Consensus on the web

— Published September 8th, 2009  |  by Randall Snare

Miscommunication: an honest mistake

I did a favor for a friend. She was in grad school for clinical psychology and her class needed subjects for iQ tests. I was all too eager to have my genius recorded for the good of the psychological profession, so I agreed.

The girl came over in the morning. The test would take 4 hours, she said. The third part of the test challenged my spatial awareness. She took out puzzle and laid the thick, brightly coloured pieces on the table one by one.

"Now we're going to test how well you put together this puzzle." I made eye contact with her when she said this and I smiled.

I smiled because that's what friendly people do when they make eye contact. I saw her write in her notebook, "subject smiles at the mention of puzzles."

I hate puzzles. And my spatial awareness is nil; I may as well live in a giant hamster ball. I was amazed that the two of us, so able to communicate in theory, could miscommunicate so strikingly. Yet her official assessment of me - a complete misconception - was recorded as official, verifiable data.

Communication, perception and the internet

Miscommunication is a common problem with normal interactions. But it gets more complicated when we throw the web into the mix. In this feature I'm going to try and answer two questions:

I. How do we really assess a website?
II. How should we test that assessment?

Consensus | An experiment

In 1968, John Darley and Bibb Latane conducted an experiment. They had subjects in 3 different real-world scenarios fill out a questionnaire:

• Scenario 1: The subject was alone in a room
• Scenario 2: The subject was in a room with other subjects
• Scenario 3: The subject was in a room with other people - not subjects, but hired actors

This is what happened:

In every scenario, the room slowly filled with smoke. In the case of scenario 3, the actors were told to pretend not to notice the smoke, and continue filling out the questionnaire. Darley and Latane were recording how many subjects reported the smoke in each scenario.

• Scenario 1: the subjects reported the smoke 75% of the time.
• Scenario 2: the subjects reported the smoke 38% of the time.
• Scenario 3: the subjects reported the smoke 10% of the time.

So what does that tell us?

Darley and Latane called the fact that the rate of reporting decreased with the increase of people in the room 'the bystander effect': inaction due to the proximity of other people in the same situation (this is also known as 'distribution of responsibility'). When something is wrong and we're with other people, it's easy to assume that someone else will take care of it.

But there's more. There is an almost 30% diffrence between scenarios 2 and 3. More people reported the smoke when everyone was noticing it.

Because both rooms had the same number of subjects, the difference was the actors ignoring the smoke. There was a lack of consensus in the response.

Consensus | A definition

One of the ways we assess a situation is the behaviour of those around us. If everyone is laughing at a joke we don't get, we'll most likely laugh along.

According to the Attribution Process (attributing causes to human behaviours), consensus is the way we figure out how to respond to a situation.

In the experiment, the winning consensus (smoke = no big deal) won by a large majority, even though it ran counter to a primal human instinct - survival.

Consensus on the web

We're all alone on the web. Consensus doesn't apply when we're negotiating with a website; our interaction is solitary.

But we haven't changed as people; we still look for social clues in order to assess a situation. So what do we call consensus on the web?

Theory I: Standards

Standards in the real world: When the light turns green and everyone crosses the street, we know that green means go (perhaps we already knew that, but it's reinforced every day). If no oned moved when the light turned green, we'd think twice about crossing the street).

Standards on the web: In the web's relatively short lifespan, standards of layout and design have formed.

Are those standards a web version of consensus?

1. Web title and logo in the top left hand corner (From iqcontent.com
   
2. Navigation across the top and/or the left hand side of the page (from From engineersireland.ie
   
3. Contact us, search, other universal needs in the top right hand corner (from hibernian.ie
   
4. Small links in the footer (from apple.com
   

Conclusion

I don't think that standards are the web version of consensus. Here's why:

We have a filtering system in our brains - in neurological terms, this is called negative feedback. For example, we don't feel the weight of the clothes on our body as we go through our day. We must pick and choose what we notice, otherwise we'd be incapacitated with too much stimuli, i.e. the world.

This filtering system happens on the web (ever hear of ad blindness)? So web standards may not be the online version of consensus; rather it's stimuli we come to ignore online.

Theory II: Feedback

Before we talk about feedback, we first need to look at where and how feedback comes into play, with a thought experiment:

	A store in real life	A store online
Action:	You enter the store, check the décor: the music, smell, look and feel of the place	You enter (any page) and check the décor: the colors, images, layout, the look and feel of the site
Consensus and assessment:	There are a few people in the store -- not too crowded or empty. The customers are very well dressed so this adds to the high end feel of the place.	Here is where web standards come into play: where can you browse items, search, or purhcase?

	A store in real life	A store online
Action:	You browse the products, by the way they are physically placed around the store.	You browse the products by the way they are placed on the web pages.
Consensus and assessment:	The other customers are a bit hoity toity; they are quiet, so you are too. Browsing is done neatly. This store is probably out of your price range, but it is a very pleasant shopping experience.	You're all alone here, so your assessment is the site's usability: when you click on something, it goes where you expect it to go. Your assessment is a pleasant one

	A store in real life	A store online
Action:	You've decided to purchase something and go up to the counter. The store clerk starts speaking in tongues and foaming at the mouth.	You decide to purchase something and proceed to checkout. When you click ' 'buy' you get an error message that says: 'Errror QXv217k' The back button is greyed out.
Consensus and assessment:	Other people in the store are looking at the clerk like he's crazy. You think the clerk is crazy too.	You don't know whether you've done something wrong or it's a glitch in the site.

	A store in real life	A store online
Action:	You leave the store	You close the window, leaving that site.
Consensus and assessment:	Other people leave as well. Your money's in your pocket, so you know you didn't spend anything.	You don't know if your card was charged or not. You do know that your experience was unpleasant but you don't know what caused it.

Conclusion:

Feedback may be the way we indicate consensus online.

So now what?

Adding consensus to design principles

It may be safe to say that consensus on the web goes beyond standard usability principles: to make a site goes beyond just useful - one that's intuitive, gives you an indication that you're on the right track, has a sense of space, and talks to you. In a word: lovable.

Coming up with design principles around this idea is a big job. But here are some of my ideas:

Error messages: this is a way that a website gives us direct feedback based on our actions. To broaden this, we should think about extending them beyond just errors. If we can give feedback on things a user is doing correctly, without cluttering up a user journey, then we're on the right track.

User comments: this is another way a website gives us direct feedback - based on what we're looking at. To broaden this, we should think about incorporating feedback on other types of sites, and doing something more with them than simply listing them out as a thread.

Link text: this is a more subtle way that a website speaks to us (click on this, trust us, it will take you where you think it will). To broaden this, we should think about how to let a user to know where they came from once they've left page, and why you've directed them there.

A sense of space: This is something that is missing and difficult to render on the web. But it's a type of feedback that may be essential. To work with it, we should think of other ways of showing related content, like indicating how it's related.

Adding consensus to user testing

Knowing more about how people find consensus on the web will help us understand how they have really assessed what we have designed. If we keep in mind they ways that feedback is shown on the web, then we'll be able to glean information from a user without having to directly ask (which may be fautly evidence).

Applying consensus to user testing is also a big job, but here are some more of my ideas:

Studying where a user hesitates during a sales funnel. If we see where a user really needs feedback, we'll know where to put it and what it needs to do.

Record related page clicks: Without having a user tell us what they think, we can record which pages a user 'traversed' to get where they wanted to go.

This is obviously the way to test labels and information architecture, but it can also tell us, depending on our methods of analysis, the user's idea of where they are in the context of the site, and how they think pages are related to each other.

Actions versus reporting

Exploring these ideas will help us better understand the user and their interactions. If we think about the different methods of assessment on the web, the same way we're always thinking about assessment and the world, we'll know how to make even more usable sites. And not just because somebody said so.