The following post is a bit longer, but it really helps how you think about crowdsourced systems. It’s also very much unpolished, and I welcome feedback as to ideas or examples that I may not have considered.

What sort of motivation does one have to participate in a crowdsourced system? To do so means virtual anonymity, becoming a single name within a sea of many. It means a compromise of control, a sharing a managerial responsibility and credit that some may find disconcerting. Numerous crowdsourced projects have failed, because they did not present a compelling reason to participate. Yet, there have been a great number that I succeeded; what did they offer that the others did not? Crowdsourcing needs crowds, so I considered how you can get them.

If you build it..they won’t particularly come…

So, what’s the magic answer? Of varying levels of importance, there are basically eight hooks that make a crowdsourced idea work: academia, charity, money, fun, community participation, forced participation, self-benefit from the product, and interest in the content.

The first two motivations to consider are academic interest and charity: the goodwill factor. Some potential users will participate in a system not for any returns from the system, but simply for the sake of its success. Wikipedia is the great benefactor from academic interests: while it has developed into a mature system where time invested by a user is time returned in some way, its initial growth was different. With a wiki, any knowledge you enter will itself be useless encyclopedic knowledge to you, given that it is something you already know. Wikipedia grew because users looked forward and imaged how delightful the system would be if it suceeded. A volunteer encyclopedia is not a new concept—the OED was started by hundred of volunteers, attempting to improve upon the lackluster documentation of the English language. Academic motivation, then, is motivation based on an ideological foresight. A motivation of charity also is centered on the actual system, rather than particularly what it provides. An example of charity is participating in a well-polished product, out of appreciation for the quality put into it. Consider three social community sites: Virb, a design-centric re-imagination of Myspace, or Tumblr and Dopplr, two sites with a small purpose but a solid implementation. Working against goodwill is the fact that few users actually do feel inclined to support something on those grounds. Also, such motivation is limited in time, and without more direct returns, people will eventually give up on supporting a system.

The other reasons for crowds to support crowdsourcing systems are much more self-motivated. Money is the most direct of these, as it is for most of our society. However, there is a problem with paid crowdsourcing: the amount of people there is to pay. This invariably leads to low wages, simply because anything higher would not be viable for most projects . Money is a great motivator, but the extent varies with the amount. It is a straightforward concept; people are more likely to pick up a quarter from the street than a penny. If the pay is low, the impact of the monetary incentive will be low. Tying the project into a commercial system also introduces problems of valuating to the work. I can spend hours sharing ideas in a free forum, but if you offer me 5 cents for it then I’ll feel like I anything more than a few seconds would be too generous. Last is the most concerning: the idea of buying community and the problematic connotations that come with it. The Mechanical Turk is the most prominent platform for paid crowdsourcing. On the Turk, we have seen that it’s implementation has fallen short of it’s ideological concept of an “artificial artificial intelligence”. What it has overwhelmingly become is a mix of people trying to buy crowds for questionable reasons, such as Internet traffic and reviews, and people trying to take advantage of cheap labour. Where money makes a good incentive is when you have a lot of it, and as a last resort, where there are no other incentives for the work that needs to be done (e.g. transcribing audio/video or editing texts).

On quite the opposite end of the spectrum is using fun and boredom to attract support. The concept of fun is basic: make it fun! ‘Games with a Purpose’, headed by Luis von Ahn at Carnegie Mellon, is an exploration of this method. It includes a series of games, most notably the ESP Game and Peekaboom, that make semantic analysis into a game. ESP game, already covered by me, has two strangers try to guess what the other it thinking. The game has become so popular that, according to von Ahn’s talk at Google, they had to introduce limits to how many hours a single user can spend logged in. While on the user-end it’s an addictive game, on the other side, the system is left with a database of independently confirmed semantic tags for digital images. if the 9 billion annual human hours spent on gaming (suggested by von Ahn) are true, then there is definitely potential here. However, ‘fun’ is a hard concept to put in practice, and with so much competition online, it is likely that such games would very much fall back on the goodwill factor. Furthermore, to produce ‘fun’ within the constraints of your crowdsourced purpose may very well be the most difficult task of all of these. It requires cleverness in competing mindsets: those who have large-scale human-computation tasks are likely the ones without a finely-tuned sense of what is fun for the masses, just like those who understand low-entry gaming probably won’t be looking to solve blue-sky problems.

Appealing to the bored masses is not much different from using fun. Fun is certainly one of the ways to do so. However, I’d describe boredom-motivated crowdsourcing not just to be games, but also ephemeral toys. Even though they have a purpose, they inherit a status as a time-waster. Little thought is asked of the user, and few rules are imposed. A bored Internet user does not generally want to mentally exert themselves; rather, they want some pretty, something “cool”, or something new. Consider this click survey, where users simply click on the page, and their click results in relation to others are presented in a heat-map type graph. Very ephemeral, but interesting, producing some fascinating psychological results at the same time. Another example is Human Brain Cloud, which takes an impressively free-form, anything-goes approach and succeeds in it. The marvellous visual brain cloud, however, is as cool as it is useful, not to mention funny (e.g. I typed in ‘racist’ and was given ‘bigot’, ‘xenophobe’, ‘intolerant’ and ‘Mel Gibson’). Human Brain Cloud also has scoreboards and lets you define a simple identity. This strategy of adding goals and achievements is used often, and keeps many users coming back. Arbitrary achievements and badges aren’t even limited to games (e.g. GasBuddy’s status icons), but they introduce a game-like level of competition.

In the next incentive, participation, the crowd is the focus. For one, there is an inclination towards human interaction, even if it is with strangers. Nowhere is this more evident than in the popularity of massively-multiplayer video games, when you would assume (well, I would) that digitally beating up a stranger from Finland would be comparable to destroying an AI player. In the realm of crowdsourcing, communities working on the same thing create emotional bonds, whether of respect or hate. In a community such as Wikipedia’s, the propensity for respect drives the content and quality of the most active users (who power a considerable slice of the site). Returning to the ESP Game, von Ahn claims that one of its addicting qualities is that users feel a strong bond to strangers they agree with, while having a human to blame when the two don’t do too well. Social recommendation is also powerful online incentive: when “everybody’s doing it”. With blogging (and even more powerfully, the Facebook Newsfeed) revolutionizing how users share information, a very likely foil for other incentives is when a familiar participant pulls you into it.

In some cases, the motivation for a crowdsourced action can be attached to an unrelated action. In such a way, it effectively becomes forced, unto any user that wants the other action done. The premier example of this is reCaptcha, which reworks the concept of the Captcha, a skewed character image where the user writes the characters to prove that they are human. ReCaptcha appropriates this common process of proving one’s abstract cognition and applies it to the digitization of OCR-problematic words in scanned texts. To access the action that the user wants, such as online commenting or a service sign-up, they have to join the distributed crowd and do a few seconds of of work.

The last main incentive for users in their own self-benefit from the content created by the crowdsourced system. There are two ways that this can embody itself: direct and indirect. Direct benefit is when the content created by the effort is of immediate use and value to the individual participating. This will be usually related to initiatives that require sharing of knowledge or experiences. By adding your own facet of information to the project, it becomes framed on a greater scale. For example, you can submit the local price of milk, lettuce, or beer to WNYC’s Are You Being Gouged, knowing that by supporting the map, the ultimate product (a map showing the prices of these items throughout Manhattan) will be of use to you. In-direct self-benefit, meanwhile, can be best described as karma. A wiki is an example of this because, as mentioned earlier, what you contribute to a wiki is generally something that you don’t need returned to you. Rather, you may contribute knowing that, by supporting the system for others, the system will likely return the favour in a different area.

Those are my preliminary thought on motivation in crowdsourcing. Are there any I missed? Since beginning this, I’ve identified at least two which I hope to put more thought into: the sense of achievement in the action being done by the individual and of course, and interest in the content. Any others?

Advertisements