The Lab

I have been working for a few days on optimizing my CMS/SSG Genesis. I currently have 9 pages. 7 are at 100% on mobile and desktop. The others sit between 96 and 99% for performance, with the rest at 100, and 100% on desktop.

I am realizing that I could spend days, weeks, months optimizing and still never get a page with any kind of style to 100%. The smallest carousel: -2. One image a hair larger than its display size: -2. After a fair amount of research, I realized the problem is their reference frame. It does not match anything real. A desktop size that feels like four fifths of a 1990s screen. A phone resolution that matches nothing, and worse, one that breaks on a real phone if you do what they expect. The only way is to use no video, no large images, no carousel. Even less a carousel with videos. I found that if I wanted the whole site to hit 100%, I would have to do what everyone who reached that holy grail did: make ugly sites.

So no. That is clearly out of the question. If I have to spend days building algorithms to generate images for every use case, I will. If I have to build my own lightweight carousel with parallax, I will. I will get 100% on every device and every page of my sites. I will prove that even when obstacles are put in my way, I always find a path.

My goal is not so much to reach 100/100/100/100 in Lighthouse on every page. It is to improve Genesis (CMS SSG) so it can generate highly performant graphic sites by default. It was already decent at the start. I had the same kind of performance as a WordPress or a Wix. But in 5 days I reached 100% in performance, accessibility, best practices, and SEO on 5 pages out of 7. I only have to improve the hero and slideshow creation tool I built first, before I even started Genesis, so that everything is perfectly responsive, optimized, and clean. My goal is to have this level of precision on every site that comes out of Genesis.

What was really interesting was the hunt for weak points. Looking through every part of Homeostasis for friction, for the small things that were blocking the loading of a given image or script. I ended up spending almost more time on these optimizations than on each feature of the App. It also let me fix small broken or inconsistent things. Sometimes the funny part is getting stuck for half a day on something tiny, for example the "alt_text", then seeing that once it is fixed, it improves all the insights at once, sometimes by several dozen points.

What is great is that from now on, as soon as I create a new block, a new type of section, or a new template, it will inherit all these technical improvements natively. In fact, if I could have a PageSpeed Insights for every kind of app and SaaS, it would let me build tools that run perfectly on their own, stay very light, and have no back doors to attack.

First article. First words on this site, on this blog. Today marks 20 days since I started Genesis, and I can finally create articles. So I am taking the opportunity to begin.
First, welcome to my laboratory. I hope you will find things to read here. I am not especially used to documenting my work, even less to commenting on it, but I think this site gives me a reason to start. What interests me is not talking about the site itself, but about Genesis, the software I am building to create it.

I needed to build a car, so I built the factory.

In fact, I thought that, as with my other projects, there was a good chance that each line of code could be useful somewhere else. I like the Lego principle: create bricks > assemble them to make an infinite number of projects. I hate wasting time. If I build a site and spend 20 days on it... I only have a site. I have just spent 20 days on Genesis, and I can create as many sites as I want. Of course, there is room to improve it, but I can export sites with very strong performance.