The rising costs of publishing technology

The biggest reason we love and hate the Internet is that it’s been so miraculously cheap to use. But perhaps you’ve noticed that your tech costs are shooting up, or your tech keeps breaking down. It costs more and more to maintain a digital presence and products.

What changed, and what does it mean for publishing?

The story here could fill a book, so I’ll summarize. And if you don’t need even this summary, read the headings and skip to ‘We get what we pay for’.

‘The March of Intellect’ (c.1828) by Robert Seymour (England, 1798–1836)

Investors need a return

Promising new technologies are always subsidised by investors. These calculating gamblers are willing to burn cash in the hope of selling their stake to a bigger risk-taker. In the swells between crashes, investors have made it possible for unprofitable companies to offer us free or heavily subsidised services. Every corner of the Internet has been flooded with them.

But the supply of hyper-subsidised, near-free services started shrinking a few years ago, as more investors started asking for a return, perhaps in response to a slowing US economy. This led many online services to raise prices faster than ever. Like me, you’ll have received more price-raising notifications from your services over the last couple of years than ever before.

AI hype has now hoovered up what was left of those investor subsidies, and any business not selling an AI promise to its investors has no choice but to charge a profitable fee for its services.

As a result, services that we assumed would always be free or cheap are suddenly significant costs.

This, and the factors I’ll describe next, don’t just affect certain organisations. They apply as much to a small company website as they do to the New York Times.

Open-source dependencies demand attention

In recent years, software capabilities have exploded. We can build more and build it faster because millions of developers share their code freely.

It’s a wonderful thing to see people creating code that others can use for free. They do it for the satisfaction of it, and a badge of honour. Even though fund-raising initiatives exist to try to pay some of these people, most software still runs on the goodness of human nature and a measure of bloody-minded grit.

Even the simplest website today depends on pieces of code from hundreds of volunteer developers. And their code depends on others’ code, and so on. We call these pieces of code ‘dependencies’, and they are the best and the worst thing about software development.

These volunteer developers update their code all the time, just to stay compatible with each other and the wider Internet, or to satisfy the endless demands of their users (who generally are not as polite as the developer is generous). Or they do not maintain it, creating headaches for everyone else: just one slip-up deep in the stack can break thousands of websites.

If you need your tech to keep working, you have to devote real human attention to your dependencies. That might be daily or weekly; or monthly at the very least. You can’t just set dependencies to auto-update either, because an update to a dependency is as likely to break your website, or the systems behind it, as your failure to update it.

It costs money to pay smart developers to manage these dependencies. (And, no, AI can’t do this for us, because AI is software, which has dependencies.)

Platforms evolve

One way we’ve managed the rapid expansion of technology is by relying on cloud-based platforms to centralise control.

For instance, we use a service called Gitpod for our production work, because we can centrally control dependencies across our whole team, rather than having each team member manage dependencies on their own computer separately.

But when platforms evolve, we have to evolve with them. For example, as Gitpod makes changes to its own infrastructure, we have to spend days updating our systems accordingly.

Or when the Google Play Store says it’s time for all apps to support newer versions of Android, we have to spend hours or days testing and updating code across different projects, just to keep our clients’ apps in the store.

As regulation evolves, the paperwork comes due

When governments try to regulate the Internet’s problems, as they should, there is work to do whether or not we’re part of those problems. There are regulations that address money laundering, privacy, election interference, public health, child safety, human rights, accessibility, and more. They usually apply first to platform providers (like Google and Facebook), who then change their policies and what they require from website owners. And sometimes the regulation affects us directly.

Soon enough, we’re all fixing cookie banners, filling out documentation, registering DUNS numbers, auditing databases, and updating code.

Given their sensitive nature, these changes can’t be automated or delegated to junior staff. So the implementation costs are serious.

Social costs demand an ethical response

Even when we aren’t forced to change, most of us are ethically compelled to respond to the social costs of the Internet. When your organisation removes Twitter logos, invests more in content moderation, adds alt text to images, avoids paying money to bad actors, or looks for ways to reduce your carbon footprint, you need sensible, experienced tech folk in the room, working through the implications.

User expectations keep getting higher

The gorgeous websites we built five years ago look old already. They might work fine, but as long as everyone else is burning cash on bells and whistles, we’re under pressure to keep up.

Should you feel that pressure? Maybe not, but the urge to keep your products looking fresh and relevant will seep through your organisation. Inevitably, you’ll want to add a button here, refresh a logo there, or – a common refrain – ‘make it more interactive’.

And as you make these tweaks, you’re not just spending money on them, but you’re also adding layers of complexity and fluff that will eventually be so onerous to use and maintain that you’ll have to rebuild the whole thing from scratch.

The careful balance of keeping up without overloading your infrastructure takes skill, which takes more expensive people.

Anxiety about IP requires legal expertise

There was a time, in ancient Internet history, when few people knew or cared about its implications for your intellectual property, or whether you were violating someone else’s. Today, the volume of the IP conversation has risen to deafening.

It’s unusual not to have to care how others use your IP, or how you use theirs. At some stage, someone experienced and expensive has to put time into research and decision-making for your organisation, and then roll that out in your products and systems.

The content flood is infinite

Now that anyone can spew out huge amounts of good-enough content, users have a near-infinite array of choices. Keeping your stuff visible and valued takes much, much more work. What was once the domain of a freelance SEO consultant is now the unending, collaborative, in-house work of bot whispering. Like it or not, we must grok the tides of The Algorithm, and ride them or be dunked.

We get what we pay for

I built my first website in 1999. It took me weeks to hand-code. If you showed that young man what we can do now, and asked him what it should cost, he’d give you a very big number. And he’d be right.

We’ve tricked ourselves into believing that this extraordinary technosystem is cheaper than it should be.

And yes, perhaps it should be more expensive. A cheap Internet was an Internet we took for granted, and one we’ve fallen out of love with.

When technology costs what it costs – unsubsidised by the revolving door of venture capital, and guided by ethics and guardrails – we care more about what we do with it.

It’s also healthier because the humans we pay for that technology can do better jobs without the absurd expectations and perverse incentives of VC investors.

Two of my favourite services, Basecamp and Consonance, have not grown from venture capital and have always charged a decent price for their software. Their relatively small teams produce some of the best, most reliable software you’ll have the pleasure of using.

What does this mean for publishers?

There was a time when we published a book and it was done. We kept copies till they sold out. A reprint with corrections was a fairly simple matter. Old books went out of print and were lost forever. We focused on the next book.

Today, our books are no longer finite projects. They are ongoing technological commitments. They are source files in the cloud; blurbs on our websites; metadata at the print-on-demand distributor; HTML at the ebook retailer; apps on endlessly updating phones; web pages behind the logins of users in databases; and digital assets listed in IP agreements.

So as the age of the subsidised Internet fades, we can untrick ourselves about the costs of this marvel, and consciously invest in maintenance, in-house expertise, and technology built by thoughtful humans with a sensible business model.

A while ago I asked myself, ‘how can my business pay for these exploding maintenance costs?’ But of course we were already paying them: with the non-renewable resource of late nights, panic-driven fixes, and broken systems. The real question is how to turn the invisible costs of degraded performance into the managed costs of regular maintenance.

It’s unsexy, but it’s authentic, and that’s an Internet I can fall for once again.