- “So, this is a great technology!”
- “What are you talking about?”
- “Well, everything’s just automatic. You don’t have to click anything, you don’t have to do anything. And it saves you 5 seconds.”
- “Hunh… That doesn’t sound all that great.”
- “No- it’s incredible!”
- “I mean, right now, I can just click on it, and it just works. I don’t see what you’re getting so worked up about.”
When people hear about technology that shaves 1, 2, maybe 3 seconds off of some common task, or removes just 1 or 2 user interactions with something, they frequently say, “Why should that matter? What does that have to do with anything? Why does anybody care?”
Some even get defensive:
- “Why would I need that? Are you suggesting I can’t wait the 5 extra seconds it requires to do this by hand? Why do you have to try and go so fast all the time?”
There’s something about reaching a magical threshold, that suddenly means people use technology differently.
This is not just the case for technology, but for economics as well. When something reaches a magical price-point, suddenly there is a rush of traffic to buy that item.
change changes effect effects small tiny great large big tippingpoint priority priorities
- The difference between ssh’ing to another server to hand-edit HTML blog posts, to actually using a blogging engine to write your blog posts.
- The difference between wiki markup syntax, and actually writing out HTML by hand.
- The difference between looking up URLs, and just using LocalNames.
- The difference between using CVS to collaboratively write a document, and using wiki.
In economics, there’s the concept of the “criticial price point” or something like that. After the magic price point is reached, the gates flood open.
Causes: The Priority
Why does this happen?
A lot of things, where SmallChangesYieldGreatChanges, happen because a process breaches the priorities queue.
- Someone might have something to contribute to a discussion.
- But there’s a cost in contributing to the discussion.
- If the cost is sufficiently greater than the benefit, there’s no contribution.
So it’s like a binary switch.
Now, that comment may have resulted in a dramatic change in the group’s decisions. Many comments, from lots of people, may have resulted in still greater effect.
Cost can be:
- time (it may take too long)
- trouble (requiring sustained, boring, concentration)
A TechnologySolution such as InkscapeToOddmuse or the ZergHive? can help alleviate the LongImageIncorporationProcess, and other ContentRouting problems. The result can be radically different behavior on the wiki, or wherever. In our case, it was far greater use of VisualLanguage. Before, we still wanted to add pictures, but it just took way too long to go through the hassle of constructing SVG and PNG images, to really be perceived to be worth the bother.
Usability & Performance
There was an interesting study (link?) that showed that usability ratings would go way up, when computers simply performed faster, rather than slower.
Some features that are never used, because they are too slow, suddenly become star features, once they pass a particular responsiveness threshold.
As many of you know, I run into this a lot.
I frequently run into examples of this, but have difficulty recalling them.
I’m immediately thinking:
- Web sites need to be rendered within X many seconds, otherwise people would rather just pick up a magazine next to them. (PhilipGreenspun’s own words.)
- Something that is supposed to be “live” interactive must response within 200-250 ms, otherwise people perceive significant lagging, and distrust the device. (There’s a reference to some psych papers about this in Stevens.)
I frequently run into explaining why this matters when explaining:
- Why OneBigSoup:LocalNames matter, and why it changes, fundamentally, how people think about linking, and frequency in linking. (The classic stop sign: “How does saving me 5 seconds of googling mean that I’m more likely to link? What difference does it make? I can just look it up, any time I want.”)
- How establishing fluidity between mediums will make a very different user experience than we experience now.
Most of my examples are on a much longer timescale. If something takes more than 10 years to do, most people aren’t going to even bother starting. Examples:
- manned mission to the moon: “… commit itself to achieving the goal, before this decade is out …” – President John F. Kennedy. Some people claim that there was no way of maintaining that momentum for more than 10 years.
- Clock of the Long Now
- writing an encyclopedia
- Travel. When it took months to travel to the other side of the continent or oversees, most people never met a “foreigner”. Now that people can travel to the other side of the world in a day, most people have met quite a few people from other countries.
- Before 1450, only incredibly wealthy collectors had an entire shelf of books, because it took so long / was so expensive to produce each book.
Perhaps closer to what you’re looking for:
- Once every engineer had a book of trig tables close at hand. Now that hand calculators can get the same result to adequate precision faster than flipping to the right page in the book, why bother ?
- dictionaries are on the verge of suffering the same fate, with “pocket dictionaries” and http://dictionary.com .
I have a few examples at the millisecond timescale:
- Once computers got fast enough to play arbitrary sound waveforms in real time, suddenly I started hearing arcade games emit relatively human-sounding speech.
- Once computers got fast enough to redraw the entire screen 60 times a second … no longer limited to erasing and carefully re-drawing tiny little characters moving across a fixed background.
- Once computers got fast enough to render the entire screen in 3D 60 times a second …
I’ve also heard of the “magic $499 threshold” – keeping something cheap enough that people who have a $500-per-transaction limit on their credit card can put it on credit.
Other than time and money, what other thresholds are there ?
Some interesting “facts” from the world of usability experts:
- The expected time of seeing feedback after performing an action is 0.1s (100ms). Anything longer feels sluggish and not trustworthy. Note that the “feedback” can be as small thing as changing the color of a menu entry, moving a cursor to next line or changing the button graphics to “pressed”.
- The average wait time for response in conversation is about 1s. It’s also the expected time to see the actual results of your action, not just a pressed button. If you don’t provide results or a progress bar within 1s, expect random clicking.
- The maximum focus time is estimated as 10s (this takes into account people who are busy or distracted too). No single operation should require more than this to complete – if it does, people will forget what they were doing before and fumble.
- The maximum capacity of short term memory is estimated at 4-8 items, but we have to remember that other people don’t live in front of the screen and they usually have some unfinished items in their short term memory already when they use the computer.
Thank you, Radomir. Especially for pointing out the 2 kinds of feedback – the immediate “yes, I’m responding to you”, and the possibly much later actual results. Many kinds of asynchronous communication completely skip that first kind of feedback. Is it a good idea to try to manually add it back in? For example, is it a good idea to immediately respond to email with something like “Just got your email; I’ll be pondering this and should get a real response by next Thursday”?
Another example of SmallChangesYieldGreatChanges:
"In brief praise of DVCSs" by Aristotle Pagaltzis says: “It seems like small potatoes, but when someone turns something that takes five actions and has a free variable into a single action that requires no thinking whatsoever, it changes the quality of the game. You start creating throw-away repositories for the littlest reason because the only decision left to make is whether to create one or not. It’s a classic example of the tyranny of choice.”
(tangent) So what is this “TyrannyOfChoice”? Is it related to ThePowerOfQuestions? (/tangent)
I see that "User Interface Design For Programmers" by Joel Spolsky 2001 discusses how tiny frustrations that “seem too petty to dwell on” have a large effect. There’s a little bit about LearnedHelplessness?. A little bit about how adding completely non-functional chunks of rubber to a camera can subtly guide a person into holding it properly.
There is a (French, I think) saying stating that “good first impression consists of around 700 irrelevant details”. I would say that it’s the same with good usability, and a lot of other areas where you depend on first impressions. It gets critical when you want the users to get into the “flow” – because the tinniest distraction can spoil it.
As for the “I’m working on it” messages, they are a good idea, as long as you do them manually. Automating them removes all credibility, and as soon as the receiver notices they are in fact automatic response, they just become an annoyance. False feedback is worse than no feedback.
- TippingPoint (may want to move this there, and redirect to it from here..!)
- PersuasiveContent lists some “little things” that get in the way.
- MixedAudioBlogs describes something that would be really nifty, if only it were a little easier to do.
- LimitationsOfMailingLists describes many minor annoyances
- PlainTalk – perhaps if we make web pages and encyclopedias just a little bit easier to understand, vastly more people will understand them.
- WhatIsAffordance – sometimes we make small, apparently “non-functional”, changes in the shape or size of buttons and handles. Those small changes can be the difference between almost everyone being able to use it right away, versus than almost everyone being frustrated and annoyed and taking much longer to figure out how to use it.