Chapter 8: What The Crowd Thinks
The author suggests the immensely popular reality TV show "American Idol" as an excellent example of crowdsourcing. The show is basically a singing competition in which singers perform and the viewers vote to decide who's best. (EN: I thought that there was a judging panel that did that, and that the voting was inconsequential to the show - but perhaps the format changed?)
The use of "voting" has become quite popular across a range of industries. Polls are popular online, and companies have promotional events in which customers vote on superficial product decisions (such as what flavor of snack chips to market, what should be the color of a candy, etc.). The idea is quite natural - as most firms would use market research to gauge customer opinion before making such decisions. But instead of polling in secret, it's done in the open.
(EN: However, I'm not under the impression that it has ever done much good. I'm not aware of an instance in which the crowd's decision resulted in commercial success. That is, the public picks a flavor, the manufacturer puts it out for a little while, but within a short amount of time it disappears from the market. My sense is that voting and buying are two different things, and people vote for things they aren't actually interested in having.)
Votes can also take the form of "views" and "likes" to sort the content of a site in a way that is more appealing to audiences. The "most popular" items in any collection get immediate interest from others, and users often sort products to move the "highest rated" ones to the top of the list. And again, it's not a new practice: retailers have long advertised and promoted the merchandise that already sells the most.
Even the Google search engine prioritizes the results according to what has been most popular in the past. Ranking according to people's actual behavior (clicking a link) is an effective way to get around those who attempt to game the system, using tricks to get their site listed on the first page of search results even though it's not what anyone wants to see.
Reference to Sturgeon's Law, which states that "90% of everything is bad." The statement was made in reference to science fiction novels, but is a handy axiom for almost everything. Only 10% of it is at all worthwhile, and perhaps the top 1% is "downright good." And in a medium where millions of users post billions of things, filtering out the trash is an extremely valuable service.
While he's handing out made-up statistics, he also mentions the 1/9/90 rule that suggests that for every hundred people, one person will create something, nine will comment on what he created, and the other ninety will passively consume what others created. The relevance here is that the 9% who comment or vote often determine what the other 90% will pay attention to.
Predicting Buying Behavior
He returns to the example of Threadless, the T-shirt retailer that allows customers to submit and vote on designs. It's a very fickle market, mostly young teens, and many companies put out products that they hope will be adopted, but which fail. Yet Threadless has sold out every design it has ever produced.
However, this is not quite as foolproof as it may sound. The firm doesn't merely mass-produce every shirt that is voted a winner, it also gauges the popularity of a shirt according to other factors, such as the number of users who voted and the margin by which it beat alternatives. Also, it is not a binary vote, but a ranking of zero to five. This gives the firm several statistics to compare against a regression analysis of past winners' sales performance to arrive at an accurate estimate, and then produces a bit less than predicted.
Traditional market research used to determine consumer tastes and preferences is not as comprehensive - firms cannot ask every potential buyer their opinion of every potential product. It is also heavily skewed by selective listening - executives have the sense that they know their market, and listen to research findings that agree with their preconceptions (and disregard research that does not).
He flips back to "American Idol" - behind the scenes, the show is a front for a music label and a talent agency. Part of the agreement that contestants sign gives these firms exclusive control over their recording and performance careers. The companies make far more off of the contestants than they give out in prize money, and the show serves as a research vehicle to determine what will sell. And they sell hundreds of millions of albums by their contestants. "American Idol isn't a TV show - it's the largest focus group in history." Every member of the audience is a potential customer.
He switches to the topic of fan-produced advertising, in which companies hold contests that allow people to create their own commercial - mentioning that there were seventeen such contests on YouTube at the time of writing the book. The company gets free creative work (Doritos ran such a commercial as a Superbowl ad), consumers voting helps select the one that is most appealing.
He admits to doing some modified crowdsourcing, inviting the crowd to submit designs for the dust cover of the UK edition of his book. There were 300 entries and 20,000 votes - but in the end the firm flaked on going with the public's choice but had its editors select a cover to use from the top 20 entries.
Unconscious Contribution
He returns to the Google search engine's algorithm that leverages the crowd in an indirect way: sites are ranked against queries according to the words that are used to link to those sites in other independent sites, and refined by the number of people who click a given link after searching for a given phrase. People who use the search engine aren't even aware they are participating as voters in a popularity contest.
There is some historical reference to "citation analysis," which is often used in academic circles: people who write scholarly articles would present the number of other articles that cite them as evidence of their credentials. To be cited was to be endorsed by another scholar. In essence, Google was picking up on this practice, considering a link to be the equivalent of a citation.
This may seem commonplace today, but consider the early search engines, which simply used page content to classify and rank a page - which led to an ongoing battle of "spamdexing" as marketers would cram keywords into their pages in order to get attention from people who were searching for other things. By avoiding this problem (along with stripping the site of omnipresent annoying advertising), Google took the lead in the search engine wars, leaving behind competitors that are now virtually extinct.
He mentions a few early attempts at taming USENET, the internet's first bulletin-board system that was highly useful until it became clogged with trolls and advertisers. Both Xerox PARC and MIT came up with systems that helped sort through the articles by allowing readers to like/dislike articles so that other users could more easily find worthwhile content.
Suggestive selling is another method of leveraging the crowd - to suggest items based on ones that have been selected (based on what other shoppers purchase as a set) or based on the purchasing habits of similar users. The data is a byproduct of normal shopping activity. (EN: The author doesn't cite its success, and most of the references I've seen to it online are a bit comical.)
Another brief mention is made of eBay's rating system, in which buyers and sellers rate one another based on their transactions to help reduce fraud or at least increase consumer confidence in unknown vendors.
He describes the practice of tagging (users suggesting what an article/photo/film/etc. is about) and "folksonomies" that attempt to create categorization systems based on the way in which users categorize things in their own collections. (EN: This was very faddish but I am unaware if anything ever came of it. The author does little more than mention it ad describe the intended function.)
Case Study: Digg
The author shares some basic information about Digg, a site that enables people to publish links to things they found interesting. The site was just a database that offered no content of its own, nor any staff of editors or curators, but it drew millions of users.
A link on Digg could send a deluge of traffic to an otherwise unknown site. One specific benefactor suggested around 50,000 unique visitors per day during the week it was featured on the front page of Digg.
One notable incident occurred when a site that disclosed the code to override copy protection for DVDs was featured on the site. After receiving a cease-and-desist from the technology group, Digg dropped the link from its home page - and then a link to another site was posted. The incident gained publicity, and the top five pages of the site were overrun with links to different sites that disclosed the code. Ultimately, the patent holders had no case against Digg, as it was simply a channel and not the publisher of the articles that disclosed the code, not could the owners of Digg possibly comply with demands to remove the links that were constantly being re-added to different sources.
Under the old regime, silencing news outlets was simple - there were very few of them, and all were skittish. But with the Internet, there is no way to silence the crowd, short of pulling the plug on an entire site - but even then, the practice would continue on other social bookmarking sites.
Mob-ocricy
The dark side of the crowd is, naturally, that it is a crowd. It tends to reduce human behavior to the lowest common denominator, and often quite lower because the anonymity of participants in a crowd reduces their inhibitions and concern for the impact of their actions on personal reputation. In essence, it brings out the worst in people.
Consider the amount of content on YouTube that is very low-brow, catering to mean-spirited and childish humor, and the "most popular" items are often violent or sexual in nature. There is the concern that the crowd will usher in "an age of mass mediocrity" in which the vulgar tastes of the swinish multitude will become infectious.
Given these problems, the author still views it as the lesser of two evils, considering that there has been a long-standing oligarchy in business and the media. Economic and cultural activity has long been controlled by very few, who pander to the lowest tastes - and lower still for the contempt they have of the crowd.
While the crowd allows many to behave worse than they normally would, it allows others to behave better. There are diamonds in the rough, individuals constrained and rejected by the oligarchy, that now have freedom of expression. They may be hidden in the masses, but they do exist.