This reminds me of trying to use File Explorer in Windows 11. I wish I could turn all their electron-app "improvements" off, to make it useful again, like it once was.. Case in point: Explorer now has tabs. I don't need tabs, I need a single tab, and a window title bar so I can drag the damn thing around. And.. my single tab, now tries to show the folder name, truncated to a few useless characters, so I now have tabs called "C:\folder\sub1\...", while the rest of the row is EMPTY SPACE (which I, admittedly can still use to drag the window around; thank you for that, but it will probably be filled with ADS come next month.)

"Oh, but you can just see the folder name in the address bar in the next row instead then!"

NO I CAN'T. Because they electron-css-screwed that up too.. It now shows a bunch of toolbar buttons <- -> ^ , then a computer screen??, then >, then [...] Then they truncate the file path to only show parts of it, starting the rest with ... Is it because we are out of space? I don't know, every part of the folder path has been separated with [ > ] (because / or \ was obviously the worst idea ever.) Then, to the right of it all, we get a big [Search log ] edit field, followed by a spyglass. So, I get two broken displays of the actual folder path, and a lot of 'candy' I did not ask for. Why does the search tool need so much space, before I am using it at all? What does it need, apart from maybe the single spyglass icon? Instead, the actual path that my object by necessity ALWAYS will have, has been chopped up to unrecognisability.

It reeks of KPI and bonus performance reviews, "we must improve the round shape of the wheel, to get our bonus and not be downsized".

In a similar vein, I've argued in many a corporate meeting that there's no such thing as "empowerment".

People start out wanting to achieve things, change things to be better, do a good job.

The active issue is disempowerment, created by other people (usually but not always senior) within the organisation.

So the question isn't "how to empower people", but rather "how to prevent disempowerment of people".

This isn't always popular, as it shifts the focus and responsibility for different behaviour away from the disempowered rank and file, towards the dysfunctional leadership.

I think codebases and optimizations are a lot like this.

A lot of people seem to think the way to make things work better and faster is to add elaborate caching layers and layers and retries and GPUs and multi threading and...

I find the opposite tends to be true. Make things fast and reliable by doing as little as possible. If an API is flakey, make it not flakey, don't cache the result and add a retry loop.

> Trust isn’t something a brand builds with an ad campaign. It’s what’s left if the marketers don’t ruin it.

So much this. Are ads still a measurably good investment for businesses? I'm assuming they wouldn't run them anymore if not, but they feel so out of touch these days that it's really hard to imagine them really working on anybody.

Sorry for the side-tangent, just felt like that last bit of the post really drove home the point best - at least for me.

> Customer delight isn’t something we add to our projects. It’s what’s left if we don’t ruin it.

my anecdotal experience in this is that getting back X (customer delight / curiosity etc) once you’ve ruined it will usually take longer / be more costly than having just not ruined it in the first place.

also, at some point you will ruin it. at that point it’s a question of by how much and if you choose to un-ruin it.

sometimes doing nothing is a more useful skill than doing something.

Don't let the brevity of this post dissuade you from its value. I believe Seth is getting at a very good psychological insight.

By default, people give a lot of trust and benefit of the doubt. Everyone's account in life starts out a little positive when it comes to trust, welcome, empathy, and believe.

But the flip side of that is that people have a very good memory for past transgressions. When someone has extended you a little trust, or given you some time to learn your product, they will absolutely remember if you turn around and harm them.

It takes only a match to burn a bridge, but a year to rebuild it.

As someone who works in marketing, this is extremely true. Right now, LLMs are causing a lot of one-time cashing in of trust.

I've seen this pattern a bunch:

1. Person builds trust on X/LinkedIn or via an insightful blog/newsletter (substitute your channel of choice here) for a few years because they have unique opinions, interesting stories from personal experience, are entertaining/charismatic, or share data/insights nobody else has.

2. They realize "AI can do this now" and use AI trained on past content to generate the content.

3. They post the content

4. People initially keep engaging because their AI-generated content inherits some of the trust they built up

5. People realize their posts are AI slop and feel tricked or simply no longer enjoy the posts.

6. Engagement falls off a cliff because the assumption has changed from "If I see this person/company in a feed, it's got a good chance to be interesting" to "If I see this person/company in a feed, it's guaranteed to be AI slop.

There's a temporary "Have your cake and eat it too" phase where you get the results without doing the work. But once that ends, you have to build the brand all over again because it's been tarnished.

(Fyi my take isn't that everything needs to be hand-written and no AI can ever be used in writing. Just that this cycle keeps repeating because people don't do the work anymore. You can use AI and still be doing the work of generating genuinely good writing)

Wait, the example held up for "Stop ruining it" is a company that sells snake oil audiophile bullshit?

Sounds like subsidiarity to me

It is in the nature of capital to ruin it - if users feel great about a product it implies that there is more to wring out of them. The ideal product leaves the user with nothing but the utility the product provides with no extra pleasure. If your employee loves to work for you, you're paying them too much. They can't hate to work for you (unless they have no other choice) but if they feel really good about it, that is a sign of a problem.

The irony that the packages system looks so obviously and clearly 'baseline claude' designed is a sign of the moment itself.

I've read haikus that made more sense than this streak of random words.

Feels like an article generated using GPT-1.