3 ms·
Unclear how this is new, even Google, Amazon, etc. have either been doing this internally, offering it as a service, been susceptible to man-in-the-middle explo
by nxzero 11y ago
Unclear how this is new, even Google, Amazon, etc. have either been doing this internally, offering it as a service, been susceptible to man-in-the-middle exploits to mining real world data for training sets, released data, etc.
- awinter-py 11y agoSpot on. One recipe to become a tech acquisition target is to collection a 'new kind' of user data -- all big companies are hungry for this. This phenomenon is not at all new; data has been informing investment models forever and access to that data comes from having the right customers, and is closely hoarded once gotten. Some of the largest companies in the late middle ages were wool buyers -- they weren't permitted to trade internationally, but they used locally owned franchises and market knowledge to corner the market anyway. And many of the largest ag commodities futures traders in this century also own substantial farm acreage. Those capital one guys who were SEC'd for trading options on credit card receipts were leveraging customer activity. Point being -- you've always needed data to train a good model.