remy sharp's b:logTop Front-end Bloggers1 min read
AI can operate a screen reader. What does that actually test? [link]
Really like the framing of this article. It uses some very visceral (or easy to understand) examples of how people vs. AI would approach the problem space, and my feeling is that this kind of thinking works well when expanding well beyond the scope of accessibility (though this is an excellent place to start). The floor plan example is good, but it was this that particularly won me over: An AI agent starts with advantages a person doesn’t have. […] The agent knows the goal before it begins. […] A pass tells us the agent found a route, not what finding it was like. AI performs; people experience. Original link: nicolas-steenhout.com/ai-screen-reader-testing
remy sharp's b:logTop Front-end Bloggers1 min read
The Correspondent [book]
A lovely story that held all the marks of a real person full of real faults, discovering late in life that there's still more to be seen. I can't remember the last time I shot through a book so fast. 6 days it took me to read (fast for me). I think it's partly due to the technical nature of the book. Specifically, it's written as (the title would suggest) correspondents from Sybil (the main character) and the people she writes to. Due to this structure, there's really no particular moments when the story has any "what's next" moments, but rather just flows from letter to letter. Although it is written as a series of disjointed letters, it works really well and at no point did I feel lost - even when the letter starts, if it's addressed to Sybil, unless it's an email, you don't really know who it's from until the clues in the letter. Sybil is in her twilight years and has enjoyed a good career and when we meet her she really isn't looking for much other than exchanging letters with her friends, family and others. One thing is very clear from the letters and the book: Sybil is flawed, but in ways that we can see as the reader, but she doesn't see from her perspective. In ways that are frustrating but don't (or didn't to me) come across as contrived for the story. I've known people with these kinds of blind spots. I can imagine myself being just as stubborn. I really enjoyed reading the book, it ends a little cliche but not so much in a sickly way. Although my last two books were centred entirely around 70+ year old characters so I think the next book will have to try to be a little different!
remy sharp's b:logTop Front-end Bloggers4 min read
My AI policy [blog]
There's no shortage of commentary on LLM usage and I've penned several posts that have never left draft status because in part I can't articulate exactly what I want, and partly because there's so much noise around the subject. That's been true for the last few years but as an senior (and old) developer I think it might be useful to at least put in writing what happens to be true for me this moment in October 2026. Fair notice: this post will only briefly cover any policy I personally hold and quickly move into my opinions about LLMs and our community.
remy sharp's b:logTop Front-end Bloggers1 min read
Tell Me Everything (Amgash, #5) [book]
I wasn't a fan of the narrative technique, or, particularly the characters. I'm sure it didn't help that it turned out that this was the last book in a five book series, where I've been told the previous books had built up the characters, and this was more of a bringing together of the characters from the previous books. This book was picked up because of recommendation, but I'd obviously missed the fact it was entry 5 of this world. The narrative tone reminded me a lot of the narration used in the movie 500 Days of Summer. Which, felt fine for a chapter or two, but for me, it felt a bit tired and forced the further I got. I also often lost track of the characters that weren't the main Bob and Lucy. One character is given months (or weeks?) to live, but I couldn't tell who they were or whether they were important (thankfully the book did give me a helping hand as a reminder once the character was dead!). I found some of the choice of dialogue odd and jarring at times, like the random use of "dickwad". Then finally the characters, literally telling each other everything… I just didn't care. The stories the characters Lucy and Oliver exchange are questioned once as to the point. As the reader, the thought had crossed my mind already. I feel like my review is particularly hard going on the book and maybe this story, even with a brief randomish murder thrown in, might have made a good conclusion to the characters, but as someone picking it up "cold", it didn't quite keep my interest.
remy sharp's b:logTop Front-end Bloggersread at source
Firefox version 157 is very round (and how to make it less round) – creolened.com [link]
This is very useful. I do like the direction Firefox design is going, but the very round tabs make me think the hit target for tabs - especially when pinned and muted - is much smaller than it actually is. Original link: creolened.com/firefox-version-157-is-very-round
remy sharp's b:logTop Front-end Bloggers1 min read
There are no "rogue" AI agents [link]
I've read so many posts in the last few months about AI agents going "rogue" and this post does a good job of calling absolute bullshit on the language being used by media outlets (and the tech companies themselves). AI cannot think for itself, nor can it take independent actions. But anthropomorphizing language suggesting it can is how the technology is described. I keep saying this, but the companies are responsible and should be prosecuted as such. Any other smaller company or individual caught in a cyber attack (which is what these reports actually are) would be likely jailed and banned from using computers again. Lest we forget Aaron Swartz. Original link: eoinhiggins.substack.com/[…]/there-are-no-rogue-ai-agents
remy sharp's b:logTop Front-end Bloggersread at source
A Trump team wants to automate U.S. Web Design System accessibility testing. Experts are worried. [link]
This is gut wrenching. This is another example of cost cutting exercises that affect minorities. This isn't a story about AI, it's (another) story about pulling funding from the places that need to support those that need it to increase the profits of those who don't. Most accessibility testing is evaluating how a real human user — particularly a real human user with accessibility needs, like a screen reader user or other assisted technology user — would interact with the site. It is about what a human would experience. Original link: fedscoop.com/trump-team-automate-uswds-accessibility-testing-ai
remy sharp's b:logTop Front-end Bloggers4 min read
Redesign: fonts, alignment and nav [blog]
As I continue occasionally tinker on the redesign I've started (and often think "what's the point, current design is fine"), I wanted to share some of the problems I've run in to. Problems mostly for me because I'm not a designer, and I don't have the wiring to know what works.