Halvor William SandenJan 15
Large Heydon Collider
Stop A/B testing every fucking thing. You don't need to do a fucking study to know that link underlines are expected and useful. You don't need to commission research to work out if making radio buttons look like checkboxes will make people want to defenestrate you. You don't need to A/B test whether a thing that DOESN’T LOOK LIKE ITSELF will fucking confuse people. It's self-evident.
Jan 14 at 5:18pmWeb
Show threadRaúl RedondoJan 14
@heydon It's useful for ending a discussion about design when you don't want to listen to the designer.
Show threadBruce Lawson ✅ ♫ ♿ ✌️♂️✊Jan 14
@heydon Did you A/B test this message with and without the fruity Bury St Somerset O'Groats cussing?
Show threadLarge Heydon ColliderJan 14
@brucelawson It's under review LOL
Show threadRoma KomarovJan 14
@heydon “A/B testing”, also known as “Deceptive Patterns Justification”.
Show threadLarge Heydon ColliderJan 14
@kizu Haha, spot on
Show threadAmadeus MaximilianJan 14
@heydon Thank you for using the word defenestrate. 😊
Show threadJared White (ResistanceNet ✊)Jan 14
@heydon A/B testing inherently misses crucial data. You don't know if something being slightly more “popular” is also pissing off a ton of other people vs. a slightly less popular solution is also much more neutral. NUMBER GO UP doesn't always mean much. And sometimes a feature is actually good if it gets used *less*…maybe it's a sign some other part of the system is fulfilling a need better or even some other product entirely. (Shocker!)
Show threadLarge Heydon ColliderJan 14
@jaredwhite metrics for charlatans