Labeling binary pairs
[I wrote this in June 2006 but never got around to publishing it anywhere. It's not terribly interesting, but having gone to the trouble of writing it, I feel it ought to live somewhere other than my laptop.]
A while back I noticed a real-world design pattern which I'm fairly certain has no online equivalent. I began to see it, or the potential for it, everywhere I looked.
My best description of the pattern would be: given a binary pair of objects, only one of the pair need be labeled. The value of the other can be deduced by process of elimination.
A quick example of this would be the hot and cold taps on a sink: if one of them is marked as the hot tap, the other must be the cold tap. It's not necessary to label both taps for a person to successfully use the sink.
# Informing on a budget
This basic concept dawned on me when I got up to go to the loo at a cheap Vietnamese restaurant. The staff pointed me towards an alcove containing two doors. I glanced at the first door, saw it was marked "Women's", and started moving towards the second door, which I presumed was the Men's. As I reached it I noticed this second door had no "Men's" sign. It was blank. Apparently it didn't need any kind of marking for people to know what was behind it; it only required simple deduction.
Since I immediately had a little time to ponder things, I put the old noggin to work on why the restaurant might have labeled only one of its two bathroom doors. I concluded it was a question of cost. Why should a low-budget restaurant pay for two toilet signs when it only takes a moment of the patron's time to work out which door is the right one for them?
# Manufacturing
That evening, as I got ready for bed, I noticed that my contact lens case only had one of its two lids marked with a letter. The right lid had a big 'R' for 'RIGHT' on it, while the left lid was smooth and blank.
I presume that this allows the manufacturer to reduce tooling costs somehow, perhaps when it comes to the international market. The right lid can be labelled in a variety of languages, while the left lid remains generically and affordably blank.
# When the idea breaks down
I mentioned to Chris Wetherell what I'd observed and he pointed out its weak point; namely that when the objects become physically separated it's left to people to remember which is the marked and which is the unmarked object.
To illustrate this flaw, think about a salt pot and a pepper pot on a dinner table. One pot is labeled 'S' and contains salt. The other pot is blank and contains pepper. Both pots are opaque so you can't see their contents.
Placed together at the start of the meal it would be easy to say which pot contained salt and which contained pepper. But as the meal progresses and the two pots are separated and passed around the table, a person who picks up the unmarked pot might start to wonder which one they have in their hand. Was it the salt that was unmarked? Or the pepper?
So Chris is right -- any time it takes more than a flick of the eyes to confirm which of the two objects you're dealing with, the whole idea seems to fall apart.
# Extending this beyond a single pair of objects
I also talked to Paul Nixon, and together we realised that this selective labeling isn't limited to things that come in pairs, but extends to things which come in lots of pairs all joined together, like the measurement marks on a ruler.
By only labeling major points along a ruler's length, manufacturers save money and reduce visual clutter. Yet we're more than capable, albeit after a pause, of finding the value of any point on the scale.
And you know, watch and clock manufacturers do exactly the same thing: label a few points, save some money, keep the interface clean, and let the human mind fill in the blanks.
# This idea is everywhere in the real world, so why don't we see this on the internet?
Since noticing this I've looked everywhere for examples of it on the internet, and I can't say I've found any. Why is that? Well, as a quick guess I'd say:
(1) It's not really any cheaper to produce a blank button than to produce one which has some helpful text on it.
(2) Very few things on the internet always come in pairs, and have consistent labels across sites. Without consistency of function and name (hot tap, cold tap) it becomes very hard to use this idea with any success.
(3) Bugs! If we chose not to label one of a pair of buttons people would just presume there was a bug in our code. On the whole the internet's so rubbish that we couldn't blame people for thinking we'd just made a mistake.
(4) As for the joined pairs scenario, the internet's not big on ruler- or dial-type scales for controls. (Though I guess that's slowly changing.)
# This reminds me...
This whole thing reminds me of a Jimmy Carr joke:
"I was talking to my Nan about Ant & Dec. She said she didn't know which one Ant was. I said 'Do you know which one Dec is?' She said 'Yep.'"
Labeling binary pairs
[I wrote this in June 2006 but never got around to publishing it anywhere. It's not terribly interesting, but having gone to the trouble of writing it, I feel it ought to live somewhere other than my laptop.]
A while back I noticed a real-world design pattern which I'm fairly certain has no online equivalent. I began to see it, or the potential for it, everywhere I looked.
My best description of the pattern would be: given a binary pair of objects, only one of the pair need be labeled. The value of the other can be deduced by process of elimination.
A quick example of this would be the hot and cold taps on a sink: if one of them is marked as the hot tap, the other must be the cold tap. It's not necessary to label both taps for a person to successfully use the sink.
# Informing on a budget
This basic concept dawned on me when I got up to go to the loo at a cheap Vietnamese restaurant. The staff pointed me towards an alcove containing two doors. I glanced at the first door, saw it was marked "Women's", and started moving towards the second door, which I presumed was the Men's. As I reached it I noticed this second door had no "Men's" sign. It was blank. Apparently it didn't need any kind of marking for people to know what was behind it; it only required simple deduction.
Since I immediately had a little time to ponder things, I put the old noggin to work on why the restaurant might have labeled only one of its two bathroom doors. I concluded it was a question of cost. Why should a low-budget restaurant pay for two toilet signs when it only takes a moment of the patron's time to work out which door is the right one for them?
# Manufacturing
That evening, as I got ready for bed, I noticed that my contact lens case only had one of its two lids marked with a letter. The right lid had a big 'R' for 'RIGHT' on it, while the left lid was smooth and blank.
I presume that this allows the manufacturer to reduce tooling costs somehow, perhaps when it comes to the international market. The right lid can be labelled in a variety of languages, while the left lid remains generically and affordably blank.
# When the idea breaks down
I mentioned to Chris Wetherell what I'd observed and he pointed out its weak point; namely that when the objects become physically separated it's left to people to remember which is the marked and which is the unmarked object.
To illustrate this flaw, think about a salt pot and a pepper pot on a dinner table. One pot is labeled 'S' and contains salt. The other pot is blank and contains pepper. Both pots are opaque so you can't see their contents.
Placed together at the start of the meal it would be easy to say which pot contained salt and which contained pepper. But as the meal progresses and the two pots are separated and passed around the table, a person who picks up the unmarked pot might start to wonder which one they have in their hand. Was it the salt that was unmarked? Or the pepper?
So Chris is right -- any time it takes more than a flick of the eyes to confirm which of the two objects you're dealing with, the whole idea seems to fall apart.
# Extending this beyond a single pair of objects
I also talked to Paul Nixon, and together we realised that this selective labeling isn't limited to things that come in pairs, but extends to things which come in lots of pairs all joined together, like the measurement marks on a ruler.
By only labeling major points along a ruler's length, manufacturers save money and reduce visual clutter. Yet we're more than capable, albeit after a pause, of finding the value of any point on the scale.
And you know, watch and clock manufacturers do exactly the same thing: label a few points, save some money, keep the interface clean, and let the human mind fill in the blanks.
# This idea is everywhere in the real world, so why don't we see this on the internet?
Since noticing this I've looked everywhere for examples of it on the internet, and I can't say I've found any. Why is that? Well, as a quick guess I'd say:
(1) It's not really any cheaper to produce a blank button than to produce one which has some helpful text on it.
(2) Very few things on the internet always come in pairs, and have consistent labels across sites. Without consistency of function and name (hot tap, cold tap) it becomes very hard to use this idea with any success.
(3) Bugs! If we chose not to label one of a pair of buttons people would just presume there was a bug in our code. On the whole the internet's so rubbish that we couldn't blame people for thinking we'd just made a mistake.
(4) As for the joined pairs scenario, the internet's not big on ruler- or dial-type scales for controls. (Though I guess that's slowly changing.)
# This reminds me...
This whole thing reminds me of a Jimmy Carr joke:
"I was talking to my Nan about Ant & Dec. She said she didn't know which one Ant was. I said 'Do you know which one Dec is?' She said 'Yep.'"