Firehose

Sometimes people message me to say thanks to me for sharing my code publicly, because it has helped them understand something.

Such messages reach me sending electric strikes of fear into my heart. I can’t stress enough how bad of an idea referencing any of my code is. I feel I must apologise.

Just discovered this is a thing you can do if you’re so inclined. https://youtube.com/tv works on non-television devices (as long as you spoof your user agent).

I love when platforms shut down due to a flawed business model lacking profitability, so another platform comes along aiming to fill the void left by the old platform but also copies the flawed business model, also fails to profit, and then also shuts down.

One of my favourite genres of business.

‘Did you add an extra page to the site just so the footer links would balance nicely?’

Me? No. Pfft. I— Y’know— Me? Ha. I— I’d never! Not something I’d do… nope…

If you suffer from acne or get the occasional pimple, you may be familiar with pimple patches. The good (non-psuedoscience/gimmick) patches are hydrocolloid patches, which absorb the fluid and form a gel from it.

You can buy rolls of adhesive hydrocolloid dressings for prices that are negligible compared to a comparable volume of pimple patches at their premium prices. Patches cut from hydrocolloid rolls aren’t usually quite as invisible as purpose-designed patches, but they are transparent. The raise around the edge of the patch due to its non-tapered thickness is the most obvious part, but you could feasibly trim them yourself to minimise the visibility.

You can cut these rolls to suit, only using as much as you need, and they’ve been proven both scientifically and anecdotally to work on popped and unpopped pimples while deterring scarring due to the moist healing process. They do not work on deep, blind cystic acne or blackheads, however. In addition to hydrocolloid’s properties, a patch also disincentivises idle picking and protects the site from muck and bacteria. Even if you buy a hydrocolloid roll and find it useless for treating pimples, it is still a valuable tool more generally for burns, cuts, and scrapes.

To use, you should first gently wash and then dry the application site. Next, cut the patch to size, peel off the backing, and apply it where needed. I’ve had great success with them.

The only agent I’m trusting is Agent 86, Maxwell Smart. He is a long serving member of CONTROL, a secret US government counterintelligence agency.

Was helping out a (seemingly) nice client in her 80s who was locked out of her Facebook, and based on the vibe I got from her (and her very vocal support of Donald Trump and Pauline Hanson), I’m almost certain her account was suspended due to spewing hate speech and harassing people. 😬

There’s simply too much good human made stuff out there to justify wasting time consuming AI made slop.

Shaving all the hair off your head can be great but is an ordeal. I don’t own clippers or an electric razor, so this is how I do the task. It takes me roughly three hours from a full head of long hair to bald, including prep and clean-up.

I’m usually beginning from over a year of unmanaged hair growth, so an important first step is to remove the bulk existing hair. I do this as part of a multi-part process. First, I clean my hair and wait for it to dry, then separate it out into strands and perform many plaits, holding it together with rubber bands. Then, I use some scissors to cut it as close to the scalp as possible. This process keeps the hair orderly, so it can be used for alternative purposes in the future, such as wigs or paintbrushes.

I then follow up very carefully with scissors, getting as close to my scalp as possible and removing as much hair as I can. Once I’ve gotten it as close as I feasibly can with scissors, I move on to using a razor. I always ensure I have enough razors on hand, because going shopping for more with uneven clumps of hair stuck to your head isn’t a great look. I shave my head with the same process I use when dealing with any other bodily hair I shave, though it is worth considering that hair on the top of one’s head is usually much finer than that found elsewhere. For a first shave, going with the grain to minimise irritation is certainly the best choice, even if it doesn’t let you get quite as close of a shave.

The back and top of the head are a particular pain to shave, so I use a second mirror (often just my phone with the front camera). Holding a razor in one hand and mirror in the other, I look at the reflection of my mirror in the main bathroom mirror to see the back of my head. I take particular note of the hair atop and around my ears and at the back of the head, especially between the upper portions of the trapezius muscles, which forms a nook that can be hard to access. The crown can also be difficult, as it is hard to track the grain of the hair.

I pay particular attention to these parts and avoid going over the same place too many times in succession to avoid irritation. The head is a big place, so don’t focus on the same small part for too long – give it a break for a little while and address another part, coming back to the previous bit later.

After shaving, I have a shower and then touch up any remaining bits before another quick wash. Then I apply whatever creams/lotions I have at hand to sooth my scalp. Shaving one’s head isn’t too tricky but is an effort. You must of course ensure your hair does not end up down the drains, causing a clog.

There is a point during the process of my annual haircut down to bald where I survey the sink with a certain trepidation.

I’m gonna have to update my profile picture, aren’t I?

Google Search Central made a post on LinkedIn, and it has the most genuine human-written comments I’ve ever seen on the platform. It is just a bunch of SEO experts saying variants of: ‘We’re fucked’, ‘Our careers are ruined’, ‘All our traffic is gone’, ‘Google wants us dead’.

:(

Also, oh, okay. Thanks Google. I was wondering where all the traffic went.

2026’s Google I/O (Google’s annual developer conference) has been a disaster for the web. The conference-driven development’s forcing through of the Prompt API, a set of Modern Web Guidance skills for AI systems to use that are already showing major accessibility shortcomings, and a whole ton more AI-spangled sloppery, is rushed and unwelcome.

I think the most damaging announcement is the changes coming to Google Search. Rather than a list of relevant links, a search on Google will be more aggressively prioritising the LLM-generated summary, now complete with vibecoded tables, graphs, and interactive elements.

There has until now been a social contract. Website owners let Google scrape their sites and present them in Google Search, and, in exchange, Google Search sends traffic back to those sites. Google wins via adverts on the search page, and sites win due to however they monetise traffic. More largely, everyone wins because there is a financial incentive to create and produce new content.

However, Google killing their side of the contract ends this. If Google only takes and never gives, then sites cannot profit. What is the incentive to publish if the only outcome is feeding Google’s AI with no return? What sources will LLMs have to pull from if all the sources are defunct? How far will Google go folding adverts into their AI output?

I can see the huge short-term gain for Google, but I see no long-term path – not even an unsustainable one. This feels like the end, but of exactly what I’m uncertain.

All of my peers (bar the ones that work at Google) are shattered in a way I’ve never seen before. I don’t know where we go from here.

I need write some words coherently proper please. Grant me strength to word put together into sentence so can read well and goodly. Brain all frazzled can’t word readable way thank you. Oh dear. Too many language.

EmDash caught my eye when Cloudflare released it at the start of April. April 1st, no less. It is a simple content management system built upon the lovely Astro web framework. EmDash looked like a solution to some problems I encounter, those being that WordPress is too much to manage for simpler sites and is too complex for less computer-adept folks to manage. I love static sites, but a CMS is a must for people without much tech experience. A bespoke static site is also a much greater undertaking than an off-the-shelf offering.

WordPress and I get on as well as any developer can get on with WordPress. We might not be the best of friends, but we are amicable. However, I sometimes yearn for something simpler I can plonk in front of an older or less technically inclined client. Something that is easy to host, easy for people to update content on, and looks decent out of the box is the dream. Sure enough, someone who could barely manage their Mac managed to figure out EmDash. Hooray!

Unfortunately, EmDash isn’t ready for prime time. It is primarily AI authored, and that shows. Some UI elements overflow off the screen. Many features lack documentation or have documentation that is ambiguous in a way that only an LLM can manage. ‘Failed to publish. Failed to publish content.’ is an error as frustrating as it is vague. It also isn’t triggered by some unknown edge case, as I can replicate it with the CMS’ online playground.

There is no polish here, and it is an exercise in frustration. I can’t in good faith put EmDash in front of a client. EmDash will probably improve (it is only a month and a half old), but it isn’t fit for use yet. Far from it.

Playing around with training a small language model exclusively on my own writing and work again.

I love this output: https://www.com/wiki/posts

It is trained on links and the presence of URL parts like ‘https’, ‘www’, ‘com’, and ‘wiki’ (from Wikipedia) in association but doesn’t quite manage to put them together correctly. This mode of failure really represents how language models work under the hood.

They say you mustn’t Log Off.

A far from perfect little table, but one that hopefully gets the point across and works as a reference. Scripts ≠ languages, with many scripts being used by many languages.

Trying to kill WordPress is like trying to kill a god. You will be made an example of, humiliated, and quite possibly smote by powers otherwise unknown in this realm.

I love dark mode, but it mustn’t be the only option. For a lot of people, white text upon a dark background is illegible.

Common eye conditions like astigmatism, myopia, and presbyopia can make content look foggy, become hard to focus on, and cause eye strain.

The above effect is especially exacerbated by strong contrast, like pure white on pure black. However, different conditions benefit from different levels of contrast. For example, people with conditions such as cataracts, aniridia, and achromatopsia can benefit from low-contrast visuals. The prefers-contrast CSS media feature is fantastic for tailoring to your user’s needs.

I’ve successfully broken Hacker News. This submission has -4 points, somehow.

https://news.ycombinator.com/item?id=48104663

I got jumpscared by John Lennon’s face appearing on my computer screen, but it was actually my own reflection. It was late ’60s John Lennon I thought I saw as well, so evidently my hair is looking dishevelled.

You can’t put important links exclusively in your site’s footer if you have infinite scroll.

You can’t have infinite scroll if you put important links exclusively in your site’s footer.

Please stop.

A lot of developers mistakenly think that the code is the product. In reality, for most businesses the product is the product, and the code is merely means to an end.

I think this explains why so many developers’ efforts are so horribly misguided by modern technologies. They’re building the code, not the product.

I’m trying to reduce my reliance on the increasingly unstable United States of America and move to more ethically run services. As part of that, I’m moving from GitHub. I’ve been steadily moving off GitHub for months. I’ve moved lots of site hosting off of GitHub Pages, removed GitHub-based comments from this site, and have been rearchitecting systems to avoid being GitHub-specific.

Unfortunately moving to another Git forge is a big effort – in part due to choice. Though I am comfortable and capable of self-hosting, I don’t want to have to self-host my Git forge. That isn’t a maintenance burden I want, and unless federated, it adds unwanted friction for contributors.

1. GitLab: I’ve been on GitLab for ages. It is pretty fully featured but has an identical vibe to GitHub. If they had the chance, I think they’d make the exact same mistakes.

2. SourceHut: So complicated yet so lacking of comfort features. The interface is too finicky and email workflows are clunky.

3. Gitea: A tad shady for my liking.

4. Forgejo: Fantastic, but I don’t want to self-host.

5. Codeberg: Pretty good hosted version of Forgejo, and I’d probably be happy with it.

I’ve ended up with Tangled. Tangled is built on the AT Protocol, which is nice given my adoption of the ecosystem. However, it also introduces my biggest gripe, which is the lack of private repositories, meaning I won’t be able to leave GitHub entirely. I very much do like that you can self-host Git repositories with ‘knots’ and CI runners with ‘spindles’, allowing better data portability in future. Simple repo migration is also on the roadmap.

Tangled also unfortunately isn’t supported by a lot of build tools and systems I use, which is a shame from the perspective of ease-of-use. I won’t be able to make a full migration until Tangled adds support for releases and private repositories and until other services get support for Tangled, but I’ve already moved over what I can. Most notably, I’ve moved my Stoat bots.