Knowledge is a tool
The effect on something else
Let’s say I’m thinking about opening a tea shop in Boise. Google tells me its population was 223,154 in 2016. That fact matters because I have a mental model that connects the size of the population to the success of my tea shop: the number of people who will pass by, the assumed percentage of tea vs. coffee drinkers, etc. I have these factors connected in a spreadsheet. The spreadsheet is a model because it identifies the relevant factors and connects them in their relationships with one another.
When you get used to thinking about knowledge in terms of models, it becomes something that matters because of its effect on something else. After all, a cell on a spreadsheet that has no effect on any other cell isn’t really part of the model. But that makes knowledge a tool: something that matters because of its effect on something else.
The mark of knowledge
If using models has hastened our embrace of knowledge as a tool, it’s not hard to see why this has happened now and not 30 or 40 years ago. Personal computing put a spreadsheet machine into every business and then onto every desk. We could even play “what-if” with knowledge, an activity formerly reserved for science fiction writers.
We’re now poised to take the second step. Machine learning systems create their own models from the data we feed them. Those models can be inexplicable to us humans. This cements the more unsettling idea behind pragmatism: The world does not necessarily look like what we know about it. Knowledge is so much a tool that all it has to do is work. That is the mark of knowledge. And that does not necessarily require that the world be much like what we think it is, any more than the wind is like the coolness we feel on our cheek.