Fascinating section in The Humane Interface by Jef Raskin, arguing against software customization.

I don't agree with his conclusion, but he still raises some great points. Some thoughts: ⬎
Quick caveat before jumping in: I don't think it's productive to ask "is software customization good or bad?", although that might fit better into 280 chars

Much more fun to consider: when is it good, for whom, in what contexts, what are the tradeoffs... ⬎
Alright, diving to the heart of the matter:

For Raskin, the "central point" is this: if the designer can make an "optimal" interface, why would you let the user mess with it? They can only screw it up. ⬎
I think this idea of a single "optimal" interface really gets at the core tension here: how much do people's needs really vary?

The Air Force found out that there's no single optimal cockpit seat for every pilot: ⬎
The more variance in people's needs, the more this idea of an "optimal interface" falls apart:

* niche professional tasks where people have very specific needs
* people w/ disabilities, further from the average ⬎
On the other hand, there are (mostly) universal truths that hold across humans, many of which Humane Interface eloquently explains.

Seems like the key question here is, how much do people's needs vary in a given domain? ⬎
Here Raskin makes 3 claims: 1) customization sounds nice and fun and all. 2) But it doesn't improve productivity, 3) and it harms learnability. Let's take each of those arguments in turn. ⬎
His first point hits hard. I often worry that I want customizable software because it's fun, and this is dangerous if we're deluded into thinking we're making useful things.

But, it's worth remembering that fun can be a goal too: ⬎ https://twitter.com/geoffreylitt/status/1254774726074884096
Second point: there's no evidence customizability improves productivity.

Maybe true, but my own experience as a programmer makes me extremely skeptical of this claim. I think it's clear that customizability can improve productivity, in the right contexts. ⬎
Third point: customization harms learnability. This one seems generally correct to me. So then it becomes a question of learnability vs usability. For power tools, learnability matters less. ⬎
quick sidenote: getting a bit tired of learnability being prioritized infinitely higher than utility. I have a ton of respect for learnable design, but feels like "no manual necessary" is dogma now, built on assumption of web apps that people are only using for a few minutes ⬎
OK, few more points from Raskin:

Customization creates modes, and modes are very very evil. Set something in Word preferences, and it's a "time bomb" ⬎
I see the point here, and have had bad experiences with hidden config. But also, calling it a "mode" feels slightly off to me -- once I set up a config, it's often fixed for years and I build muscle memory around it. Not modal in the moment. ⬎
Raskin points out we need stability in our software.

Totally agree! In fact, this is a common reason that people customize software -- to change it back after an unavoidable upgrade (see this paper by Wendy Mackay https://www.lri.fr/~mackay/pdffiles/CHI91.Triggers.pdf) ⬎
I have a lot of respect for Raskin, Humane Interface is a fantastic book. And the Macintosh was pivotal for empowering more people to use computers. So I think his critique is worthy of serious consideration for anyone who wants more malleable software...
Also, always fun to realize that you've already read a page in a book on Twitter... 🙃 some good discussion of this here: https://twitter.com/stevekrouse/status/1229883092606869508
You can follow @geoffreylitt.
Tip: mention @twtextapp on a Twitter thread with the keyword “unroll” to get a link to it.

Latest Threads Unrolled: