Make an easy-to-use tool for companies to set up internal prediction markets or forecasting tournaments, so that even medium-sized companies can freely experiment with in the technology (rather than just Google/Microsoft/Facebook/etc), and it’s easy for employees to use so you get plenty of engagement. Thus addressing what Nuno considers the #1 most important challenge to prediction markets in a corporate setting, “maintaining a prediction market given current technologies”. Individual companies shouldn’t have to do all the work themselves and reinvent the wheel; they should be able to buy prediction market tech as a service!
Maybe some high-quality tutorials, UX design, and demo examples to help employees understand what prediction markets are and help companies understand how to use them (how to write good questions, what business functions might benefit from getting predictions, etc). Thus attempting to address challenge #2, “Writing questions that are both informative for decision-making and attractive to traders.”
With a bit of experimentation, I feel like it should be simple enough to tune the sweet spot of challenge #3, “attracting enough predictors to get accurate forecasts while also preventing the market from taking up too much of the employees’ time and attention”. Should the market just be run on fake internet points like Metaculus? Fake internet points plus prizes for top predictors? Stock bonuses based on performance? Varying amounts of real money? Surely it wouldn’t be too hard to find the right knobs to turn so that companies can incentivize their employees to participate the right amount on average. This is another benefit of offering a service to many companies—experiment once and then offer that expertise to each customer, rather than forcing each new internal prediction market initiative to jump in blind and risk wasting a bunch of employee time if they mess up the rewards structure.
An external tool means companies get to use advanced features (like conditional prediction markets) right away, but (like Gitlab and other software-developer tools) they also have lots of control to self-host and customize the software if they want for their specific needs. (This is different from most prediction market software today where, at best, you can create private markets for your organization and that’s it; no customizability or ownership.)
Instead of a weird standalone experiment, offer integrations with other sales/marketing tools (like salesforce business software) so that the prediction markets can be seamlessly integrated with sales targets/quotas, advertising and A/B testing analysis, financial models, etc. Do the same with software developer tools like git, issue tracking, kanban boards, schedule tools… all the usual Jira / Asana stuff. That way, the act of predicting how long a product will take to come to market can be directly integrated with schedule & issue tracking tools to spot problems and reallocate resources. Prediction markets shouldn’t just be standing weirdly off to one side of everything a company does, but should become another tool in the integrated management/productivity toolkit of philosophies like “scrum”, “agile”, etc. I think this would greatly help companies find a use case for prediction markets and other forms of forecasting in their existing workflows.
As for a roadmap of who the early-adopters are going to be—I think your first market is probably going to be “medium-sized tech companies of the sort that regularly attempt to use prediction markets (and maybe have a CEO who happens to be excited about the idea), but smaller than Google/Microsoft/Facebook and thus would benefit from an external pre-packaged solution”. You could probably get some EA funding as initial seed money, and some of the larger EA/rationalist orgs might be willing to help out and experiment with the system internally, in order to get a better sense of which use cases you should focus on. (For-profits like Wave might be especially valuable to get feedback from—are prediction markets more useful for comparing marketing strategies and maximizing sales, or to predict software dev timelines, or for high-level strategy questions like “which direction should we take the company” / “will this product flop”?) Then build your integrations towards whichever application (sales vs dev vs strategy) seems to be getting the most traction, and roll out the product to more normal tech companies and eventually non-sillicon-valley non-tech companies.
Alternatively (caveat: this is a weird idea), maybe early on you could pitch prediction markets as a good way for especially loose/decentralized teams to get agreement on some debated issue. Like cryptocurrency projects looking for consensus on adopting a change to their protocol (easy to raise a lot of money if your project has a light coat of crypto-paint, perhaps!), or as a way for two companies considering a partnership to jointly and fairly asses the value that each brings to the table (seems to go against every rule of negotiation but might just be crazy enough to work??), or as a tool that helps build agreement on big questions of company direction and get employees more mission-aligned around key initiatives at otherwise large, dispersed, slow-moving organizations like General Motors or Boeing.
This is such a great idea, I really hope someone does it! My company is small (~80 person startup), but might be an example of a potential customer type—we’re an infectious disease intelligence company, so we’re literally trying to predict epidemics/disease events/operational disruption and an internal PM could be a useful way to organize our ongoing analysis and judgment.
Generally anyone working in the risk/critical event/threat intelligence/OSINT space might have a similar use case around collecting and coordinating their internal experts’ assessments.
Also, this may be taking the Gitlab comparison more literally than you intended but the ‘/estimate’ quick action has always seemed like a juicy UX for collecting predictions!
The comment I linked (laying out my “Gitlab of prediction markets” idea) is replying to a very high-quality post by Nuno Sempere and pals which examines exactly these questions of what stops corporations from enthusiastically adopting prediction markets / forecasting. My “gitlab” idea is an attempt to address some of the potential bottlenecks to wider adoption:
Make an easy-to-use tool for companies to set up internal prediction markets or forecasting tournaments, so that even medium-sized companies can freely experiment with in the technology (rather than just Google/Microsoft/Facebook/etc), and it’s easy for employees to use so you get plenty of engagement. Thus addressing what Nuno considers the #1 most important challenge to prediction markets in a corporate setting, “maintaining a prediction market given current technologies”. Individual companies shouldn’t have to do all the work themselves and reinvent the wheel; they should be able to buy prediction market tech as a service!
Maybe some high-quality tutorials, UX design, and demo examples to help employees understand what prediction markets are and help companies understand how to use them (how to write good questions, what business functions might benefit from getting predictions, etc). Thus attempting to address challenge #2, “Writing questions that are both informative for decision-making and attractive to traders.”
With a bit of experimentation, I feel like it should be simple enough to tune the sweet spot of challenge #3, “attracting enough predictors to get accurate forecasts while also preventing the market from taking up too much of the employees’ time and attention”. Should the market just be run on fake internet points like Metaculus? Fake internet points plus prizes for top predictors? Stock bonuses based on performance? Varying amounts of real money? Surely it wouldn’t be too hard to find the right knobs to turn so that companies can incentivize their employees to participate the right amount on average. This is another benefit of offering a service to many companies—experiment once and then offer that expertise to each customer, rather than forcing each new internal prediction market initiative to jump in blind and risk wasting a bunch of employee time if they mess up the rewards structure.
An external tool means companies get to use advanced features (like conditional prediction markets) right away, but (like Gitlab and other software-developer tools) they also have lots of control to self-host and customize the software if they want for their specific needs. (This is different from most prediction market software today where, at best, you can create private markets for your organization and that’s it; no customizability or ownership.)
Instead of a weird standalone experiment, offer integrations with other sales/marketing tools (like salesforce business software) so that the prediction markets can be seamlessly integrated with sales targets/quotas, advertising and A/B testing analysis, financial models, etc. Do the same with software developer tools like git, issue tracking, kanban boards, schedule tools… all the usual Jira / Asana stuff. That way, the act of predicting how long a product will take to come to market can be directly integrated with schedule & issue tracking tools to spot problems and reallocate resources. Prediction markets shouldn’t just be standing weirdly off to one side of everything a company does, but should become another tool in the integrated management/productivity toolkit of philosophies like “scrum”, “agile”, etc. I think this would greatly help companies find a use case for prediction markets and other forms of forecasting in their existing workflows.
As for a roadmap of who the early-adopters are going to be—I think your first market is probably going to be “medium-sized tech companies of the sort that regularly attempt to use prediction markets (and maybe have a CEO who happens to be excited about the idea), but smaller than Google/Microsoft/Facebook and thus would benefit from an external pre-packaged solution”. You could probably get some EA funding as initial seed money, and some of the larger EA/rationalist orgs might be willing to help out and experiment with the system internally, in order to get a better sense of which use cases you should focus on. (For-profits like Wave might be especially valuable to get feedback from—are prediction markets more useful for comparing marketing strategies and maximizing sales, or to predict software dev timelines, or for high-level strategy questions like “which direction should we take the company” / “will this product flop”?) Then build your integrations towards whichever application (sales vs dev vs strategy) seems to be getting the most traction, and roll out the product to more normal tech companies and eventually non-sillicon-valley non-tech companies.
Alternatively (caveat: this is a weird idea), maybe early on you could pitch prediction markets as a good way for especially loose/decentralized teams to get agreement on some debated issue. Like cryptocurrency projects looking for consensus on adopting a change to their protocol (easy to raise a lot of money if your project has a light coat of crypto-paint, perhaps!), or as a way for two companies considering a partnership to jointly and fairly asses the value that each brings to the table (seems to go against every rule of negotiation but might just be crazy enough to work??), or as a tool that helps build agreement on big questions of company direction and get employees more mission-aligned around key initiatives at otherwise large, dispersed, slow-moving organizations like General Motors or Boeing.
This is such a great idea, I really hope someone does it! My company is small (~80 person startup), but might be an example of a potential customer type—we’re an infectious disease intelligence company, so we’re literally trying to predict epidemics/disease events/operational disruption and an internal PM could be a useful way to organize our ongoing analysis and judgment.
Generally anyone working in the risk/critical event/threat intelligence/OSINT space might have a similar use case around collecting and coordinating their internal experts’ assessments.
Also, this may be taking the Gitlab comparison more literally than you intended but the ‘/estimate’ quick action has always seemed like a juicy UX for collecting predictions!
Hey!
[I somehow missed this comment from 6 months ago!]
Are you still interested in something like this?
I am! Would love to discuss, will DM you!