I have found all the time that psychology is not only important but vital for web design. I don’t think we should dismiss it just because it is difficult to measure or we can’t quantify it adequately. All this also bothers me, but there are methodologies in usability that can measure user reactions to a certain point, so it is not totally un-scientific and un-measureable. I think we should take psychology as a difficult-to-use-and-measure-not-to-mention-prone-to-error scientific tool.

Releting to the confidence hypothesis, I have observed that if users are sure that they will find their information at the end of a certain path, they will always follow that path, even if it is longer than other paths. The confidence hypothesis was right here. This one situation happens to be quite measurable.

In the case I related in my former post, users always followed a 4-click path with pages with some graphics and which left little screen space for seeing the information, and then had to click three times for each piece of info (about ten pieces), so 4+10*3=34 clicks.

The short path, required two clicks per piece and loaded very light pages, (or only one click and some scrolling, if they choose the long format), so 2+10*2=22 clicks.

The kick here is that the pieces of info were on exactly the same format, only that at the end of the “long” path they were shown inside a frame ocupying only 40% of the screen.

After showing both paths to 3 differents users, they all claimed that the longer path was “easier” and “shorter”,a dn that the injformation was easier to find and read. They also didn’t mind that they couldn’t see the long format, where they could compare directly the pieces of info. The rest of users simply refused to see the short path, and demanded that that we set up the long one, because it was more “logical”.

Had we gone on studying why users preferred one path to another, we would have probably found reasonable explanations to this phenomena, and even measured them, but we hadn’t got the time or resources.

The confidence hypothesis is based on the psychology of the user. The higher his confidence, the more he will persist in his search, hence the higher the probability that he will locate the information he seeks. This finding is not of particular scientific interest to the web designer. The confidence from Sofi makes the user undertake 10 search attempts where he would otherwise have undertaken 5. In fact, if you don’t tighten the axioms of the theory, one may argue that, in this case, Sofi actually encourages poor design. It seeks to hold the user longer on the path, rather than shorten his path, to the information. There would be a price to pay somewhere. As an analogy, if you pay for 10 consecutive tosses of the die and I pay for 5, you will have a higher probability of throwing a six. However, it will cost you more.

The shortest-path hypothesis is the one that gives Sofi its power. It has a solid intuitive and scientific basis. Intuitive, because it tells you what you yourself can work out, namely, good design is the one that aims to minimize the number of search-steps between the user and the information. Scientific, because it can actually be written down as a formula.

Imagine you are at step 0. The information you seek is at step 10. You may only make one step at a time. If you have 100% certainty at each step then you will get the information in 10 steps, the minimum number possible.

However, consider the other extreme, and suppose you have 0% certainty at every step. For example, you can only get from step 4 to step 5 by chance. If you get there at all, the next step you make would also be at random, with only a miniscule chance that it would take you to step 6, as required. This is the well-known Drunkard’s Walk problem. Starting from a given spot, a drunkard takes a random step. From his new destination, which is now the centre of a one-step circle around him, he takes another step at random. His movement is therefore analogous to the search with 0% certainty we’ve just described.

There is a mathematical formula that shows that, after N such random steps, the drunkard’s current distance from his starting point is, on average, less than the square root of N. To illustrate, if you make 25 steps, moving from one step to the next entirely at random, your end-position will on average be less than 5 steps from your starting point.

The morale is clear. The purpose of the Scent of Information has to be to reduce the randomness along the path to information. The increased certainty would then reduce the number of steps to the information.

For example, we had a list of activities and an area besides it clearly labelled as “Answers”.

But users simply didn’t expect to find the answers in the “Answers” area.

It turned out that they first entered the activity they wanted the answers for, then they attempted to find an answer list somewhere inside the activity but never outside its boundaries or back into the parent page.

This was because users were searching answers for only one of the activities, and somehow thought that an area labelled “answers” was too general. Entering the specific activity seemed a better idea to them.

This was solved by creating inside the activities a special area with direct links to the relevant pages inside the “Anwers” area. Users were convinced that they were still inside the activity itself and used “back” to go back each time to the activity in spite of every page in the “Answer” area having its own navigation menu and completely different looks and layout…….

We have never seen an user actually using the “Answer” area directly, but we keep because it is useful for a certain type of power-users who also use the same site. Those power-users need the answers of all activities at the same time. Those users always expect and demand the answers to be in the “Answers” area, and they would never enter an activity in order to find the answers for just one activity