Crowds are not always wise

“The wisdom of crowds” has been a popular phrase since James Surowiecki’s book was published in 2004. I have heard it used in conversation a few times, usually to stress a point that crowds are better than experts in just about every scenario. That isn’t entirely true and last night’s TV provided two perfect examples…

The book makes a very clear point about crowd behaviour:

A group of people… is far more likely to come up with a good decision if the people in the group are independent of each other.

This is no small challenge – groups are usually sustained by the relationships between members of the group. Those relationships (and the hierarchies they support) can make independence non-existent. You can take the quote further – any size of crowd is only likely to come up with a good decision if all bias and influence are removed. And that isn’t just a big challenge, in this networked world of ours it can be an impossibility.

Example #1: The UK record of the year

This show was about a public ‘phone vote for the record of the year. There were 5 contenders in the final running. At least three of them (I think) had written their songs (Black-eyed Peas, James Blunt, Daniel Powter), not sure about the 4th (McFly), but at least their song was original. Did any of these 4 win? Did they hecks. How can you when you’re up against West Life doing yet another cover of a smoochy ballad? West Life have a huge fan base, who probably spent half the night hitting ‘redial’ on their telephones. The crowd is neither independent or unbiased. Did they honestly select the song because they believed it to be musically superior to the other 4 contenders? I think any one of the five songs could have been attributed to West Life and they would still have won. So much for the title of the show…

Example #2: The X-Factor

This show also relied on a public ‘phone vote. Same outcome but a completely different situation. This time you had 4 contenders and the one with the lowest vote leaves the competition. But these contenders are all unknowns meaning there is no loyal fan base to influence the vote. How could it not be fair? All 4 contenders are talented but there were two who performed flawlessly and received high praise from the judges. Then there were two who didn’t perform so well and were criticised by some of the judges… and so the bias appears. Any show involving judges and audience votes is prone to becoming the revenge of the crowd. Everyone loves supporting the underdog.

The part of the show I saw started with the presenter saying ‘just 24 votes separate the bottom two acts’. That comment likely influenced the remaining votes – the focus was placed on the acts likely to be in the bottom two. And what happened? The two you would expect to have been at the bottom filled the top 2 places. The two who performed the best ended up at the bottom. And the one who everyone thought was a certainty to reach the final, “best singer this country has ever produced” according to the judges, could not possibly be in need of another 24 votes… Yup, you guessed it – she was shown the exit. The best ended up last because everyone assumed she had enough votes… crowds aren’t always wise.

(In both cases, the studio audience started to boo when the results were announced – that’s good ol’ hindsight for you, when wisdom has a habit of sticking its tongue out.)

The Wisdom of the Crowds‘ is a great book and well worth reading but as quotes go, it’s as misused as ‘Good to Great‘ (also a very good book)

[Should add a disclaimer: I didn’t actually see either show in its entirety, just saw enough to witness the wisdom of hindsight in action.]

Why is KM so difficult?

It’s a popular question and there are many reasons but, after 10 years of talking about it, I still find the following 5 to be the most common:

  1. Organisations do not sufficiently recognise they are already doing it
  2. Knowledge is treated the same as information
  3. Information technology is often regarded as a substitute for social interaction
  4. Most knowledge management solutions look like traditional techniques
  5. First understand what you want to achieve…

1. Organisations do not sufficiently recognise they are already doing it

Too often a KM system is designed without due consideration for the knowledge sharing already taking place. Informal social networks are rarely documented and do not resemble formal organization charts, but their existence can not be denied and they are where the majority of knowledge typically flows. The challenge with social networks is that inner circles start to develop. We tend to go back to the same people we know and trust for advice and this can lead us to exclude people we don’t know who may have better knowledge to share. In other words, the flow of knowledge does not necessarily map to the quality of knowledge.

Identifying social networks and profiling skills can make it easier to find the right people to consult regardless of their inner-circle membership, helping improve the quantity and quality of knowledge flowing across the organization. And never forget, we are still more likely to ask a person than a system for the answer to a question (3 times more likely, according to Gartner; five times more likely, according to Tom Allen at MIT).

2. Knowledge is treated the same as information

I risk sounding like a broken record on this subject (I do tend to bleat on about it), but I first saw these statistics back at the beginning of the 1990s and they haven’t changed much in over a decade, despite (or, perhaps, because of) improvements in information technology. Far too much time, resource and money is dedicated to the pursuit of managing accessible information (the shared data in the chart, approx 10% of corporate knowledge), often at the expense of simpler and more effective solutions that focus on helping people find each other and work together.

According to Gartner, at their European Symposium in May 2005, a good sales system can improve revenue by 6%. An experienced sales person can outsell a novice by 40%. Do the maths – which system should you invest in? The master-apprentice model is still one of the most effective methods for sharing knowledge.

3. Information technology is often regarded as a substitute for social interaction

Do the following phrases sound familiar?

  • The “e-lance” economy – individuals will work as freelancers rather than as members of firms
  • Telecommuting will replace traditional forms of commuting
  • The “Paperless Office”

Yup, all popular ideas in the 1990s that we are still waiting to see take off. Why haven’t they succeeded? The first two fail to allow for a crucial element of human nature – we are innately social animals. We like to share experiences, we like to gossip, we like the feeling of ‘belonging’ to a group. It’s a rare person who, when asked for advice, doesn’t offer it willingly. And that social interaction is the most powerful vehicle for learning – for generating and using knowledge. As for the third phrase: there are clear benefits to be realised by converting many paper-based processes to electronic ones but we are unlikely to completely eliminate paper. Why? Maybe, for some uses, we have already invented the optimum format – paper.

4. Most knowledge management solutions look like traditional techniques

A paper-based file is not the same as an electronic record. A meeting is not the same as a virtual meeting. Too many KM systems have tried to take a physical form, activity, whatever and simply replicate it online assuming exactly the same processes will continue to apply. The physical and virtual worlds may exhibit similarities, but they are not identical. To use technology to manage knowledge (and information) often means developing new tools and eliminating old ways of working. If you are not prepared to change the process then you should reconsider whether or not to change the tool…

A great example is provided in the book ‘The Myth of the Paperless Office’. London Air Traffic Control Centre (LATCC) needed to replace their traditional paper flight strips with a digital process in order to maintain more accurate and up-to-date data about flight movements. However initial attempts failed because the digital process was unable to replicate all of the physical actions and team work built around the paper strips. To eliminate the flight strips meant reducing the complexity of the work. At the time, airplanes were able to come in from all directions. Today, they follow a set flight path in strict sequential order – a pattern that had already been adopted by other air traffic control centres. The process had to change.

5. First understand what you want to achieve…

And finally, too many attempts at KM still fall at the first hurdle, as identified by McKinsey in their 2001 global survey on knowledge management:

“Without knowing how to apply knowledge, there is little point in worrying about cultivation and distribution”

I continue to be amazed by how often people struggle to articulate the reasons why they are investing in KM systems. Fluffy descriptions such as “to make it easier to find information” or “to help people collaborate” are usually given. ‘So what?’ is my usual response. What’s the point in being able to find information if you don’t know what to do with it. To succeed, you need to be able to define how the KM system will benefit the business? I look for responses such as:

  • “We need to reduce the average time taken to repair ‘within warranty’ faults on our products – our best engineers fix stuff 40% faster than inexperienced ones”
  • “We want to increase our customer satisfaction ratings by 15%, existing customers purchase 18% more than new ones”
  • “We need to improve R&D and get better products (20% fewer ‘within warranty’ faults) to market quicker (current process takes 387 days, we want it down to 360 days”)

Well defined reasons help ensure that investment in KM is focused on what the business needs and provide metrics that can help measure the return on investment once the system is in use.


Related blog entries:

[Update: Jack Vinson has located the URL for ‘Why is Knowledge Management So Difficult’, now included above]