Tuesday, August 14, 2007
Why voting in sessions may be a bad idea
Siddharta Govindaraj has an excellent take on why voting in sessions may be a bad idea: it encourages homogenisation of the audience, thereby undermining our efforts at promoting diversity. Siddhi says:
Lets take a simple example to see how this works. Say we have an event and 100 people turn up. 60 of them want to attend startup sessions. 40 want to attend photography sessions. There are ten speaking slots. Common sense dictates that having 6 startup sessions and 4 photography sessions is a “fair” distribution for the given audience.
But what happens when topics are put to vote? In every slot, the startup crowd can out-vote the photography crowd. Therefore when put to vote, the most likely outcome will be 10 startup sessions and no photography sessions.
Point taken. I think the voted sessions proposal can be safely put to rest now. (Update: Or maybe not. See comments below.)
Kingsly John — Aug 16, 2007 5:22:05 AM — # ↩
Not necessarily...
The scenario used by Siddhi isn't "typical" for what you were talking about.
Voting doesn't necessarily have to be one person one vote one topic.(There are many different <a href="en.wikipedia.org/wiki/Voting_system">voting systems</a> to chose from.)
When there are scores of topics and hundreds of voters, they can vote on all the topics or on as many talks as they can attend. And also rank them by preference.
This will see a wider spread of popular topics rising to the top.
The one person one vote principle is possibly of the biggest flaws in current day democracies. It was probably the easiest way to poll and tabulate results from a largish population a couple of hundred years ago. But in this age of electronic voting we need to look at more fairer systems.
In todays polarised world it is possible for someone to win an election in with less than an absolute majority of the votes polled. Someone can win an election with less than 25% of the popular vote and yet be hated by remaining 75% of the electorate. But there could be candidates with a much broader appeal who aren't the first choice of lot of people but could be a really popular second/third choice of more than 50%.
Kiran Jonnalagadda — Aug 16, 2007 9:39:34 AM — # ↩
Siddhi's scenario may be typical where voters come in collectives and essentially vote for more visibility for their collective. If they voted for individual preference and had multiple votes, however, it could be fairer. You're right.
Part of my intention with having voting pre-event is to encourage online interaction. Presenters seeking votes must engage their participants and form a little community instead of just expecting to deliver to a waiting audience. It's meant to facilitate the ongoing drive to turn Barcamp from a series of standalone events to a series of milestones in a continuum. If collectives break off to form their own online group or event, that's great news, for it means they've formed some sort of continuum. Whether voting will facilitate or hinder this remains to be seen.
Siddharta — Aug 16, 2007 3:53:36 PM — # ↩
The system I had mentioned is one vote per time slot. Since everyone will vote for each time slot, you get the problem mentioned.
The blog post was initially about reddit stories where you get one vote per story and how it has led to homogenisation. If you go to the reddit front page for example, almost all of the topics are about american politics. The linked article - onstartups.com/home/tabid/3339/bid/2086/Lamenting-The-Loss-of-Reddit.aspx - has more on that pattern.
There are many different ways of doing the voting. The central problem remains though.
The reason I chose startups and photography is because they are on the whole somewhat exclusive. A person coming for startups is not likely to have too much interest in photos, so even in a system like ranked voting, they will vote 1,2 and 3 from the list of startup topics. Plus ranked voting has the big problem of equating votes. If some talk has got 20% first choice votes and another talk got no first choice but 40% second choice votes, then which is more popular?
As an aside, this problem only occurs where there is a big variation in topic areas. An event for startups, where all topics are about startups will work out well with voting. This works because everyone is interested in each topic to an extent. So if the event is themed, eg: 'web 2.0' or 'social tech' and all topics are taken from the theme, then voting might work out.
Non-voted sessions have an obvious defect - what the speaker talks about could be something that no one is interested in. Thats why a voted track sounds like a good idea.
Kingsly John — Aug 17, 2007 2:38:22 AM — # ↩
I don't believe scheduling should be replaced by voting(while it can and should influence it)
Organizers can never satisfy all the people all the time in a multi-track event, so the objective so as long as every single talk is good it would add to the overall success of the event.
Assuming there are 4 tracks and 40 possible slots and someone can only attend 10 talks at most. We only let people rank their top 5(or 7) picks. So assume we have enough people voting there shouldn't be too much of a conflict.
As to the question of whether 20% first choice or 40% second choice it wouldn't really matter except in case of a tie since in this kind of scenario we will have multiple winners. Tie breaking can be a run-off vote or a flip of a coin.
Once the most popular talks are decided, the organizers can get figure out the best way to schedule the talks. And there'll be fewer complaints of clashes since organizers can talk into account the overlap of interest from the voting results.