Recently I was looking for some placeholder images for an app I was building.
\nIn the past I've used unsplash, but its overly complex for just:
\n\n\nget me a pic, specific or random, this size, grayscale, blured etc.
\n
Enter picsum
\nSmall, perfectly formed quality of life improvement.
\nIts the little things.
\n","tags":["utils","picsum"],"date_published":"2026-06-16T00:00:00.000Z"},{"id":"http://localhost/posts/genesis/","url":"http://localhost/posts/genesis/","title":"I have a cunning plan...","language":"en","content_html":"The best laid schemes o' Mice an' Men Gang aft agley
\nI've been toying with an idea for a project for quite a long time.
\nI think it could be great, and if successful, could potentially be a small business.
\n"Ideas are easy" they say and they're right.
\nIt takes commitment and time away from other pursuits.
I've certainly become more discerning as I've aged, but it might be time to stop the procrastinating and get building.
\nIt will be fun. It will be hell.
\nI don't even know the things I don't know I'll need to know.
\nBut one step a time.
\nI'll see how it goes.
I ask it questions and it tell me things.
\nSometimes it is confidently incorrect.
\nI have to know the correct answer first of course, therein lies the rub
\nIts a magic box, spitting out wonders.
\nI'm cursing, same mistake on a loop.
\nIt's the prompts.
\nOr maybe the skills.
\nEasy as MCP.
\nNot enough context.
\nWhy can't it remember things, we already covered that.
\nIt stalled again... <continue>
\nIt's just fancy predictive text.
\nIt's a pattern finding machine.
\nIt's NOT AI, its a large language model
\nIt's too expensive!
\nIt's so cheap!
\nIt's so productive!
\nIt's killing productivity!
\nIt's great for learning.
\nIt's teaching me junk.
\nI only verify the stuff I suspect.
\nHow much slips under the radar? Lots
\nCreeping into daily life; an opiate of convenience.
\nIt's the future.
\nIt's a dead end.
\nHype
\nSlop
\nBots
\nCost
\nDatacentres
\nPower
\nWater
\nJobs
\nWar
\nPoison
\nAll these things and more.
\nAI is wonderful
\nAI is terrible.
\n","date_published":"2026-01-14T00:00:00.000Z"},{"id":"http://localhost/posts/reboot/","url":"http://localhost/posts/reboot/","title":"sudo reboot _","language":"en","content_html":"Its been more than a few years since I've felt like writing anything. I haven't been doing much coding either.
\nI've ignored the dev-world chatter I used to hoover up on a daily basis.
\nI shuttered the old blog I rarely ever touched, I stopped posting and reading dev social media. My github activity flatlined.
\nWhen you've been doing app development long enough, multiple decades at this point, not much feels truely new or different in that space, just continual refactoring of the greatest hits.
\nThe new and shiny often seems a lot like the old with slightly different trade-offs and syntax.
\nBeen there, done that, got the stickers, the t-shirts, the slogans.
\nMost of the design trends that emerge to excite the masses, end up distorted beyond recognition, often applied wildly-out-of-context by enthusiastic versions of our younger selves, ignoring the rising frictions from pairing the wrong solutions to a problem context and scale; just as I ignored the advice of my older selves at the time I was doing the same.
\n\n\n"When you've a hammer, everything looks like a nail"
\n
I've had several periods of burnout. Years of working under high pressure and tight deadlines, often with additional hours cramming in research and study to keep up with the churn. Not just the best-practices of the stack du jour, but also all the alternatives of the platforms, languages, frameworks and tooling used in the professional environments.
\nIt was a lot of fun at times but over the years, exhausting. I've read countless posts from other software devs covering burnout, the bellcurve of people lasting more than 10 years in the industry. I'm pretty far to the right of that curve at this point.
\nThe wholesale change in tooling, frameworks, platforms and languages roughly every few years locking in redundancy or requiring painful refactoring efforts, sometimes complete rewrites.
\nCloud-native, your infrastructure, but under the control of external unpredictable roadmaps and billing.
\nWhether it is things like Python 2 v 3, the framework shifting-sands of the Microsoft world, angular 1 v 2, the data contract abstractions of odata, soap or graphQL. Restful or grpc, HAL or hateoas. Monoliths or microservices. Mono repo's or modules. Imperative or functional, Stateful or stateless. Static vs dynamically typed. Branches or Forks. The npm left-pads.
\nnosql vs sql. ACID vs BASE, The php hammer.
\nThis list just never ends...
If there is one thing most would agree with, is that the choices made were probably the right ones at the time but everything decays over time, cpus, languages, frameworks, platforms, standards, processes, apis, data, infrastructure.
\nThere's no global concensus or standard on anything when it comes to software development that really lasts, it continuously reinvents itself.
\nAs I reflect on my career and journey in software development, I've gotten nearly everything wrong at times, Picked the wrong abstractions. The total re-write, the big refactor. Gigantic PR's. Tiny PR's. Code smells and anti-patterns. I've been too naive and too clever. Too conservative and too ambitious. Too many tests, too complex. Not enough tests. Chosen the thing only for it to be doomed or abandoned down the road.
\nI've continuously oscillated between the peaks and valleys of imposter syndrome.
\nI want to be "good" but the development universe is so vast. There are entire galaxies out there of which I still know little or nothing at all. I've mastered a solar systems worth.
\nAt times I've stagnated. At others made giant leaps forward. I've also stayed in the comfort zone of the familiar to "be productive and get things done".
\nSome of the best projects have been that way, just manifesting code for hours 'in the zone' with few errors or corrections. What felt like entire cathedrals built in days.
I've gotten a few things really right.
\nMost of the applications and services I've put into production over the past 20 years are still out there, doing their jobs and running mostly error-free for employers past and present.
\nI checked my neglected overflowing inbox on github recently. One notification stood out to me, one of those times I'd made really good tradeoffs.
\nA user had commented on a years old issue the following:
\n\n\nFor what it's worth, some years later, the above still seems to be the best approach - you have to manually figure out which project files to run against dotnet test, cover each of those, then merge and run reporting separately (which is actually quite useful anyway, because a solution can contain tests that aren't actually unit tests, or only run on special hardware).
\n
This was related to a library I wrote around 2016/2017, Cake.Incubator, to extend Cake (make for C#) in a few ways that I needed.
\nIt still gets around ~2k downloads a day, over 5.5M since first released on nuget and hasn't really generated many bugs nor required much refactoring over the years, despite its widespread use in hundreds of projects large and small.
\nOver 7 years later and its still useful, still widely used, almost error free and as close to future-proof as possible from the Microsoft obsoleance factory.
\nThe cake foundation looks after it these days but its really only had a few dependency bumps over the years or a couple of tweaks to fix up deprecated Cake Core APIs
\nIt was planned, deliberate code and based on years of knowing how Microsoft tended to extend and change things as the versions rolled on.
\nThere are always tradeoffs creating things, and as I've aged, I've gotten better at choosing when to pick the quick hacky thing over the clever one. How much testing is needed and where. I learned from, and developed muscle memory to avoid some of the self-laid traps of the past.
\nI've left most of the cults behind to join the last one I ever needed or cared about...pragmatism.
\nSo what now?
\nI'm not sure.
\nI still enjoy how productive I can be with C# and Typescript as I know them so well, but for the time being, I've left the Microsoft and Windows ecosystem behind where I've spent the bulk of my career.
\nMy OS's are linux. My homelab is linux.
\nI've run diff distro's which keeps things interesting as I continue my journey through the tech galaxy.
\nI'm un-enshitifying, homelabbing an ever-growing-list of self-hosted services, leaving behind the oft-hacked obnoxious ad-pushing, data-mining and telemetry-riddled freebie services that became so entwined in my digital life.
\nI'm unplugged from the noise.
\nI've been having fun again.
\nTinkering with home-automation. Lately making my first tentative steps into the world of microcontrollers.
\nNow AI is everywhere and its only just begun...
\n> sudo reboot\nMy journey continues.
\n","date_published":"2024-10-23T00:00:00.000Z"}]}