Notably, "pay for performance" (compensation porportional to measured value add) is much more common in online ads. I think this is likely a combination of better measurement tools and 3rd party ad platforms that can make strong promises about low-noise performance measures. This is because 3rd party performance measurers don't have the same incentive conflicts as agencies evaluating themselves (or internal ad providers measuring outsiders' performance!).
This is what I was thinking as I read the article. Interestingly, Facebook and Google are the 2 largest advertising firms and could easily offer advertising services (especially AI created ads). However, many firms that advertise on Facebook and Google are small businesses whose exec’s main economic value is advertising ability. In a way, the people who run Shopify stores act as the competing ad agencies on behalf of overseas manufacturers.
The firm has a fixed, large amount of risk involved in selling their product. Pay-for-performance either:
1. makes the ad agency take on all of that risk, which carries a high risk of sinking the ad agency if it is much smaller than the hiring firm and the product does poorly,
or,
2. makes the ad agency take on only a small fraction of the risk, say 10%. This is safer for the ad agency, but it means that if the ad agency could spend $1 more on their campaign to increase the firm's profit by $9, it's not in the ad agency's interest to do it. (Since the ad agency only gets $0.90 of that $9, for a net loss of $0.10).
Do you think your futarchy solution somehow gets around this? What fraction of the risk would the ad agency take on, what fraction would the original firm take on, and what fraction would other investors in the prediction market take on?
When you invest in a prediction market that pays off proportional to the profit of the firm, you're taking on some of the risk of that firm. And whoever you bought the share from is taking on a corresponding negative amount of the risk. So risk is still being transferred. The question is how much and what financial motives result from the transfer.
Notably, "pay for performance" (compensation porportional to measured value add) is much more common in online ads. I think this is likely a combination of better measurement tools and 3rd party ad platforms that can make strong promises about low-noise performance measures. This is because 3rd party performance measurers don't have the same incentive conflicts as agencies evaluating themselves (or internal ad providers measuring outsiders' performance!).
This is what I was thinking as I read the article. Interestingly, Facebook and Google are the 2 largest advertising firms and could easily offer advertising services (especially AI created ads). However, many firms that advertise on Facebook and Google are small businesses whose exec’s main economic value is advertising ability. In a way, the people who run Shopify stores act as the competing ad agencies on behalf of overseas manufacturers.
The firm has a fixed, large amount of risk involved in selling their product. Pay-for-performance either:
1. makes the ad agency take on all of that risk, which carries a high risk of sinking the ad agency if it is much smaller than the hiring firm and the product does poorly,
or,
2. makes the ad agency take on only a small fraction of the risk, say 10%. This is safer for the ad agency, but it means that if the ad agency could spend $1 more on their campaign to increase the firm's profit by $9, it's not in the ad agency's interest to do it. (Since the ad agency only gets $0.90 of that $9, for a net loss of $0.10).
Do you think your futarchy solution somehow gets around this? What fraction of the risk would the ad agency take on, what fraction would the original firm take on, and what fraction would other investors in the prediction market take on?
Yes, futarchy avoids this, by NOT using the risk transfer approach at all.
When you invest in a prediction market that pays off proportional to the profit of the firm, you're taking on some of the risk of that firm. And whoever you bought the share from is taking on a corresponding negative amount of the risk. So risk is still being transferred. The question is how much and what financial motives result from the transfer.
Only vaguely related, but curious if you might respond to this recent criticism of prediction markets (which also calls you out by name): https://reducibleerrors.com/prediction-markets/