This article is more than 1 year old

Google boffins PROVE security warnings don't ... LOOK! A funny cat!

Designing a UI people care about is harder than it looks

The revised SSL warning interface introduced in Chrome 37, designed to teach users more secure behaviours, was only a partial success – according to the Chrome security team's own analysis.

Confusing security warnings serve only to make users more insecure and normalise risky behaviours, according to Google. To try and beat that, Google tried to make its interface easier to understand, and try to restrict itself to warnings when a real risk existed.

This Association for Computing Machinery (ACM) paper and presentation explain that the attempt was only a partial success.

Even with Google's resources behind them, the group that authored the paper says it's difficult to craft a security warning that is readable down to grade-six level, and will give users either the information they need to stay safe, or to guide them towards a secure decision.

“We ultimately failed at our goal of a well-understood warning. However, nearly 30% more total users chose to remain safe after seeing our warning,” the paper says.

With a redesigned SSL dialogue designed to strip out jargon and tell users they might be in danger, the best the design team could manage was about 58 per cent compliance.

Google testing warning dialogues

Best of a bad lot: this SSL warning nearly doubled user compliance, but too many users still don't get it

The Chromium security team, led by Adrienne Porter Felt, reckons there are many degrees between a genuine attack on and end user and a browser generating a warning when none is needed, but they generate the same generic popups.

If a client clock is wrong, or if an employer has a DPI box or content filter between a user and the Internet, or a client is missing a root certificate, a user will get an SSL warning that tells them nothing, and serves only to teach them to click “OK” on a dialogue.

As the paper states, “An ideal SSL warning would empower users to make informed decisions and, failing that, guide confused users to safety”.

“We attribute the low comprehension rates to the difficulty of creating an SSL warning that is simultaneously brief, non-technical, simple, and specific”, the paper concludes. ®

More about


Send us news

Other stories you might like