by Hubertus Hofkirchner -- Vienna, 08 Jun 2018
In the fast-moving 21st century, CEOs and other strategists need more powerful tools than just their gut sense to forecast the future and long-term trends. Without such foresight, all strategy development and innovation programs is built on shaky or even unsuitable foundations.
One relatively well-known such tool is the Delphi method which originated in the 1950’s at the Rand Corporation. A newer and still unfamiliar one are prediction markets, with their earliest industrial applications at Siemens AG in Vienna and Hewlett-Packard in California in the 1990’s, concurrent with the dawn of the internet age.
- Delphi is in essence a group of experts with different perspectives doing iterative rounds of questionnaires asking for their strategic forecasts and a commentary. A human facilitator aggregates the responses after each round and provides feedback to the participants. This process is repeated until a consensus has been formed.
- Prediction markets use a stock market mechanism where experts or consumers trade possible futures, try to hype “their” future stocks with fellow experts, and glean real-time feedback from market price changes. They do this until the market stabilises which means that a consensus on the probable futures has been formed.
Both methods aggregate individual opinions from diverse viewpoints, both can work remotely and asynchronously, there is no need for physical meetings or travel. Participants can remain anonymous with either method, nobody needs to worry much about taking a contrarian stance. Anonymity also reduces bias from celebrity or hierarchy deference.
So, are prediction markets only a fancy, shiny new high-tech tool, where good old paper and pencil from the 1950’s will do just fine?
Despite the commonalities listed above, there are also quite a few pertinent differences between the two methods:
- Participants in prediction markets are incentivised for accuracy, which motivates them to think harder and more attentively than Delphi which often produces poor results.
- Second-generation prediction markets like Prediki include an online debating feature, which facilitates a lively exchange of arguments and ideas. Delphi is turn-based which hinders interactivity, its commentary is unresponsive, often redundant and adds less value over what others know already.
- Prediction market feedback is authentic, derived from other participants’ reactions, and aggregated by neutral mechanisms. Delphi feedback needs a human facilitator whose own opinions and abilities will invariably influence the feedback provided to participants for the next rounds.
- With prediction markets, observers can watch the forecasts in real-time. Any misunderstandings can be recognised and clarified without delay. New information, such as competitors’ strategic moves or new product innovations, is reflected nearly instantly in the market forecasts.
- Recruiting for prediction markets is much simpler, they can distill experts or superforecasters from any readily available online crowd by simply running one or more qualification rounds. Delphi experts must be hand-picked with time-consuming recruiting processes.
- Like all markets, prediction markets integrate the valuable voice of contrarians in the dynamic price balance of bulls and bears, while Delphi has a known groupthink problem (Woudenberg 1991).
This has been an unashamedly one-sided argument for prediction markets, I know.
It is not just that I side with the technology because I am a prediction market evangelist. Now in 2018, I really cannot see any redeeming argument for Delphi anymore – other than where PCs, laptops or tablets are unavailable. Or where the participating experts very much dislike computers or electricity, like Chuck McGill or the Amish and such.
But hey, there is a comment section below if you want to strike a blow for Delphi.