Making the website (and some information about my home server / regarding AI)
GBG Site (formerly GBG Recipes) has been a project born out of procrastination. I initially wanted a reliably accessible documentation of my recipes, especially so that my partner and friends could access them when they decided to bake/cook. I decided on a simple layout to ensure that readers wouldn't be too taken aback by having everything all in your face, all at once. Many recipe websites nowadays, and this may just be a comment on the current state of the web, are needlessly complex in such a way that can reduce readability, and encourage frustration. My Mum is particularly stunned by all of the pop-ups that make websites a wrestle to navigate, and as a result she avoids web-navigation like the plague. Why would you want to make it hard for the average viewer to take in and actually use your recipes?
One of the pet-peeves that I had as I just started to get into cooking at the start of university (curse the chair in the blatantly too-small-for-chairs kitchen I had at home), was that the recipe sites I was using for references had long introductions, weren't practical, and were filled with lots of affiliate links / advertising cookies - and now may even be an AI-generated hallucination. Lots of cooking technique was either taught to me by many Youtube chefs (Adam Ragusea, J. Kenji Lopez-Alt and Baking On A Budget alongside many others) or techniques inspired by foundational knowledge of chemistry and pharmaceutical formulation (which highly complements food science - this gives me flashbacks to when I said "Baking is the chemistry of the home" in my university interviews.) Navigating through an author's life story tended to get in the way of scrolling to the ingredient ratios that I wanted in the first place. I just wanted a little reference website, so I got to making it.
In school I had web-design classes. I loved the concepts taught to us at the time, so the foundations stayed with me. I just searched up any issues I had as I went along. I liked the idea of bare HTML and CSS; I was looking for the least amount of layers when it came to complexity and they managed to provide everything I required for a static website. I decided against making my website with popular open-source tools such as Hugo solely because I wanted the process of problem solving that comes with writing your own code - credit must be given to HTML and CSS for their simplicity, which makes the problem solving easy for a non-coder like myself.
My first iteration was hosted with porkbun.com, which is also the domain registrar that I use. This was okay, but it had issues with updating my CSS. I was going to self-host a server anyway, so I decided to host it using apache, all by my lonesome. Getting a server set up portably did, however, confuse me. This changed when I discovered through my partner a wiki based on a presentation ran by FUTO software on self-hosting servers. This sparked inspiration within me to embark on a journey to host my website, as well as expand it to include a blog, a professional profile (akin to Linkedin) and RSS feeds so that site updates wouldn't need announcement via social media - something I have tried to distance myself from partially due to privacy concerns.
Starting up the server was being punched in the face with networking lingo. I was living in a managed student accommodation with no access to my router when I first tried setting the server up, which was an unsuccessful attempt. I then tried when I went back to my parents' house to set up my own personal router on a raspberry pi, which worked. I set my router up as a static IP on my parents' router, outside of the range of their DHCP (Dynamic Host Configuration Protocol) which automatically assigns a new IP whenever the connection is disrupted / time has passed. I then set up my own subnet on their network which included my server and router (which were physically attached via ethernet.) This meant that I had successfully created a LAN that was connected to the outside world.
Having your LAN connected, unprotected, to everyone and everything is a terrible idea. This can be avoided by use of a VPN. VPNs tunnel your connection, meaning that you can establish a secure link to your LAN from anywhere, remotely. I used Wireguard for this, which was the most lightweight option for my router firmware (OpenWRT). The only issue I had was that this now meant that everyone couldn't access my website unless they had my VPN details, so I had to run my site off of my router instead, which luckily, OpenWRT has integration for.
So I had a website now, publicly accessible, but it had to go back down because I was moving again (you can understand why I wanted a portable setup.) I decided I wanted to give the site a makeover, especially because I was moving back into my old apartment for a bit prior to starting a new job. This gave me and the site some downtime to rehaul the whole thing. I had started with a very pure HTML experience, but decided to modernise by adding a little bit of modern design, especially given I wanted it to have a professional section.
That leads us to the time of writing this, I have just installed Fedora KDE Plasma on every computer I own. I have a phone with GrapheneOS. I have pruned my former internet presence to get rid of old accounts off of applications I never cared enough about. I've changed my email to de-Google myself. I changed my TV media centre from a TV box to a Mini PC, also running Linux. The website that I have made for myself is just another extension of all of these activities, it's an assertion of digital sovereignty. Most of this is spurred by the recent improvements in large language models, and the public acceptance of AI into mainstream life. We have existed with an internet that has been scraping our information for advertising ever since Mark Zuckerberg hit the market. I do think the rise in AI use in arts, culture and science is an extremely blatant overstepping of boundaries, at least my own. With so much Legalese in a set of Terms and Conditions, there will always be a loophole whereby people can use your content to form a product. This site is me opting out of that.
To end my tangent, I think it's fairly difficult to not be disillusioned with AI use. I went through my university course openly shirking the use of AI. We were one of the first academic years that were officially briefed on an AI library tool that digested papers and showed you a summary alongside references. This brings up many questions. What will be the longer term impacts on critical thinking skills? Handing control over your factual information and opinions - that will be used in academic research - to one of a few corporations? A little bit strange. Information bombardment is an increasingly prevalent issue in today's society, and many people will not fact-check the results of a large language model (despite warnings to do so - similar to Terms and Conditions.) People need to develop their skills in discerning real from fake, and reliability from unreliability, in a world rife with so much data. They need to become comfortable with navigating it all, instead of shutting themselves away and asking a server to give them what they need. AI is completely ambivalent to context and bias. Learning how to notice issues and reason with what to do regarding them is an essential skill for an academic, especially one who will practice evidence-based medicine. Medical misinformation is already a tough enough enemy completely disregarding any potential loss of skills from over-reliance on AI.