Monday, June 8, 2015

Crowdsourcing Revisited

There is great potential in networked technology to bring together the ideas and work of millions of minds to accomplish substantial things – but to what degree has that potential been realized?   In over two decades, examples of successful mass-collaborative efforts are very few, which begs the question of whether the potential is real, or merely has not been realized.

Granted, the wave of enthusiasm about “crowdsourcing” has largely died down.  So it’s likely a good time to reconsider it from a more sober and less starry-eyed perspective – because there does seem to be a great deal of untapped potential, though not quite so much as the fanboys would have us believe.

With this in mind, I’ve read a couple of books on the subject and considered what the authors have to say:

  • Jeff Howe’s Crowdsourcing speaks directly to the phenomenon – the author attempts to be objective, but is still rather an enthusiast who goes on long tares about what should be possible (but hasn’t been accomplished) and trots out “success stories” of businesses and projects that have failed since the book’s publication.
  • Gustave LeBon’s The Crowd was written a few hundred years before the Internet was invented, but remains a valid exploration of crowd psychology.  And the behavior of crowds online, like that of people online, is very similar to their behavior in real life – albeit a bit worse given the ease of participation and the expectation of anonymity gives people the sense they can behave very badly without consequences, and they often do. 

Considering what Howe claims a crowd is capable of doing and what LeBon says about their habitual behaviors and cognitive capacity, it becomes clear that the crowd is good at a few things, but bad at most:

The crowd is very good at gathering

When it comes to collecting things, the crowd is a very good resource.  Expedia, YouTube, Pinterest, and similar sites that aggregate a huge amount of data benefit greatly from crowdsourcing, on the premise that many hands make light work – with the caveat that the work will not be particularly good.

The crowd is not good at creating

There is a significant difference between gathering and creating, in that creating requires effort and talent to produce something whereas gathering does not (it’s merely bringing together things others have created).   User-generated content is notoriously awful – and while there are members of the crowd who have the talent and the willingness to contribute it, they are few and far between.

Consider Sturgeon’s Law, which maintains that “90% of everything is crap.”   It’s more like 99%, and maybe 99.99% - but the Law of Large Numbers reminds us that if there are ten million contributions, the 0.01% of quality contributions means there will be a thousand good items – though finding them in the mass of inferior works is difficult.  This brings me to the next point …

The crowd is fairly good at moderating and organizing

Finding needles in a haystack is another instance in which many hands make light work.  Crowds are good at flagging inappropriate content or suggesting which items among a multitude are worth giving closer intention.   However, the reason a crowd is only fairly good at moderating is that opinions differ as to what constitutes quality – and the crowd is only capable of reflecting popular opinion, which may be inaccurate (see the later point about knowledge).

Sorting things into categories is rather more complicated than moderating.  Whereas moderation requires sorting them into two categories (good or bad), organizing often has an open categorization system, and the crowd must develop its own taxonomy – which again is a task that requires knowledge if the system of categorization is to be meaningful.

The crowd is very bad at anything that requires talent or knowledge

From the previous points about creating, moderating, and organizing, it becomes apparent that crowds are not good at anything requiring talent or knowledge.   Ask a crowd to paint a mural, and you get graffiti.   Ask a crowd for medical advice, and you get highly questionable and potentially dangerous folk remedies.   You’re better off commissioning one artist or consulting one physician for those things – and the same is true to some degree of an task that requires talent or knowledge.

The crowd is not very good at predicting

The financial markets are ample evidence that when people fall in line with a herd, they do not think very clearly about the future – they tend to be emotional and move en masse based on very superficial evidence.  Hence, there are bubbles and crashes.   They are also highly subject to groupthink – a person stating an opinion is prone to agree with the opinion stated by someone just before him, or express what he thinks will be acceptable to others.

This has been discovered in the field of marketing research: while it is easier to interview twelve people in a focus group, the results are less accurate than interviewing them individually and aggregating the results because the interaction causes them to normalize, and generally to the lowest common denominator.

The crowd is very bad at sustaining interest

Crowds are prone to get excited about doing something that is simple and quick, but they tend to lose enthusiasm and disperse very quickly – and this is the main reason they fail to accomplish much.   They follow fashions, which change rapidly, and have the attention span of a hyperactive teenager.   So if a task requires more than a few moments’ effort, or engagement over a long period of time, it will be left undone.

***

All things considered, crowdsourcing is an idea that should be considered – but before getting into the details of how the crowd can do something, it’s very important to consider whether the crowd is capable of doing it sufficiently well.   In many instances, it is not, and you’re better off hiring an expert (or assembling a team of experts) rather than throwing your project to the mob.


1 comment:

  1. I read the same book years ago and agree with your statement that the author is overly enthusiastic. I have been involved in a lot of crowdsourcing projects (my boss is sold on the idea in spite of our constant failures) and find that they require a lot of time and money and produce very poor results. There’s the old saying that a million monkeys banging random keys on a million typewriters will produce a perfect novel just by random chance. But the problem is you have to sift through a lot of garbage to find anything useful. And asking the monkeys to help identify the good material isn’t as good an idea as the author says it is.

    ReplyDelete