Chasing Perfection: Talk and Links - SeaGL 2026
Morgan Murrah
Editors note: Life has been hectic. I put this online very fast. I will update it and prune it later for better readability and access. This is a medium draft.
Slide (marker where each slide is intended in the talk)
Hello everyone… Conditions are perfect… For 20 minutes we are going to look at building simple effective websites to achieve particular kind of perfection. This kind of perfection may not work for everyone but I aim to show that it can work and is not just a dream. This talk is called Chasing Perfection. Websites that pass all the tests.
Slide
My name is Morgan Murrah. My personal website is https://morganwebdev.org.
I see web development as a craft even though times can be challenging. I stick with it and believe it you are here listening, in person, remote or recording, you may be interested in it as a craft too.
I contribute to web standards at the world wide web consortium. Ive made contributions to WCAG and the Web Sustainability Guidelines. These are large scale comprehensive community projects to help people learn from evidence and experienced based knowledge.
Slide
We are going to be talking about a couple of things about the why and what kind of perfection we are aiming for with websites:
Performance - How fast does it load, is it glitchy or sketchy looking when it loads? Or does it kind of appear flat in place without any delays?
Accessibility - Can people with varying abilities use your website as much as possible and use key features.
Sustainability - Are you using the most efficient solutions for your service, such as deploying a static website with a folder of assets instead of an application server that adds friction you may not need.
Slide
In terms of What, the concrete experiences we hope for are things like:
First vs. Repeat visit experiences. Ideally a strong first load, and an even almost instant repeat visit.
Goodwill. This is an accounting term I like that means the intangible value in a transaction. Your brand, your org or you yourself gets a boost when the website creates a nice experience even if you dont exchange anything tangible.
Search engine results are kind of grey topic but if you do things right you put your best foot forward to appear well in search if thats what you want.
A static website is possible to quickly deploy a website on a variety of providers for free or cheap, without feeling locked into an application server or other configuration. It becomes possible to drag a folder of assets and drop it in instead of installing on a service. Freedom.
Usability by people you want to reach who may be using assistive technology such as a screen reader or relying on a keyboard. They may be an intended audience or just the benefit of making things work for everyone as much as possible.
Slide
In terms of testing to try and get to perfection, we are going to test the actual code thats delivered into the browser on the front end. Testing what reaches the user, at the end of the line rather than testing purely on your code in your dev environment, even if it has benefits too. Having said that, today we are just considering lab data - this is sometimes called synthetic data - in contrast to real user measurement or field data of real people.
These tests often produce a score or ranking or grading. The Green or mostly green results can often be a significant hit of dopamine. Many people take screenshots of their scores and share them if they reach 100s. And then if they get really over confident post them on Linkedin, and then people drag them in the comments. for being misleading but I digress
Slide
Lighthouse for example if you achieve perfection shows you a confetti celebration animation when you achieve a top score. Thats a fun little perk, but what does it mean really in terms of being a lab test? Is this confetti meaningless?
If you go online, often you will find people saying it is. That this is a toy test. Some trivial.
In terms of perfection, We are going to entertain running these tests on websites that get the 100s. The 0 issues found results. The A Grades. They do exist, and can do things in the real world that are non trivial. And also ask ourselves, why isn’t it like this good feeling all the time with these tests? do these tests even matter in the real world, most of the time?
My chase of perfection all started with one website a few years ago and why I am here today: https://WhatcomPublicHospital.org Whatcom is the name of the county in the State of Washington, where I live and it has no major public hospital district… yet. But I made a website about it.
Slide
The simple design and effective content of the website resonated with people. and I did it on a freemium hosting plan and I have never had to pay for bandwidth or a server, or place ads, used no analytics. It communicated something that mattered to people.
- Indexed in 2024 in search, was quickly and strongly indexed
- Very influential in engaging local conversation
- Interviewed and cited in Newspaper, based on this website
- Has a simple form and interactive dropdown elements that work with assistive technology. Important because of the audience.
Slide
So, regarding running tests on this website, Whatcom Public Hospital.org .
Lighthouse is an open source test that can run on the command line or is commonly run in chrome browser dev tools. It has 4 major audit areas, Performance, Accessibility, SEO and Best Practices. You can run it locally on your own machine, with your own internet, and see how it goes. Or can use a service like PageSpeed Insights to run a remote test from somewhere else.
https://yellowlab.tools is an open source performance testing tool with a free implementation online and also can be run by command line. It is more basic and less flashy than some of the other tools, but offers good a good checklist of items that most sites should be following. Really useful for gut checking if your website is basically too big, images are too big, if youve got too much CSS, still.
The W3C Validator is a powerful but potentially a harsh guide, as it can be very strict. Try running it on almost any production website and it will likely find some issues. Great for making what people call Semantic HTML with the feedback it gives as it really will insist on full semantic usage of elements. I do this on the website.
Axe Devtools and Arc Toolkit are browser extensions to catch automatically scannable accessibility issues. These are just a small footprint of all the potential accessibility issues. But if you have issues on the auto test, you are going to have issues on the manual tests too and real users. So we want to have a perfect score here.
As of a day or two ago, I can say I tested it with a bunch of these tests and it got perfect or close to perfect scores. Ive made two other websites, https://wsg-o-matic.com, and https://guideline-journey.com, that also get amazing scores and are worth a look.
Slide
WSG-o-matic is about clicking a big button or an emoji of a box and getting a random selection of guidelines from a massive document. It only loads the data on a first interaction and is entirely static. Something like less than 15kb on first load. Everything else after that is cached in its design.
Slide
Guideline-journey.com is an accessibility educational project. Ive been told by maybe like 5-10 students its been an aid to them directly studying WCAG.
I can’t emphasize enough how my little static website made me feel so empowered and could affect some kind of change. I want to show you some thoughts about how to make one like this for yourself today.
Slide
Lets start somewhere with no constraints. Lets consider your intended audience. Your own project, for your own purpose, with nothing riding on the experience saying it has to be done a certain way… Except for the front-end tests that are to follow I suggest.
Lets set this up this worldview about chasing perfection: Using boilerplate, process and tools. There may be no shortcuts in some areas, but this is the way.
Slide
- Boilerplate: Our chosen beginning. We have to start somewhere.
- Process: Crafting our website. Building it. Probably adding imperfections as we go.
- Tools: A sample of tools, to test the website as we go and mend it back on track.
First, Boilerplate is basically a selection code for starting a project that is ideally pretty good, it ideally prevents errors and astonishment. Importantly it is easy to delete whatever is included that you dont actually want. That is Why we start with https://github.com/h5bp/html5-boilerplate#quick-start .
It has lots of downloads, lots of commentary, thought that went into it. Better than some random boilerplate from somewhere.
Slide
I suggest there is a 5 step process to creating from boilerplate a useful website:
This is a generalized high level view, but this is the way I see. There is no easy shortcut for learning about these processes or HTML or CSS but ideally you will have a really apparent feedback loop that helps you along.
This process primarily works from the front end, whats delivered to users, and then the journey is up to you to trace it back to your build process. Not the other way around. This is I suggest why so many tests fail: failure to put yourself into the shoes of a real or at least even simulated user first. Or, as we can consider, how it will work for Bots and even AI Agents benefit from these kinds of tests and their output if those are things you want.
Slide
Chasing perfection with so many tests? Surely that must be difficult.
Indeed, the real world is full of quizzical facts. and sobering ones too.
A lot of websites that do our daily business, daily bread, work fine for millions or billions of people potentially, don’t get perfect score on any of these tests. I thought about trying to ask the audience for a random website to try one from the wild, but that seemed chaotic and would anger the live demo gods, so I decided not to.
Slide
Did you know if you run the W3C Validator on itself, it comes up with informational warnings about itself? There are lots of potential warnings and issues that may not seem to really matter to many people’s noticeable experience.
Yellow Lab Tools still does a check to see if you have duplicate jQuerys installed. Thats a bit of a throwback if you dont know that reference, as jQuery is a bit dated for a new project. But you dont want a duplicate either.
It is very possible to have a not perfect Lighthouse score and do really well in the real world because of faster connections and strong devices possibly than your tested conditions. That’s where you may want to consider field data, but thats mostly for big websites as it takes a bit of setup, where as these checks can be run on any sized site – big or small.
Accessibility is deeply important, but you will be dissapointed how many websites have common major errors on simple quick scans. Over 95 % of major website homepages have at least one automatically detectable Web Content Accessibility Guideline (WCAG) failure in them.
- Axe DevTools uses powerful open source libraries but has business upsells in it
- Arc Toolkit uses a different rules engine which is good to try but sometimes has false positives at different viewports
Sustainability is incredibly important and ofter undervalued. Ecograder is great and new on the scene about sustainability, but well… its a bit opaque. Shifting your hosting provider if its not considered green hosting is not always easy. What is Green hosting? Its debated a lot and discussed still. Watch this space.
Slide
So, this perfection thing sounds a bit well… it sounds a bit less clear than when we started. These tests do matter, but ultimately you cant let yourself never finish, you need to put what you made out there to users and test it yourself to make progress.
Slide
Use your imagination for a moment and imagine a performance dashboard that kept lit up with perfect scores. Would we continue to learn anything? Would we keep going or would we consider ourselves finished and limited?
Well its true that a perfect score doesnt mean a good usable website under the principles we outlined at the beginning. But thats not exactly what we are entertaining for this talk. This talk is still about the chase of perfection. And I hope to have showed you a little about it today. Thank you.
Slide
Questions - Thank you all!!