13 October 2025 – Baldur Bjarnason

It’s now been roughly six months since I slowed down this newsletter to give me space to think and reconsider.

I’ve done this before – several times in the history of this website, in fact. Sometimes the blog has stayed sporadic for a couple of years. Sometimes only a few weeks passed until I found my motivation again.

Writing just for yourself is its own kind of fun. Whatever else is going on in my life, even when the mental landscape that is my blog or newsletter is lying fallow and slowly regaining its fertility, I still update it occasionally because writing is both fun and a tool that helps me make sense of the world around me.

But writing – any media – only reaches its full potential if it’s done with an audience – not just for an audience, but written as a part of the field or community in discourse with others.

And that kind of writing has to have a purpose. It needs to be of service and add value to the field that frames it.

That’s the bargain. That’s why people read what you write and, occasionally, help support the writer in some way.

When you feel that you’re burning out on writing, one of the possible reasons is that the bargain that’s specific to the writing you’ve been doing isn’t working for you.

When an audience wants anger and hostility, feeding that can be the writing bargain you offer and the audience will often reward you with attention, praise, and waves of hostility that lift you up, as if it were support, because the poison is directed elsewhere.

But the poison is still there. Feeding hostility in others requires feeding the hostility in yourself.

This is why I’ve been struggling for a while now to write constructively about tech and “AI” specifically. Constructive criticism has a purpose, but we’re at a point in the current iteration of “irrational exuberance” where the stakes for the pro- side are so high that nothing constructive registers and what we’re left with are anger and the expert chroniclers of “just how bad is it anyway?”

Academics and researchers who specialise in machine learning and related fields catalogue the hype and promises. The few genuinely numerate journalists, such as Ed Zitron, do the maths on just how big and potentially catastrophic the bubble is. We’re at a point where the finances, environmental impact (which is currently expressing itself mostly on a community level), workplace dysfunction, and outright debasement of the entire software industry’s user experience mean that there is simply no reasonable justification for supporting these tools. They are obviously destructive.

Sensible criticism is unnecessary in this environment. You need chroniclers – the experts who are cataloguing the various examples that show just how damaging these systems are – and you need the analysts – the journalists who can tell us just how many orders of magnitude the industry numbers are off by – but there is little outright need for low-key constructive criticism.

Because the correct response to those who are explicitly collaborating with authoritarians and fascists, as much of the “AI” industry is doing, isn’t to tell them that their tech is likely to cause substantial delays at the organisations that adopt it.

The correct response is to tell them to fuck off and then move on.

Wallowing in the anger isn’t productive. Mockery always works, but I’ve never been much of a humour writer so I already know that I should be leaving that kind of writing to others.

So I had to do a bit of contemplation. I’m not a particularly fast thinker. I tend to linger on details for too long and overthink things, but with enough time I get there and I think it’s time to put this newsletter back on a schedule. I’m putting together something that wraps up my coverage of “AI” specifically, at least for now, but other than that I’d like to continue my explorations of the impact of and problems caused by tech, but with a slight shift in focus:

Software as a mass medium and the effect it has on other media.

Why?

The reason is straightforward: pretending software isn’t a medium, pretending that it doesn’t have more in common with the fashion industry’s tight interplay between disposable materials, faddish pop culture, and a hybrid both bespoke of mass manufacturing than it does with engineering, just doesn’t work.

It doesn’t work for those of us who have been making software.

It doesn’t work for those of us who need to use the software.

It doesn’t work for the societies that have, inadvisedly, come to depend on what’s effectively a bunch of flared jeans designers cosplaying as engineers as the fundamental drivers of their economies.

Pretending this made-up bullshit makes sense doesn’t work any more. At least, it doesn’t work any more for those of us who have avoided getting caught up in the bubble irrationality.

The best way to deal with software – making it, using it, dealing with it as a creative medium, coping with it as a maker of other creative media – is to understand it as it really is:

A creative medium driven by fashion that operates with limited regard for practicality or productivity.

This isn’t far off from what I was already doing. Systems-thinking is still one of the lenses I apply and I’m firmly of the school of media analysis that takes business factors and processes into account, so it’s mostly a question of broadening the focus slightly, coupled with a small change in perspective.

Maybe it won’t work and I’ll have to rethink things again, which is fine. Maybe it’ll only work for a few weeks until the topic runs out of steam, which is also fine.

But at least it should be a bit fun.

---

I’ve finally released print versions of all my books:

• The Intelligence Illusion: Why generative models are bad for business
• Out of the Software Crisis: Systems-Thinking For Software Projects
• I am Uncluttered (Yellow)

If you’d like to support this newsletter and make it likelier that I continue to publish in print in the future, this would be the time to order a copy. 🙂

Previous entry

24 September 2025