Here’s how to make your post as hard to read as possible.
Step 1: Make it all one paragraph, the longer the better.
Step 2: Don’t put in any subheadings. If you’ve followed step 1, you won’t be able to do this anyway.
Step 3: Resist the urge to reformat some sentences as bulleted lists. Keep those sentences nice and long.
Step 4: Most importantly, make everything lower case. This includes the starts of sentences, the word “I” and any proper nouns.
This final step ensures that readers will balk at even attempting to read your post. Seeing a formless blob of grey, they will move on to something more appealing.
Good job. You’ve accomplished your goal.
What’s that you say? All lower-case is more conversational and authentic? No, it’s an unoriginal affectation that makes your prose hard to read.
There’s a guy who occasionally publishes reviews on Reddit that are honest, informed and non-monetized. I trust him and have in fact bought one of the things he recommended. I use it often, and I’m glad to have it.
Lately, though, I’ve noticed something different about his reviews. They seem to flow a little better, the sentences are a little less awkward, the rough edges have been sanded off. Has he suddenly become a better writer? I would like to think so, but more likely he’s taken to running his reviews through AI.
What a shame.
Before, he had a voice, a personality that I could imagine with a passion he enjoyed writing about. Now, he’s . . . I don’t know what to call it. Some words on the internet?
If you really want to improve your writing, there are some great books to help you do that. A good start is On Writing Well by William Zinsser, which is itself well written and thus an easy read.
Here’s what he wrote in the book’s introduction in 2001:
Ultimately, however, good writing rests on craft and always will. I don't know what still newer electronic marvels are just around the corner to make writing twice as easy and twice as fast in the next 25 years. But I do know they won't make writing twice as good. That will still require plain old hard work — clear thinking — and the plain old tools of the English language.
Whatever you do, keep your own voice, even if it is unvarnished. It’s part of who you are, and we appreciate it.
I’ve been interested in the concept of the indie web for a few years now, but have struggled to take part in a meaningful way.
I wrote about an early attempt in 2024 when I discovered that you can get a random RSS feed from indieblog.page. To this day, I continue to get 10 random posts every day in my feed reader.
The problem, though, is that those posts are rarely of any interest. They tend to be extremely niche. Typically, they are about a technology I don’t understand or something personal suitable mainly for friends and relatives.
I haven’t given up, though.
Kagi Small Web is another way to randomly find indie web posts. I find that it’s good for filling in a few minutes of boredom, but I’m pretty sure I’ve found only a few articles that I enjoyed reading.
There is a more recent effort called Blogosphere, where you can go through the latest posts of indie sites chosen by the developer. With a fair bit of scrolling, I can find one or two articles worth reading — so it's not bad. You can also sign up and follow your favourites, but why not do that with your feed reader?
The latest I’ve come across is Bubbles, which came out about a month ago. It allows for up-voting and comments, both of which help surface articles worthy of your attention. It’s like getting recommendations from like-minded people. There is also a daily briefing for the top posts, and a hall of fame for the top blogs.
It might seem a little too close for comfort to the corporate world of Reddit, but I’m willing to live with that. If the indie world is going to flourish, it needs a way to make itself known. Bubbles might not be the best way to do this, but it’s the best I’ve found so far.
Apparently, Apple likes those big-ass round corners so much that they have decided to go all-in with what they’re calling “circular” design. Here’s an example I found for TextEdit.
In seemingly miraculous fashion, basic functions that I thought were lost to my Mac have returned. Spotlight is now able to find stock Apple apps, and Mail is now able to send emails.
So what happened?
I can only guess, but I do know what I was doing just before the miracle — fixing an error in Terminal, along with cleaning up brew installations.
The error was a missing quote mark in a file, something that has plagued me for years. Before AI, I would copy and paste errors into Google and hope for an answer. Typically I would wind up going down rabbit holes, becoming increasingly frustrated until I finally gave up.
What saved me was Le Chat from Mistral. It showed me how to find the offending file and how to fix it. Out of curiosity, I asked if the the missing quote mark would cause any problems. To my chagrin, it had the potential to create a long list of problems, many of which seemed embarrassingly significant.
But this was a long-standing error, so how would fixing it solve problems that just cropped recently? Again, I can only guess that it was a cumulative thing over the years that finally reared its ugly head.
While I was at it, I upgraded all the stuff that I’ve installed with brew. In some cases, I decided to uninstall — things that take up a lot of space and that I no longer have use for. These included Node, Ruby and Hugo.
Why were these installed in the first place? Because I was curious about static site generators. Fussing around with Terminal is not my idea of a good time, so I eventually moved on. Good old Textpattern can never be replaced.
Recently I learned that Font Awesome has acquired static site generator called Eleventy. They’re rebranding it as Build Awesome.
This is surprising given that Font Awesome’s icons are meant to make life easier for developers. The same goes for Web Awesome, a collection of code components.
I’ll be curious to see how or if they can make a static site generator easy for the average person who has never hear of Terminal, and would be scared to touch it if they had.
I’ve had my iMac for almost a decade now, and aside from slowing down a bit over the years, it’s worked just fine.
Until now.
It started with Spotlight no longer being able to find any apps or files. After a lot of back and for with Claude that resulted in more time in Terminal than I normally like, along with a trying number of reboots, I was able to force Spotlight to re-index.
Spotlight is now back to normal except for one weird thing — it can’t find the stock Apple apps, including, just to make things more irritating, Terminal. These apps are kept in a separate part of the system, and Spotlight — for reasons unknown — is no longer able to index this part.
Claude suggests re-installing macOS, insisting that it’s not that hard and only takes 30 minutes. I’m skeptical, and besides there are other solutions. For one thing, Terminal is now in the dock.
And Spotlight has been demoted from Command-Space. I’ve turned that combo over to Launchbar, which has no trouble finding stock apps. I’m tempted to go on a rant about why on earth a third-party app would be better at indexing than the one built in. But I'll spare you.
I tried other launchers. Raycast found the stock apps, but slowed down the computer to the point of making it juddery. Alfred, surprisingly, could not find the stock apps. Quicksilver found them, but has an interface that only a mother could love.
Launchbar is free if you don’t mind the occasional nag. I haven’t been nagged yet, so we’ll see how that goes. If it gets bad, I might be tempted to see if I can bear Quicksilver. Or, you know, actually pay for Launchbar.
Update: Now Mail won't send email. I switched to Spark, which has a free version that seems fine. But this really is annoying. My expectation is that it would be the opposite — Apple's apps should be the gold standard for basic reliability. Third-party apps should only be there for optional functionality.
I have been thinking about my characterization of OpenAI and Anthropic as being motivated by greed. This may be unfair, given that greed motivates all of us to some degree, and it is a necessary vice if you are to succeed in the corporate world.
Economists have observed that self-interest and the profit motive are the engines of capitalism. We use legal institutions to harness these basic human instincts for the greater good.
To his credit, OpenAI CEO Sam Altman has on a number of occasions mused about the need for legislation in the area of technology. I won’t argue with that, but there is another factor we should keep in mind: fear.
Why is the U.S. military so intent on getting unfettered access to AI models being developed by OpenAI and Anthropic? It’s because they fear their rivals will gain superiority in AI and use it against them. We’re in a new kind of arms race.
Fear often outweighs greed. Entrepreneurs are aware they need a rules-based playing field to keep greed from getting out of hand. But fear on a global scale can prevent this from happening — fear that rivals are not doing the same, and that our side would handicap itself with self-imposed rules.
Would some kind of international agreement work? To a certain extent, yes, but there would always be the fear that others were cheating. Would it even be possible to verify that they weren’t?
A lot of people have been deleting their ChatGPT accounts in favour Claude. It’s their way of choosing sides in the controversy surrounding contracts with the U.S. government.
I was initially tempted to do the same thing until I concluded that Anthropic, the company that runs Claude, is motivated by greed every bit as much as OpenAI.
My first thought was to try Ecosia’s version of AI. I already use Ecosia for search, and like the idea that they use their revenue to fund environmental causes.
But then it occurred to me that Ecosia is very likely using the services of other companies for its AI responses. Sure enough, it pays OpenAI and Google for their APIs — application programming interfaces that allow them to tap into those companies’ models.
They somewhat mitigate this dependency by framing their answers in an environmentally conscious manner. For example, they they tried very hard to make lab-grown meat seem like a good thing — not creepy at all! But still, I wanted to keep looking.
I’m currently trying out Lumo from Proton. They rely completely on open source large language models (LLMs), with no connections at all to OpenAI, Google, Anthropic and the rest of the proprietary bunch.
But open source LLMs don’t come from nowhere. When asked, Lumo cited these contributors: NeMo, OpenHands, OLMO 2, GPT‑OSS, Qwen, Ernie 4.5 VL, Apertus, Kimi K2.
Of these, two might be a bit shady. OpenHands is a Meta project and Qwen is based in China. Lumi promises complete privacy, though, so there should be no worries about your data making its way to the owner of Facebook or the Chinese government.
I have a feeling that many of the switchers to Claude were people who use it for programming. I don’t do much coding these days, but when I do, I might fall back to Claude.
Lumo was able to show me how to centre a div, and it had some good ideas about how to use an app to update a web page. I’m not sure at what point I would need to go for the subscription-based Lumo+ if I really wanted to up my reliance on it for coding.
With ads now showing up in ChatGPT, there is a worry that the answers will turn into sales jobs for the advertisers. That’s too obvious, and is unlikely to happen — at least in the short term.
What is more likely is that the AI will try to lead us down more rabbit holes, hoping to engage us by being more “helpful” than ever.
Already, pretty much every answer ends with an offer of further help. “Would you like me to ... ?”
I predict that initial responses will be tailored to make you want to follow up with even more questions. More questions means more opportunities to show more ads, which of course generates more revenue.
For example, instead of giving a straight-up answer, it might say there different way of approaching the question, forcing you to ask another one.
If you don’t like it, you can always switch to a paid version.
The Mac’s Finder icon has, for me, always been an artistic depiction of the phrase “friendly interface.” It’s been like that since 1995, but with its latest iteration in Tahoe, the face has lost its inter.
Let me explain.
I see the icon as a play on the word “interface” — Susan Kare’s idea of what a face would look like if it was “inter”. The icon, I believe, is meant to show two faces brought together as one, just as humans and computers are brought together in a real-life interface.
The earliest version showed an obvious split between the two halves of the face, with the lines extending outward on the top and bottom. Gradually, the line disappeared and the only way to differentiate was the colours.
The Tahoe version breaks with that tradition. It looks like a blue face with a partial white mask placed on part of the right side. It retains the friendly smile, but it’s no longer inter.
Given that many people don't like the changes to the interface introduced by Tahoe, an interface that in some ways makes the interface more difficult, this version of the Finder icon stands as a reminder that a friendly interface may no longer be the goal at Apple.
