To get a market for lemons, the following characteristics are required:
- Nonuniform products or services with widely-varying quality.
- Expensive quality assessment.
- Poor information on relative quality, whether by distortions by sellers or lack of sophistication of buyers, or both.
You find this all over the place, with one notable example being tech recruiting (which appears multiple times in the HN/Algolia search above).
This is also a characteristic which leads to worsening product quality as formerly niche markets expand. Bicycles, audio equipment, and electronics are classic instances of these. A larger market is inherently less sophisticated, and more easily distracted by spurious or irrelevant characteristics of products.
Another tendency is for cargo-culting and fads to develop. That is, as products or services become more complex, a follow-the-herd mentality appears, where (apparently successful) influencers drive follow-on behaviour. Often, of course, the influencers and early-adopters themselves have a poor understanding or capability of distinguishing between high- and low-quality offerings. Given random selection, some will emerge as either successful or lucky over others.
There are some mitigations. In the case of used-car markets, for example, the emergence of vehicle history services (e.g., CarFax), reduces informational asymmetries. In the case of appliances, certification services (e.g., Underwriters Laboratories) and review organisations (e.g., Consumer Reports) aided greatly, as did uniform trade practices such as implied warrantee of fitness and generous return policies (both of which reduce buyers' risk).
As for your assessment of AI's future market, that seems highly probable to me, and would greatly dampen actual positive prospects within the field.
<https://en.wikipedia.org/wiki/The_Market_for_Lemons>
Discussed numerous times on HN: <https://hn.algolia.com/?q=market%20for%20lemons>
To get a market for lemons, the following characteristics are required:
- Nonuniform products or services with widely-varying quality.
- Expensive quality assessment.
- Poor information on relative quality, whether by distortions by sellers or lack of sophistication of buyers, or both.
You find this all over the place, with one notable example being tech recruiting (which appears multiple times in the HN/Algolia search above).
This is also a characteristic which leads to worsening product quality as formerly niche markets expand. Bicycles, audio equipment, and electronics are classic instances of these. A larger market is inherently less sophisticated, and more easily distracted by spurious or irrelevant characteristics of products.
Another tendency is for cargo-culting and fads to develop. That is, as products or services become more complex, a follow-the-herd mentality appears, where (apparently successful) influencers drive follow-on behaviour. Often, of course, the influencers and early-adopters themselves have a poor understanding or capability of distinguishing between high- and low-quality offerings. Given random selection, some will emerge as either successful or lucky over others.
There are some mitigations. In the case of used-car markets, for example, the emergence of vehicle history services (e.g., CarFax), reduces informational asymmetries. In the case of appliances, certification services (e.g., Underwriters Laboratories) and review organisations (e.g., Consumer Reports) aided greatly, as did uniform trade practices such as implied warrantee of fitness and generous return policies (both of which reduce buyers' risk).
As for your assessment of AI's future market, that seems highly probable to me, and would greatly dampen actual positive prospects within the field.