After patiently waiting for its release, I finally got my hands on a copy of Cory Doctorow’s book Enshittification: Why Everything Suddenly Got Worse and What to Do About It. I am still getting through the work, but it is, so far, an expertly led distillation of many of the litany of frustrations we have lived through on the modern internet. Much ado has been made about Doctorow’s book in the technosphere, with The New Yorker, Vox, The Economist, and Wired, among others, picking up on the work. Doctorow articulates all too well the things we knew were wrong in the first place, and backs them up with a predictable vim and vigor.
If you still do not believe, somehow, that the pattern of:
- A platform, using a surplus of investor cash, is good to its non-business users
- A platform pivots to prioritize the needs of commercial or business customers over regular schlubs
- A platform turns into a giant pile of shit
Is real, I recommend you try to live the example by attempting to participate in today’s job market, which is almost entirely facilitated by online platforms like (especially) LinkedIn and Indeed.
Let me take you back to when I started my career and try to further explain on a technical level why platforms always seem to attenuate value for individual users. In 2008, if you wanted a job, you had a few options, in order of effectiveness:
- Know someone
- Go on Craigslist
- Apply directly to the job
There were a few job aggregators at the time, like Monster, but not many. It was unusual to be recruited for a job, or even talk to a recruiter at all in the process. Part of that is because it’s not typical to be recruited for entry-level roles anyway. But as I continued to move up in my career, the pattern remained the same. In fact, the first time I spoke with a recruiter for a role, a “recruiter screen,” was in 2022. In each case before that including my two longest-held roles, while emailing an HR person as part of the process was typical, they didn’t have a “recruiter” title and certainly weren’t conducting any interviews.
How did it come to be that we now hear horror stories from applicants about how they’ve applied for “4000 roles in 6 months” (don’t do this), and from hiring managers that they have received “4000 applications for a single role in less than 3 days”? I’m being slightly hyperbolic with my numbers here, but not much. How did it come to be that to land a successful role, in many cases, takes 5 or more rounds of interviews (here, I am not being hyperbolic with my numbers), and that that is increasingly typical?
To answer these questions means answering this one: what is the thing a computer enables us to do?
The answer has three parts, all rooted in fairly basic mathematics: the first is that the computer enables us to make a digital asset. Digital assets are typically significantly easier to manufacture than physical assets, so there’s value there. Nobody would argue that using a word processor is more difficult than handwriting.
The second is that we can transmit that digital asset electronically over a computer network at an extremely low cost.
The third part, however, is the key to unlocking the true value realized in the first two parts. The marginal unit cost of making a copy of a digital asset is almost zero. These three parts, the “value basis of computing,” are what have led us here, to the Information Age.
This model works exceptionally well for sellers and buyers of digital assets. The market for in-app cosmetics in video games, for example, is extraordinarily successful because a seller can charge a high price for something they really only made one time.
Consumers are clearly willing to pay that price, as developers pump out continuously more absurd cosmetics to consumers in what I can only reasonably believe at this point is elaborate trolling, although some would just call it “late-stage capitalism,” as if anyone knew what the difference was.
This model works well for digital assets, but not for real-life assets like jobs (or romantic partners, which is a topic for another day) when the assets are facilitated by platforms like Linkedin. LinkedIn has very much turned into a giant pile of shit, and it would be crazy to posit otherwise. LinkedIn’s value proposition originally set to connect like-minded professionals and has since evolved into a job search platform where paid B2B tools like “Sales Navigator” and “LinkedIn Recruiter” proliferate alongside a premium subscription that suggests you’d be a “top candidate” for an apprenticeship in weasel breeding.
Contributing to the rapid enshittification of LinkedIn and the unmitigated disaster of the current job market, especially in tech, is an uncomfortable truth rooted in the value basis of computing, which – remember – is only fully realizable with digital assets. In computing, we call this an M to N, or “many to many,” relationship. It doesn’t matter if 1 person wants the Beavis and Butt-Head costume or if 100,000 people do. The asset is created, priced, bought, sold, and distributed digitally. For scarce, real-world assets like jobs, the relationship is 1 to M. One role, many applicants.
So we have a scenario where we’re in violation of the third principle of the value basis of computing because a given job opening usually exists for a single candidate, but the other two principles still apply for the hiring organization, and all three principles apply for candidates. The only outcome of this scenario is a race to the bottom where applicants compete less with other applicants and more with the tools used by the organizations that are trying to hire them in the first place.
They do this by using their own tools. And the hiring org, overwhelmed, responds with more tools. Each generation of the arms race obscures the value on both sides until we have AI interviewers interviewing AI candidates, and the qualifications of the applicant matter less than the ability for the applicant to have access to paid agents hosted in the same cloud computing infrastructure operated by the companies who own the platforms.
Oh….
If you have two candidates: A and B, and A is objectively more qualified than B, but B is more willing to conform to “the algorithm” in the hiring process, what value did the platform actually generate when B gets hired?
We have hijacked the intended use of the computer and optimized it for density and volume of transactions, but not quality and outcomes. Platforms exacerbate and amplify this problem because they have strong financial incentives to do so. LinkedIn would like nothing more than to sell you LinkedIn Recruiter or Sales Navigator because you’re overwhelmed with leads and/or applicants. LinkedIn actually doesn’t have much of a financial reason to help candidates find jobs at all, because for many people it means they spend less time on LinkedIn (Doctorow goes into detail about network effects and ultimately how petrified platforms are of their collapse in his book).
I am not a recruiter, but I suspect based on the LinkedIn outreach I have received from recruiters that the quality of data offered to recruiters for sourcing on LinkedIn is only high enough to maintain LinkedIn’s status as a job website. Anecdotal evidence, and there’s a lot of it, backs up this claim.
The solution to this problem is actually to introduce friction in the process and return to a paradigm where candidates compete on quality. People offer this as a solution to the problem all the time, they just don’t realize it. The solution they offer is telling someone “your network is your net worth,” and to try to find roles outside of digital platforms and in real life to bring the M to N problem closer in balance.
A second solution is to avoid the platform, but actively approach recruiting firms, preferably local ones, who you can establish a trust relationship with and send roles to you.
A third solution to this problem is to charge applicants a small fee to apply. The fee is refunded automatically when the application is finished.
The final solution is one many will not like, but I see it as the most likely to be implemented outside of in-person networking: implement a challenge-response process using a third-party verifier to authenticate the identity of an applicant at the time of application. Privacy advocates will poo-poo this scenario, and rightfully so, but considering the influx of bogus applicants from North Korea and other adversarial nation-states, I again see this as the most likely outcome, even though right now the practice of asking would be seen as scammy.
If all of these solutions sound like awfully bitter pills, that’s the point. You have been tricked into believing anything that can be done faster and more efficiently on a computer should be. This idea, academically called the “Technological Fix,” prescribes technology to solve problems that technology is not actually well suited to solve.
Platforms do not want to take the bitter pill. There is no dopamine to be had by paying to apply for a job. They want to be as efficient and frictionless as possible so they can sell you more services and serve more ads. They don’t care about the diseconomy of scale. They will certainly not be trusted to process a payment on behalf of a hiring organization.
I’m picking on LinkedIn a lot here because LinkedIn doesn’t really have any competitor. The best competition for LinkedIn is the thing that has been most effective the entire time: in-person networking.
I’m not trying to sound like a luddite. Are computers good? Yes! If LinkedIn (or any other platform) actually was excellent at matching people to jobs and facilitated the process for both job-seekers and hiring organizations, it would be great! Could they do this? Yes!
Can they do it?
No!
Why would they when it would invariably mean people would stop using the platform? They can’t afford to have one-time or limited-time users who churn after they get a job. That’s why LinkedIn continues to add crap features like short-form video of people (and it’s always a certain kind of person) about their “5-9 after their “9-5,” and browser games like Zip.
We cannot continue to slap technology on problems that scale poorly and attempt to force them to scale well. It just doesn’t work. We’re trying it with AI too. A byproduct of this technological shoehorning is, invariably, the enshittification that Cory Doctorow details in his book. I highly recommend you read it for a deeper dive on the pathology and cures for this worsening problem and the rapid decline of the utility of the internet.
Doctorow does go into detail about this, and it’s something I’ve written about before both on this blog and on platforms: the intent of the internet was to be a decentralized network comprised of participants who owned their data and were willing to share it. It takes real work to do this. Not much, but some. There are tools that can help you. I just saw a take on this the other day, on LinkedIn, in fact: platforms will kick you out. They won’t explain why. It’s happened to me! Doctorow enumerates and elucidates on policy solutions to enshittification.
I have one more: take back your data. Start a website. Publish. Incentivize your network away from platforms. Maintain your phone book. Be thoughtful. Friction can be a valuable and useful tool. Engaging in difficult things and going through some struggle should be celebrated, it’s what makes us human. Be your own platform.
It’s up to you.