Kevlin Henney

consultant • father • he/him • human (very) • husband • programmer • keynote speaker • technologist • trainer • writer

Last Notes

2025-12-09 22:05:03 UTC

A reminder: when displaying dates and times in apps, do just that. Display dates and times. Don't withhold information have from the user. Don't try to be clever.
Just saw an app tell me that a particular message was sent to me on Wednesday. Which Wednesday? In the time zone I am currently in, it's Wednesday. If the message was sent on Wednesday 3rd December, that is the information you give the user.
Even better, include the time. That way, people can more clearly see intervals and durations.

2025-09-30 06:05:33 UTC

Blogged: Step Back
"Instead of trying harder, do the opposite. Take a break. Properly. Don't just switch to another screen to browse memes, graze cat pics, answer your email or catch up with notifications, messages and socials."
https://kevlinhenney.medium.com/step-back-39d9e005ed89
https://files.mastodon.social/media_attachments/files/115/291/738/360/989/366/original/b30787a01f06b844.png

2025-08-25 09:17:33 UTC

One of the interesting things about AI-generated code concerns reliability and correctness. I haven't come across definitive studies on this, but just rummaging around it seems that where typical rates across the industry for developers are around 5–20 defects / KLOC, with AI-generated code is around an order of magnitude higher.

2025-08-18 16:15:52 UTC

If you're interested in having me speak, run some training, consult, etc., please get in touch. There is a small chance I may be able to fit you in this year, but things are pretty booked up. Better to get in touch now for something early to middle of next year than leave it until the last minute.

2025-08-11 08:46:04 UTC

There are lots of principles that vie for developers' attention, some of which are arbitrary (e.g., SOLID), some of which are misleading (e.g., SRP), or misunderstood (e.g., DRY, Separation of Concerns), some of which are redundant (e.g., ISP) and some of which are just plain wrong (e.g., OCP), but there are some that are classic, timeless, effective and, unfortunately, still ignored.
In this category we find the balanced pairing of high cohesion and low coupling.

2025-08-07 06:26:04 UTC

It's not even funny how so many are fawning over 'modular monoliths' as if they're some great new idea and alternative to microservices.
Consequently, I'm proposing a new conference talk for later this year: Modular Monoliths and Other Facepalms.

2025-07-31 07:58:55 UTC

One of my constant frustrations with many apps (web and mobile) is the constant need to confirm via OTPs, 2FA, etc., that I am indeed me when I'm in the same session for which I've already just done this less than a minute earlier.
Security — as opposed to security theatre — requires a more joined-up approach to architecture. This kind of annoyance highlights potential problems of organisational structure and process in the development organisation, unmanaged technical debt, etc.

2025-07-22 10:21:59 UTC

So why do we prioritize things like feature development and bug fixes in software development? No, really, why? It might not be why you think.
My most recent YouTube Short for the Modern Software Engineering channel:
https://www.youtube.com/watch?v=TVAGd-Si4OI

2025-07-10 10:57:41 UTC

The limitations of Python's lambdas grate as much as their syntactic inconsistency. I can live with the latter, but the former is a continual source of frustration.
Does anyone know of any serious proposals to improve lambdas, such as allow them to hold statements?

2025-07-09 09:39:42 UTC

Tell me you don't know what reading is without telling me you don't know what reading is
https://files.mastodon.social/media_attachments/files/114/822/608/167/265/049/original/ad01b5e27d5e50ef.png

2025-07-03 09:22:33 UTC

To understand how effectively people currently use and will use AI one can look at two previous transformative technologies: the spreadsheet and the search engine.
The first point is, of course, that they have been transformative. There was a definite 'before' and 'after' for individuals and businesses.
They have been effective. That does not, however, mean their widespread adoption has meant they have been used effectively, and the way these products have evolved has also been suboptimal.

2025-06-10 09:23:59 UTC

Watching a speaker apologise for any quirks in the code they're using in their slides.
"It was generated by AI."
Resisted the temptation to raise my hand and say, "You know you're allowed to edit generated code before committing it?"
Don't use LLMs to increase the unmanaged technical debt in the world. Software is supposed to be soft; It's not written in stone.

2025-04-17 05:28:15 UTC

- reply

@nprofile…49rv I 'love' the fact that's beyond column 80 😳

2025-04-11 15:03:31 UTC

It's 2025 and we live in a highly connected and fallible world. No, "Something went wrong" is not an acceptable error message.

2025-03-06 10:41:04 UTC

One of the most interesting impacts that AI integration has had on many apps is in terms of UX: usability has declined as apps compete in a downward spiral for our attention, adding buttons, prompts and other enshittification that distracts from and gets in the way of user task focus while adding no meaningful value.

2025-02-17 16:53:20 UTC

Suggestion for a AI-based coding assistant: a comment dissonance checker.
Red squiggles under comments that contradict what the code actually does. Blue squiggles for comments that add no value to what is already obvious from the code.
Also, blue squiggles on code that could be improved to make a comment superfluous, e.g., rename a variable to be more explicit rather than having a comment explaining what the variable is for.

2025-02-11 16:34:00 UTC

Today's passive-aggressive pro tip: use the trim method to trim leading and trailing whitespace from a String in JavaScript
https://files.mastodon.social/media_attachments/files/113/986/212/684/433/711/original/0fc0222c4d72e96d.png

2022-12-06 08:57:59 UTC

As the term 'intelligence' is used in 'artificial intelligence', it is perhaps worth asking, what kind of intelligence do AI systems represent? Confident and detail focused, but naïve, easily misled and lacking in common sense, worldliness, situational awareness, fairness, imagination, empathy, emotional intelligence and social intelligence 🤔