Whys, wherefores and whatnots

Order in disorder

andycrellin — Mon, 08 Jun 2009 10:16:26 +0000

If you are in the business of causing mayhem, this is where you can have your efforts evaluated. Expect to meet The Joker, The Penguin and The Riddler in the waiting room.

And apologies for throwing in two Engrishfunny.com posts in quick succession!

Seize the day – use it or lose it

andycrellin — Wed, 27 May 2009 14:31:21 +0000

I’ve just read an (as usual) excellent post by Dave Trott over at CST on the perils of saving something good for just the right moment. To paraphrase: If you have a good idea, use it now not tomorrow, because now is real while tomorrow may never happen.

In other words, imagine that you have a low value project running now and the possibility of a high value project coming up some time in the near future. Imagine, also, that you have just had a spark of inspiration that could be beneficial to either project, but can only be used once. You may be tempted to wait until the high value project comes along to try to gain the best possible return from your inspirational idea. This sounds sensible, but think about it another way: what if the future project never arrives? What if the future project turns out to be low value after all? What if the future project changes slightly so that your inspirational idea no longer fits? Then you get low or zero return on your idea. There are too many unknowns about the future, while there some guarantees about the present.

The present is here and real. You know that you’ll get some guaranteed return for your idea in the current low value project. And on top of that, if you came up with an inspirational idea once, you can do it again. If your idea for the present project is rejected then you still get a chance to apply it to the next project – multiple bites of the cherry.

Carpe diem quam minimum credula postero: “Seize the day and place no trust in tomorrow”. *

Or, more simply, “Use it or lose it”.

* Thanks to one of Dave Trott’s commenters for this full quote.

Pure Engrish

andycrellin — Thu, 21 May 2009 12:16:19 +0000

Engrishfunny.com usually elicits the odd smirk from me, but this one had me in stitches.

I can’t even imagine what “Leek Speculation Eggs” might be…

Google Epidemiology FTW

andycrellin — Wed, 29 Apr 2009 22:17:20 +0000

I’m a big fan of Google’s way of doing things. For example, the idea of spending one day a week on personal projects is a brilliant way to tap into the hidden skills and ideas of your staff. I’m also a big fan of data. Well, not data in it’s rawest sense, but the uses to which it can be put. So, really I’m a big fan of analytics. Google is one of the biggest repositories of behavioral data there is – simply because we spend so much of our time online and using Google to search.

Google Flu Trends

Google Flu Trends

One very beneficial graduate of Google’s labs is the Flu Trends product. This measures the prevalance of searches for flu related terms in the US and can estimate flu activity up to two weeks faster than traditional methods. And last week the Centres for Desease Control in the US asked Google to perform a similar analysis of data from Mexico to see if the spread of swine flu was reflected in the data before the media got hold of the story. And it was.

Google Flu Trends Mexico

As you can see from the above graph, there is an unseasonal rise in flu related searches from early April.

The power of this type of analysis is only just being realised, but the sheer amount of data and its range means that it is going to take some impressive analytical minds to design useful products. As this Wired article says, nobody was looking at the Mexico data when it would have been useful.

650mph Car Pancake

andycrellin — Mon, 13 Apr 2009 14:05:12 +0000

I don’t normally like simply pointing you to another blog entry with no narrative or analysis, but there isn’t really much else to say about this one. Definately one for the boys. The video’s about 4 1/2 minutes long but its worth hanging on for the slow-mo.

Mythbusters – what a job!

Reaching 50 million people

andycrellin — Sat, 11 Apr 2009 14:29:36 +0000

How do you show a span of 70,000 years on the same diagram as a span of 3 years? This is actually much more difficult that it might at first seem – especially if you have multiple items with very short spans that you want to show at the same time as a few items with very long spans.

I got thinking about this after reading a couple of factoids about the length of time it took for things like the iPod and Facebook to take off. I started thinking about the length of time it took for other technologies to reach 50,000,000 people, as this was the number used for the iPod and Facebook comparison. The image below is an attempt to show this information on a single diagram. I have included the rise of the human race, writing, radio, television, the Internet, the iPod and Facebook. I’d like to include more technologies to fill in the gaps in the thousands-of-years range, so I might update this diagram in the future.

Reaching 50 Million People

What was also interesting was the rate a which the reach of technology changed over time. By plotting the log of the age of the technology against the log of the length of time it took to reach 50 million people, I got the following chart.

Access to Technology

Although this isn’t really a formal scientific treatment, it looks like that for technologies that are accepted (i.e. continue to be popular after the initial hype), reaching 50 million people should now take less than a year – and this is only going to get quicker and quicker!

If you want to use either of these diagrams I am happy for you to do so, but please make sure you reference the source.

Keeping Design Simple

andycrellin — Tue, 07 Apr 2009 08:57:08 +0000

A long time ago I was a graphic designer. I was OK at it, and enjoyed it, but it was never going to be a career for me. One of the mantras I particularly enjoyed working to was “Keep It Simple”. I loved the opportunity to design a logo using a single colour, only one or two lines or shapes and some clever typography. Sometimes this was not only appropriate but the client loved it and everyone was happy!

On abduzeedo’s blog he/she (for I know not which) has trawled logopond for a collection of simple, single colour logos as part of a reference collection for a project. You can see the full collection here, and below are some of my favourites.

Brothers Grinn

Siah Design

Million Year Art

andycrellin — Sat, 04 Apr 2009 16:05:52 +0000

Sometimes, we puny humans are completely redundant when it comes to creating great art. Just let nature get on with it and jaw dropping, adjective exhausting, sense overloading pieces of art pop up all around us. Well, after a million years or so.

This site (http://www.earth.uk.net) is a collection of photographs of cross-sections of agates and other rocks and minerals, highlighting the spectacular formations often rendered on tiny scales. I can heartily recommend browsing though the images – they are also available to buy in large format and mounted.

My top 5 (find them on the site as they look much better on a black background), which I would be happy to hang in my house, are:

: 1

: 2

: 3

: 4

: 5

Google search to return related concepts

andycrellin — Fri, 27 Mar 2009 13:09:16 +0000

Google have recently made an interesting improvement to their search product. If you are searching for a fairly clear topic the search results will be returned along with a section with a list of possibly related topics.

In this Google Blog post they use the example of searching for Principles of Physics. Along with the usual results, other suggested searches are returned including “physics special relativity”, “big bang physics” and “physics classical mechanics”.

Although it is not clear in the article, I have to assume that the associated search topics are suggested algorithmically, meaning that Google will be building a database of concept sets that will be improved as people click on the suggested search topics.

The power of this, in the future, will be the ability to measure how “linked” certain principles or concepts are, rather than simple search terms or phrases. This might be incredibly useful for companies trying to understand how their online customers move from idea to idea or concept to concept, with the possibility of finding new areas where they can present their products.

However, well defined links between concepts will mostly be already known, so it will be the unusual links that will be valuable – and these links will take some time to surface. But surface they will, and to the benefit of those companies sharp enough to take advantage.

Who’s Asking?

andycrellin — Wed, 25 Mar 2009 18:25:32 +0000

I have been thinking about how the web (and search engines in particular) is being used to answer direct questions. For example, if you put “how do I hang a door” (rather than “door hang method”) into a search engine, how good are the results in answering the question? The English language phrase “how do I hang a door” is not optimally designed to help a search engine find the best result, whereas “door hang method” might be better. However, as humans, we are not naturally programmed to design optimal search queries. This is an enormous topic and there are dozens of books and research papers on the subject, so I thought I’d look at a specific part of the question: Who is asking when, where, how, which, who and why, and are there any interesting patterns and insights?

Notice that the English language query contains the word “how” whereas the search optimised one doesn’t. It looks like when people are using English language queries they would be more likely to contain one of the “w” question words.

So let’s start at the beginning: Global searches with “why” in them. (click on the image to enlarge)

Search volume for "why" - Global

This chart shows the relative number of searches made with Google for the years 2004 to 2008 that contained the word “why”. [light blue=2008, red=2007, orange=2006, green=2005, dark blue=2004]. The horizontal axis shows the months from the most recent year. These charts are from Google Insights for Search.

You can see that most of these years followed the same pattern with slight dips in the middle of the year and at Christmas. There is an unusual spike in November/December 2007 (possibly the subject of a future post, let me know if you want to know!). The regular dips I guessed must be to do with the usual Western school year – coinciding with the summer and Christmas holidays. I tested this by narrowing the analysis to UK searches only – I could then check the dips against actual academic year dates.

Search Volume for "Why" - UK

Now, this is more interesting. There are very clear dips at the following times:

Mid-February
Early-April
Late-May
Mid-July to early-September
Late-October
Christmas.

All of these dips correspond almost exactly to the dates of the UK School Holidays:

February half term
Easter holidays (this has a larger spread due to the shifting of the dates of Easter from year to year)
May half term
Summer holidays
October half term
Christmas

From this it is easy to hypothesise – from the UK study at least – that it is school students who are asking “why” during term time. It is also clear that more students are asking “why?” – the number of queries containing “why” is higher in 2008 than in 2004. This might suggest one of many things – that students are becoming savvier and discovering that they can get good answers online, that the population as a whole is more “internet enabled”, or that students are asking more questions online because it is easier. In fact there are multiple possible roots to this behaviour – enough for an entire book on the subject!

This analysis is very simple and doesn’t really have any surprises (unless you look at the countries where most of these searches are coming from – check it yourself, but I will probably write about this at some point). For now, I just wanted to show you how powerful free online search analysis tools are.

More next time.