven harder than we expected to ensure mining is profitable and sustainable for home miners in practice. For one thing, it’s much harder to run a node from home than we expected or intended, and it’s become harder over time as the network has grown rapidly. For another, the network has grown much, much more rapidly than we hoped or expected, meaning the rewards for small miners have dwindled.
There are lots of complex technical, social and economic reasons for this trend, and while examining them would be very interesting it’s beyond the scope of this issue. From my perspective the lesson here is that “egalitarian proof of work” mining is much, much harder than I expected even with clever protocols and unit economics theoretically working in your favor. I’m not ready to declare the experiment a failure yet—the market hasn’t yet found equilibrium, it’s possible that the situation may eventually improve, and Spacemesh may yet create value in many other ways—but for now I’m unconvinced that proof of work, however clever, can ever be egalitarian or create much value for home miners.
It’s time to think even further outside the box. As I see it the goal is to design a novel form of proof of work that fulfills three criteria. First, it should require a talent that is widely endowed and that ordinary people everywhere can contribute without specialized training or equipment. Second, its needs to be difficult or impossible for a machine. Third, it needs to be objectively measurable and meritocratic so that rewards are more or less proportional to skill and effort.
This is a really interesting puzzle! It’s one that I’ve been mulling over at least since getting into the crypto space almost eight years ago. A good starting point is “proof of humanity”, i.e., a way for a person to prove that they’re both human and unique. I’ve looked at many unique, crazy schemes from physical movement and gaming to pseudonym parties to web of trust to eyeball scanning. Unfortunately none of these cuts the cake. The problems should be clear: these schemes are mostly centralized and not scalable. What’s worse, they don’t fulfill the third criteria of being meritocratic and objectively measurable and they can easily be gamed and dominated by machines.
I’ve only ever been able to come up with one idea that ticks all the requisite boxes, and I keep coming back to it: creativity, especially with respect to creating and curating high-quality content. This is the one thing that’s meritocratic, that machines just aren’t very good at yet, and that involves talent that’s quite widely endowed. The permissionless rise of bloggers, vloggers, streamers, and other social media stars over the past generation should make this clear.
Let me address a couple of likely objections upfront.
You’re probably wondering, what about LLMs and other “AI” technology? I’ve played with LLMs as much as the next guy and, while they’re good at spitting out text that sounds natural for very specific use cases, I don’t think the technology is yet at a stage where the memes or long form content it generates are anywhere near on par with the best, most creative content generated by humans. Even if it does eventually get there I still don’t consider this a problem because at the end of the day it’s just a tool, like a word processor or a spell checker. Behind every piece of LLM-generated content is a human agent and a prompt, and if that human actor is able to produce better content more efficiently than other people they deserve to be rewarded for that (it still requires creativity). In my opinion we’re nowhere near a world where AI tools are capable of acting on their own behalf.
You may also be wondering how to objectively measure creativity. It’s not obvious at first blush but it’s fairly straightforward as existing social media networks worked this out a long time ago: it can be measured through engagement metrics such as clicks, likes, upvotes, shares, etc. These metrics aren’t perfect but they’re a reasonable first-order approximation and can be improved upon over time.
In a sense we find ourselves back where we started! Starting from first principles it seems like the best model for collecting, curating, and valuing human labor in a maximally decentralized, participatory, egalitarian fashion is… social media. I think the problem with social media today isn’t anything fundamental about this process. It’s rather the way in which today’s platforms are centralized, unaccountable, and privately governed and operated. The incentives are wrong. An open, decentralized social protocol is worlds apart from the way social media operates today. There’s a lot of work to be done to wrap all of this into a protocol that’s usable and isn’t overly complex, and that isn’t trivially gameable, but we’ll work towards that over the coming weeks.
I call this Proof of Creativity (I’m not the first to come up with this term). Decentralized social media is proof of creativity.