The Illusion of 98%

At first glance, 98% sounds like an overwhelming success. However, the value of that percentage depends entirely on the context.

The Contextual Divide

Depending on what is being measured, 98% can be a miracle or a catastrophe. Consider the following comparisons:

---

"98% is a fantastic metric for exceptional gains—such as a massive leap in someone's quality of life—but it is a failing grade for basic expectations, such as a baby surviving the care of a babysitter."

The Scale of Digital Exclusion

In the realm of web design, "wide support" is often a misleading term. If a developer implements a flashy new browser feature that works for 98% of the global population, the math reveals a harsh reality:

$\text{Global Users} \times 0.02 \approx 150,000,000 \text{ people}$

By optimizing for the 98%, we are effectively welcoming evicting 2% of the audience. Imagine a physical venue refusing entry to 2% of its loyal customers simply because the owner "improved" the experience.

General Population vs. Specific Audience

It is a mistake to assume that the general population reflects your specific user base. The following diagram illustrates this discrepancy:

A Real-World Case Study

I recently encountered a situation where a feature was deemed "standard" and "safe" since 2023. I wanted to simplify my workflow by trimming out the scss pipeline. However, upon analyzing a client's actual browser distribution over the past year, the data told a different story:

/* New CSS feature that is "widely supported" */
.container {
  display: grid;
  gap: 1rem; /* Example of a feature that might lack legacy support */
}

While the general web might support this at 98%, only ~70% of this specific client's visitors could actually use it. By following the "general" rule, 30% of the audience would have been left behind.

The Philosophy of Robust Engineering

True engineering excellence is not defined by how well a system works for the majority; it is defined by how it handles the edge cases.

If a feature lacks a strategy for graceful degradation, then claiming it is "widely supported" at 98% is a fallacy. It has failed the most basic requirement for 2% of the population.

---

Final Takeaways

• Recognize that 98% is not always "enough."
• Distinguish between general statistics and specific audience data.
• Prioritize graceful degradation over "fancy" features.

Get in touch: Reach out via my public inbox: ~whynothugo/public-inbox@lists.sr.ht