The Question Of The Day:
- What do you make of the term “collective forecasting”, coined (apparently) by our good doctor Hanson? Do you think it is appropriate? Does it convey the right message about the specificity of our information aggregation mechanisms that predict (our prediction markets and the non-trading mechanisms)? Or, at the contrary, does it confuse people into thinking about data and models (as in “weather forecasting“), which is obviously something we are *not* into?
- Should we let Robin Hanson use the term “collective forecasting”, or should we stop him cold right now before more damages are done? Let us know. Please, do comment. We need your input. Mike, David, Jason, Keith, Daniel, etc. Don’t hesitate to be libelous. Be harsh. Be disrespectful. Tell the truth.
Market-based forecasts?
Market predictions? ;-p
As I explained in the comments, there, the word “forecasting” refers to models, which is exactly the opposite that we have been selling. We have been selling information aggregation that predicts. We wanted to pose prediction markets as something different than forecasting, it seems to me.
http://www.midasoracle.org/2009/05/22/confusing-prediction-markets-non-trading-collective-intelligence-solutions/#comment-24070
As an accountant, there’s nothing wrong with the term “collective forecasting”. In fact, it might be the most appropriate term for what we are trying to accomplish (via prediction markets). I mean, it is collective, and it is forecasting. To me, it has no negative connotation that may be associated with weather “forecasts”.
I, like you, have an issue with the accuracy of prediction markets. I believe they can be accurate, but I think the “industry” has lost sight of this objective, which is crucial to its acceptance. I will have a lot more to say about this in new posts coming soon. I’ve been sidetracked by paying work and golf!
Paul
It’s fine with me too. Forecasting = Predicting. Although, I prefer “Collaborative forecasting” = Group prediction making. I don’t see the connection to weather either.
Panos,
“Market predictions” and “market-generated predictions” are OK to moi —since day one.
But that is not the question. Do you think that “collective *forecasting*” (where Wikipedia tells us that forecasting is less general than predicting, and applies more to predictive modeling) should define stuff that has nothing to do with predictive modeling but rather with information aggregation mechanisms that predict (like prediction market and the non-trading collective intelligence mechanisms that predict)?
http://en.wikipedia.org/wiki/Forecasting
“I don’t see the connection to weather either.”
Daniel, when you say “forecasting” to the mainstream people, they think of the kind of data crunching and modeling that is in used in weather forecasting, which is something most people understand very well.
So, when you say “forecasting”, they understand it thanks to their understanding of weather forecasting. Now, those people are not accustomed to information aggregation that predicts, which is a horse of another color. That is whey the use of the term “forecasting” for both things (predictive models and information aggregation that predicts) can be discussed.
“As an accountant, there’s nothing wrong with the term “collective forecastingâ€.”
Paul, I don’t care what accountants think. I want to know what people in the street understand if we ask they about “collecting forecasting”.
“To me, it has no negative connotation that may be associated with weather “forecastsâ€.”
Daniel, does the term “forecasting” have a connotation associated with predictive models? That is the damn question.
I have asked prof Gelman…
Hi Chris…
I gather you are attacking the issue from a *public* prediction market perspective, whereas I am looking at it from the EPM side. Collective forecasting is a reasonable term for EPMs. I should note that *all* prediction markets are “models”, though they are not deterministic.
If we need a separate term for public prediction markets, maybe “crowd crystal-balling for fun and profit” says it all (with smaller words)? (I’m kidding).
Paul,
I am following the field of prediction markets (public and private) since 2003. I have always heard the EPM guys saying that there are attacking the corporate planning problem from a completely different standpoint than forecasting, using information aggregation, powered by traders who are not necessarily experts.
That is why I wonder whether it is clever to put EPMs and the non-trading collective intelligence mechanisms under the “forecasting” tent. PMs and CI are very different (I would dare to day, “opposite”).
Actually, I believe that PMs are a component of CI (possibly appropriate for “cognition” problems). Everyone needs to be reminded that PMs are merely *tools* to be used in the decision-making *process*. I’m not sure that the EPM guys are looking at the planning problem from a “completely different standpoint *than* forecasting.” I think it is more appropriate to say they have a new tool that can help improve the accuracy of forecasts. Similarly, if other non-trading collective intelligence mechanisms can assist in generating accurate forecasts, what’s wrong with putting them in the same boat? They’re both tools in the process.
“Collective prediction” sounds a lot more awkward to my ear than “collective forecasting.” “Collaborative prediction” sounds just as bad. I don’t think most people see much a difference between “forecasting” and “prediction”, other than that “forecasting” is a word used a lot more often in business contexts. That seems to me a good reason to use “forecasting.”
“completely different standpoint *than* forecastingâ€
Traditional forecasting is done by an expert of a small team of experts. Collective intelligence that predicts is done self selected people, or a large group of non-selected people. They all act as information bees.
Yes, that is completely different. That is why traditional forecasting and prediction markets (or else) was not put in the same “boat”.
“what’s wrong with putting them in the same boat?”
Because of the definition of “forecasting”, which I said that I believe is about mainly predictive modeling.
Now, the big boat is “predicting”, of course. The whole polemique was started because I believe that the word “forecasting” is a sub of “predicting”.
Will the professional forecasters accept that “forecasting” encompasses the prediction markets and the other non-trading information aggregation mechanisms that predict?
“Forecasting” is different than “predicting”:
http://www.midasoracle.org/2009/05/25/predict-forecast-dictionary/
Great, 16 comments on this post alone, all that on Memorial Day.
Chris,
I like “collective forecasting” , the way it sounds. A lot of your criticism is redundant because we already have the word “collective” in there. Also, take a look at the definitions and you can see why “forecasting” is actually superior to “predictions”.
“Forecast”
1. To estimate or calculate in advance, especially to predict (weather conditions) by analysis of meteorological data. See synonyms at predict.
2. To serve as an advance indication of; foreshadow
Explanation 2 is right on the mark.
“Prediction”
1. The act of predicting.
2. Something foretold or predicted; a prophecy
Now, surely you don’t have to ask yourself where all the hype came from…