destraynor

Vilfredo Federic Damaso Pareto was born in 1848. During his notable career in the field of micro-economics, he observed that 80% of the income in 19th century Italy went to the richest 20% of the population. This became known as the Pareto prinicple, and later on as the more general “80/20 rule” As you can imagine, it’s a significant stretch to relate an observation about wealth distribution in Italy to the design of user interfaces, but it does happen. I have three main problems with Pareto being used to justify decisions…

1. The data rarely backs it up
   Even with tonnes of usage data, or Google Analytics data, it’s very tricky to find this mythical 20% that accounts for 80% of the usage. Also, if/when you do find it, you have to make sure that it reflects what the users want, not what they click. I’ve worked with clients who were adamant that their “About page” was their most popular, only to see all the users end up there whilst hunting for a non-existent contact number. Analytics only tells you the what, not the why.
2. Some vital features are rarely used
   From what I’ve seen, less than 1% of users will change their password arbitrarily. If you’re just going to focus on the 20%, you can annoy a lot of people. There are a lot of high priority features that don’t see much usage.
3. What does it really mean?
   If I told you that Paste, Save, Copy, Undo and Bold account for 32% of the commands used in Microsoft Word, would you really alter the UI based on that? Paste alone is responsible for 11% of the commands, so it should have a super big button, right? At least 25 times the size of word count? Nonsense.

I know we use Pareto’s principle as a little nugget to capture clients ears and make them think about focussing on what's important rather than expanding their product to become the sum of all desired features. That's understandable. But in this new age of Less Software(tm), lets not get carried away. You can’t design software based solely on quantitative research. Especially when that research comes from 19^th century Italian economics.

The search engine world has found itself in a weird position. Weird, yet strangely familiar to me. Cuil.com launched just 20 hours ago, and is already the whipping boy of web applications. There are about a hundred blog posts showing side by side screenshots of Google vs. Cuil.com, with all sorts of hilarious search results. Cuil brought this on themselves, but that's only part of the problem.

Whenever I’m designing an application in Omnigraffle, I use the YUI stencils which are really great. There is one whole stencil devoted to Advertising styles, leader boards, banners all the usual stuff. They follow all the advertising standards (125x125 etc), and for the most part they do the job, that is they allow you to indicate where the ads will appear. They lull you into a false sense of security however …

I could do a glowing a review with screenshots and all sorts of niceties, but I think the following elevator pitch is enough for most people.

ZYB solves the disconnect between your phone and the net. Your contacts are backed up automatically. Your phone calendar can now automatically update with your work calendar. ZYB works for all phones and all calendars in all countries on all networks. Give it a try. Zyb.com.

Using ZYB, I backed up my phonebook, and synchronised my phone calendar with my work calendar. All in about 3 minutes. I'm impressed. It's a shame they're going for a social network thing as well. I think this will be really useful, especially since I won't be going next or near the O₂ iPhone anytime soon. The " O₂ deal" may as well come with a tube of KY Jelly and and some rusty nails.

It’s been quite a while since I’ve posted a programming challenge, that’s mainly because I haven’t heard a good one in a long time. I need my friends to do more interviews! Also, I’ve received emails in the past along the lines of “I really like the puzzles, and learn a lot from them, but can you post one a first year could solve”. That’s understandable, most of the puzzles I’ve posted previously might be tricky with only a few months of programming experience. So this programming puzzle is my attempt at monolithic di-ornithicide (aka killing two birds with one stone)

The longest repeated word

Write a method that takes a simple string as input, and outputs its longest repeated substring. When there are two of equal length, output whichever is first lexicographically.

Examples

• abba should return ’a’
• abracadabra should return ’abra’
• Mississippi should return ’iss’

You can get your solution to me any of the following ways, email, blogpost, trackback, IM, phonecall, even twitter it if your solution is <180 characters

Pretty please, I implore you, don’t post it as a comment You'll end up wondering why my blog ate all the rogue characters(< > ; etc). I can’t be bothered digging through 5K lines of someone elses PHP to try and stop it.