by Tiago Peixoto
I am currently a PhD researcher at the European University Institute in Florence (EUI). My main research interests are related to identifying factors that contribute to successful policies aiming a... More about Tiago Peixoto
Serendipity for Online Deliberation?
I was going to reply to Paul's comment and this actually became a post. Having suggested the idea of inversed tag clouds as a means to enhance serendipity, Paul pointed out two cases in which serendipity can be desirable: 1) where you know that all the entries are of a certain standard and 2) where the most popular ones are likely for that reason to be less interesting.
In this respect, I would like to speculate to what extent serendipity can be desirable in large-scale deliberation processes, and how this serendipity can be induced online. To give it a fancy name, "induced redistributive serendipity".
To make my point simpler, let's think for instance of a cloud based on the number of views that an article / argument receives. In this case, the most viewed articles / arguments (or their respective tags) become more visible as more people view them, generating a snowball effect of intuitive mimesis.
This is more or less what happens in flawed models such as the ranking system of phase 1 of the "Open Government Initiative", where interventions at the initial stage can (and often do) produce huge variations in the outcome. In this case, for instance, the idea of an inversed cloud could neutralize these undesirable effects.
Now, consider an online deliberation where, at least in theory, all the arguments are important and should be objects of consideration. Here, a traditional tag cloud could have disastrous consequences for the quality of deliberation. Once more, an inversed cloud based on the number of times an argument is viewed (the least viewed become more visible) could be useful for the purpose of online deliberation.
This could be particularly applicable for large-scale online deliberation. Given that people rarely take the time to go over most of the arguments that are available, an inversed cloud could have a redistributive function. As the most read arguments become less visible and vice-versa, one would expect a better distribution of the number of views each argument receives. Finally, for the participants, this could lead them to come across information that they were not looking for in the first place.
In other words, my question is: could this "induced redistributive serendipity" be used in large-scale online deliberation?No account yet? Join now - it's quick and easy.
Comments
Once you measure view counts, you can expose not the absolute number of up or down votes but the relative degree to which participants may or may not agree with any particular statement or comment.
posted about 1 year ago
Tim, I think that exposing the relative degree to which participants may or may not agree with a statement is a step further. But it still introduces a bias in the process. In this case, I believe that the relative degree of agreement should only be visible to the user after he casts his vote or at the end of the deliberation process itself.
posted about 1 year ago
As to ranking systems, 2 more options to make them to work a bit more effectively:
a) give people a limited set of randomly selected ideas to vote for
b) let people vote in a limited set of pairs (one idea compared to the other)
Of course, in both cases, the number of votes needs to be significantly higher than the number of ideas. Still in that case, the (overal) relative degree of agreement with an idea should only be visible to the user after he casts his vote.
updated about 1 year ago, posted about 1 year ago
Here's a good example where the user can see how others have rated an idea only after he submits his own rating.
http://scenic.mysociety.org/
posted about 1 year ago
Answering the original question 'could this "induced redistributive serendipity" be used in large-scale online deliberation?', the answer is yes, technically it can. This is achieved by incrementing a counter on the argument table each time an unique user views the argument, so that when the (inversed) tag cloud is shown, the font can be resized accordingly without any extra processing/calculations.
I favour this approach, because it promotes distribution while still allowing people to deliberate on matters they're more interested in.
posted about 1 year ago
How interested would the average user be in an inversed tag cloud? If there are 100 ideas to look through and I have only a limited amount of time, surely it makes sense (from my point of view!) that I look at the ideas that are currently most popular and then vote them up or down? Of course, this distorts the process but to some extent that is desireable (people don't waste time on the worst suggestions) and it certainly helps the user! If I see an inversed tag cloud, why should I look at the largest tag when I know that in this case size means probable not worth looking at? I might just change my behaviour and go straight to the smallest tags!
Wouldn't another way be to have systems that encourage people to trawl for missing gems, i.e. you somehow flag up random ideas to a user when they log it and ask them to indicate whether or not it is a "missing gem" (i.e. an idea that is at the bottom of the pile but maybe should not). Then you could have a tag for "missing gems" and so you would have a mechanism for checking whether good ideas had been missed.
posted about 1 year ago