So, think about the profiles currently used by ad systems and recommenders.
- The main kind of data is observational. Like, they watch your clicks and purchases.
- Sometimes these systems add in other information. For instance, Google and Amazon know about our goals, based on what we search for. (In advertising world, they use the word “intent” to mean goals.)
- And there are ideas of other things to add. These companies are considering putting emotions or emotional responses in. So: Facebook would track what videos make you smile or react with horror.
These different types of data are good for different things.
- Info about emotions or emotional response could keep you scrolling. Or help them recommend content that gets you to smile or something like that.
- Goals help for recommending funnels and tubes.
This data isn’t good for recommending spaces. In fact, it backfires. Say you have a source of meaning like community care. What would a goal-based recommender do?
It might try to transform it into a goal—it could tell you that, if your teeth were whiter, you might have community, then sell you a way to make your teeth whiter.
Remember, our lives are increasingly structured by these systems. They decide what we pay attention to, what we download, what we buy, even who we meet. Which type of information the corporations, and operating systems use is really important.
To fix this, you could have two recommenders. One standard one, that recommends funnels and tubes. And another, that's trained especially to recommend spaces. That second one would need a different profile of you, containing your sources of meaning.
You could mix recommendations from each of these recommenders. Or find a smart way to use one or the other depending on what the user needs more.
This would be great.
