Proposal for an indicator of obsolescence for content websites
Rédigé le 12/09/2024 par Youen Chéné
Objective of this article
This article aims to propose an indicator to detect if a content website causes hardware obsolescence due to the use of too recent technologies on a website.
This indicator is currently named WCOI for Web Content Obsolescence Indicator.
This is a proposal that will undergo several iterations internal or external to Webvert.
This indicator proposal is under the CC-BY-SA license. (You can remix, improve, use commercially… in all cases, the article and authors must be cited.)
What type of obsolescence are we talking about?
The type of obsolescence targeted is hardware obsolescence caused by software.
In 99.9% of cases, it is not planned obsolescence, but structural obsolescence of a sector that has experienced exponential improvement in the performance of its systems (Moore’s law) and therefore significant and recurrent renewal of hardware.
Now that digital technology is very mainstream, this renewal causes two-thirds of the impacts of digital technology on the environment (in greenhouse gases, minerals, water deprivation and pollution, abiotic resources).
What is the objective of WCOI?
This indicator dedicated to content websites was created to allow teams to arbitrate:
-
on one side between strong content optimization (influence on energy performance and partly on minimizing terminal renewal through slowness felt by users),
-
on the other side on compatibility with a maximum of terminals (influence on manufacturing via minimization of terminal renewal).
Indeed, in countries like France, Switzerland, Sweden, Norway, the energy mix is very low (20g to 70g per kWh), the impact of manufacturing is the most influential part if we want to minimize the environmental impact of a digital service.
In countries like Germany, the American continent or Asia, the energy mix is higher (300 to 900g per kWh), we will only mention the impact of the energy and data center part.
The objective is therefore to add a second indicator to give information to website creators and content managers so that they can make the right trade-offs between optimization and backward compatibility.
This second indicator is complementary to energy efficiency indicators such as the Sustainable Web Design Model, One Byte Model, Ecoindex… see our state of the art of environmental assessment of content websites.
WCOI Specification
Web Content Obsolescence Indicator is an indicator that ranges from 0% to 100%. 100% means backward compatibility on a maximum of terminals. Depending on renewal speeds, generally terminals over 12 years old.
The equation is as follows:
totalImages is all the images referenced in the group of pages analyzed.
totalWebp is all the WebP format images referenced without the presence of a fallback in a historical format (jpg, png, gif, svg).
totalAvif is the equivalent with the AVIF format.
totalJxl is the equivalent with the Jpeg XL format.
As a reminder, a fallback is coded as follows in HTML:
<picture>
<source srcset="/offre/img/screen_portfolio_fr.webp" type="image/webp" />
<img src="/offre/img/screen_portfolio_fr.png" alt="Screenshot of site list" width="200" height="411" />
</picture>When a fallback is implemented by coupling a new format with a historical format, then the image is not counted in totalWebp, totalAvif or totalJxl.
Explanation of weighted averages
The different totals by modern format are weighted via their level of backward compatibility.
The source data used is as follows:
| Format | Can I Use - Incompatibility Rate | iOS - Incompatibility Rate | Android - Incompatibility Rate |
|---|---|---|---|
WebP - 2020 (Safari). |
4.31% |
3.44% |
0.30% |
AVIF - January 2024 (MS Edge) |
7.36% |
11.64% |
(unreliable) 51.40% |
JPEG XL - Future Date - (safari only compatible) |
86.77% |
24.99% |
100% |
The additional information on the first column is the last event where all browsers declared their compatibility in their latest version.
The bias that we can easily have is that a small percentage can represent a huge volume of terminals in 2024. For example, 0.5% of the Android fleet represents 30 million terminals (2023 figure) and 0.1% of iPhones represents 2.5 million terminals (2023 figure). See the article on WebP for more details.
To compensate for this bias, weighted averages were created on the last 10% of compatibilities. On the data side, the percentages from the CanIUse site were used.
| Format | Can I Use - Incompatibility Rate | Capped at 10% | We determine the weighted average on the last 10% of incompatibility |
|---|---|---|---|
WebP |
4.31% |
4.31% |
4.31 |
AVIF |
7.36% |
7.36% |
7.36 |
JPEG XL |
86.77% |
10% |
10 |
Total |
21.67 |
This strong hypothesis on the last 10% of compatibilities is a basis for discussion.
The weightings will also need to evolve over the years according to the natural evolution of the market.
Roadmap
-
v0.1 (Current) - Inclusion of images not displayed in some terminals
-
v0.2 - Inclusion of videos not displayed in some terminals
If you wish to discuss or contribute to this indicator, don’t hesitate to contact us at contact+wcoi (at) lewebvert.fr .
Bonus: how was the idea for this indicator born?
The idea of the Web Content Obsolescence Indicator was born from 2 events in 2024.
The update of the RGESN version in May 2024 which is very aggressive or clumsy on energy optimization via the use of too modern formats without taking into account the trade-off to be made between optimization and the impact on terminal renewal. See the critical analysis article on RGESN.
My discussions with developers from around the world at WordCamp Turin, where the axes of discussions were solely on energy optimization. In France, due to our low energy mix, we balance discussions and actions between energy impact and manufacturing impact.