Three experiences today all drove home the point two people in different fields can both nominally ‘know’ about the same subject or technique, but when you line up the details of that knowledge, it’s actually amazingly different.
This morning, my computer vision professor put a 3x3 grid of numbers on the projector and nonchalantly referred to it as a filter. It took me a bit to figure out how it related to my concept of a filter – either a physical circuit or a computer program that took a signal and modified components of it in phase space (usually cutting out the high frequency component.) It turns out that grid took as its ‘signal’ the pixels in the grid and gave an output that applied to the center pixel. So if you squint very hard both ‘filters’ do the same thing, but in very different ways.
I spent this afternoon struggling with the fact that there are different conventions for what different components of quaternions mean based on whether you are in aerospace or computer science it also depends on where you are in time – the convention has changed over the years. Since doing the same operations on two quaternions written with different conventions in mind yields very different results, this was rather frustrating.
And this evening, in the most recent EconTalk episode, Russ Roberts referred to the idea of degenerate cases as an ‘econometrics concept.’ While I’m sure it is used in econometrics, it was kind of shocking hearing it explained that way. I know it in an entirely different garb - as an idea that stems from linear algebra and has applications in spacecraft control – specifically in controlling their attitude with reaction wheels.
Tying the three together: in broad strokes, techniques that share a name do the same thing. But between these different areas with different conventions, the nitty gritty of how it achieves that end, and what using the technique entails is VASTLY different.
There are two take home points from this fact: one is to just be aware of these differences – you can’t assume that because someone knows or used a science/math tool that you are familiar with, that they actually did what you think they did.
The other point is that there is still a lot of room for conversation between even very similar fields. I’m interested in how we could make tools that can translate between the needs of the different fields, both so that there is more efficiency both in communication and in transfer of skills.