2024. April 24.
A new website has joined to the Wloff Empire, a Linux CLI Guide. It's for those who start out, or want to know a bit more about the tools avavilable, or perhaps need to use CLI only occasionally and forget how things are done due to lack of practice. It is not comprehensive, but already has some nice tips, and the plan is it will be expanded time to time.
Further plan is still changing the code of Articles so it is more social media friendly, making individual blog entry pages more sharable. But even before that, this blog is needed to be added to the main Wloff page in some form.
I do not want to list it as part of the works, since it's kinda official blog. I'm thinking in adding a column of icons on one side of the page, and gradually add some more stuff to it, like theme switch (dark and light themes), or publications. But first only to the blog.
Problem is: have to draw icons. Drawing itself is not hard but have to find an style that fits the page's current appearence. This can take some try and will test my patience. I already experimenting with a stand-ins on the development server, at least the positioning and related css. Perhaps I'll settle with something temporary to cut the issue short, then revisit it later and polish if shiny.
2024. April 12.
Lots of people agree that the next BIG thing is Artificial Intelligence. We see huge potential, but also worry about consequences. I came across a great article among many similar on that site and elsewhere that describes vividly the likely scenario that things might won't work out how conventionally we think.
In short we wish the AI to free us from work, and fear it will take our jobs. Employers, CEOs, board of directors, and shareholders expect it to reduce cost, to replace unreliable workers with a tireless machine that needs no pay.
The reality is that these machine learning algorithms, large language models, generative AIs make mistakes all the time, they hallucinate, fill the gaps in. They do this in a very plausible fashion, and it's easy to make the mistake to believe in just about anything they generate. They cannot go unsupervised, they can't relieve people from their work, always have to keep "a human in the loop" as I read.
And here comes the problem. If implemented properly we get a worker with enhanced capabilities doing quality work. Great right? It will cost more however, and noone is willing to pay. So they make it the AI doing the brainwork, and the unskilled work can be done by unskilled workers for peanuts. "You don't need that much money just for checking errors!" The only ones who get their capabilities enhanced will be shitty bosses.
Couple of terms to make note of, gathered from these (and related) articles:
Centaur: when an "AI", an LLM augments an individual in his work, make it more accurate, quality, effective - but not cost effective; frees up the individual from the menial, repetitive tasks, doing the muscle work. The ideal, utopistic setup.
Reverse-Centaur: when the AI becomes the boss, and the tedious, repetitive, and hard muscle work is left to the human. The AI dictates the pace, and the human has to follow. The dystopia of automatized slavery.
Habsburg AI: when a generative AI's input is other generative AI's output, resulting in grotesque inbred mutant, generating content resembling to the classic Habsburg chin/jaw.
Potemkin AI: when the facade of a promised automated service powered by "AI" hides a mass of underpaid humans doing the work instead.
2024. March 31.
The first lengthy post is up, gained some insights. Should put some thoughts to paper since this is a dev blog as well, isn't it. It is.
The aforementioned post contains images and a video. The media loads well but needed to edit the post a bit more heavily to make them look good, especially with captions. Okay, kept the video's size low, converted to webm to make sure so. In case of images perhaps should implement some thumbnailing practice - basically wrapping the smaller image (thumbnail) with a link pointing to the normal sized image. I don't think such facility is needed for the blog engine itself, everyone can include images in an html however they want. Also thumbnailing might not needed for all the images or all the posts.
Now that we are at thumbnailing... shared the post on various platforms, and it could be worth it to make the individual posts "shareable friendly". For me it's not a priority, but for others, and for general convenience it could be a good idea. Perhaps chance to help with search engines too. The meta tags are needed to be edited in the heading - not in my head, but the <head> element -, present them dynamically. However this can only done programmatically with variables, can't do with pure html which makes me sad. I wished to make this thing as simple as possible with the use of as much pure html as I can.
So the plan is to break up the head.html in two parts where the necessary meta tags go, and insert those with php - in case of pages of individual blog entries. A title is needed whith is the title of the article in the 'h3' element, and a preview text, the first 'p' element should do the trick.
What meta attribute to use? The title-description might just work, but those Open Graph tags are also an option, namely property="og:title" and property="og:description" attributes, maybe rounded out with an image, depending. WordPress surely uses them, and seem more convenient, although not convenience was in mind when we begin this project.
To make the metadata change dynamically, the GET request is needed to get handled earlier, before anything renders to the page. Probably it's the better practice anyway. Why show anything from a webpage if there is an error.
So need to change some stuff quite a bit. That's okay. Rome wasn't built within a day.
Carlos II of Spain, House Habsburg - Public Domain