Tech Giants' Water War: Rethinking the Boundaries of Interaction Design

The most valuable information in this article is that the physical resource demands of AI data centers are shifting from hidden costs to explicit conflicts, and the verdict of this conflict will directly impact the sustainability of the digital services we use every day.

Water Resource Dispute: Blind Spots Outside the Design System

Imperial Valley Computer wants to build California's largest AI data center, requiring massive amounts of river water from the Colorado River for cooling. After the water supplier refused, the developer sued directly. This sounds like legal news, but from a designer's perspective, this is a classic case of "resource allocation interaction failure."

Our design systems cover pixels, spacing, and component libraries, but never touch water, electricity, or land. In ByteDance's design system documentation, I wrote hundreds of rules about alignment and spacing, but not a single one about "how many liters of water this function consumes per call." This isn't satire, but a reminder.

The operation of any digital product depends on physical infrastructure. AI training and inference consume extremely high energy, and cooling requires large amounts of water. When developers view water resources as "purchasable production materials," and water suppliers view water as "public community resources," the information architectures of both sides are completely misaligned. This misalignment is called "inconsistent user mental models" in interaction design, but here the consequence is courtroom confrontation.

From Pixels to Water Drops: The Hidden Costs of Interaction Design

The dried-up riverbed of the Colorado River in this picture illustrates the problem better than any "loading state" in a design mockup.

We optimize button click feedback, reduce page jump steps, and improve animation smoothness every day, but rarely ask: How much extra computing resource does a 0.1-second loading optimization require? And how much water and electricity do those computing resources need? The "side effects" of design decisions are usually not considered in design systems, but they exist in reality.

For example, an AI chat assistant requires massive GPU computation for every reply. If designers make the model regenerate full data on every user input for "instant response," instead of using cache or incremental updates, they are pretending that user needs outweigh physical limits. This isn't experience optimization; it's resource waste.

User Perception and System Transparency

There is a principle in design systems: Provide users with clear system status feedback. Like loading animations, progress bars, and error prompts. But "system status" only covers the software level, never the hardware level. Users don't know how much water evaporates behind the scenes when they click "Generate."

If designers could add visual hints for "carbon footprint" or "water footprint" in the interface, even just a small icon, it could change user behavior. This isn't moral coercion, but information transparency. Just like WeChat Pay showing "Recipient has received payment" during transfers, users should know "This operation consumed 0.5 liters of water."

Of course, this requires collaborative design between frontend and backend. But more fundamentally, we need to redefine the boundaries of "design systems": It shouldn't just include colors, fonts, and spacing, but also visibility of resource consumption. This isn't a technical issue, but a design ethics issue.

Imperial Valley Computer has a high chance of losing the lawsuit, but even if they win, the scarcity of water resources won't change. In the future, more data centers will compete with communities for water sources in water-scarce areas. As designers, we have a responsibility to make users aware of this competition at the interaction level, rather than masking the underlying cost with "smooth experiences."

Wrapping up directly, no summary.

Original Link: https://www.ithome.com/0/982/244.htm