Comment by IceMetalPunk | Hacker News Reader

Comment by IceMetalPunk | Hacker News Reader

I don't know that I've heard the terms "hard and soft skills" before, but I've definitely heard it about "hard and soft sciences", and it's equally frustrating and detrimental in that domain. Psychology specifically comes to mind: I've constantly heard people saying psychology is a "soft science" or even "not a real science" to discredit it, which is absolute BS. People do the same for economics. Thing is, psychology is based in neurology -- a very real physical/chemical/biological science -- and economics is based on psychology. The named fields of "neuropsychology" and "neuroeconomics" make that foundation clear. If abstracting a concept makes the result "not real", then I guess 99% of computer applications are "soft apps" or "not real apps", since they basically always use an abstraction of some sort to work (anything from assembly as an abstraction of machine code, to higher-level programming languages as abstractions of assembly, to external libraries as abstractions of their common functionalities).

People will say psychology is a "soft science" because "it's based on statistics and general probabilities, not hard rules and guarantees". But quantum mechanics is founded on the idea that nothing in it can be guaranteed and it's all just statistics and probability -- yet we consider QM a "hard science".

TL;DR: Can we stop qualifying knowledge and skills based on arbitrary rules about what make them "true Scots-knowledge"?

The concepts are different, but the similarity is that anything easy to measure is "hard" and anything difficult to measure is "soft".

Soft sciences measure complex things, like human behavior, that are challenging to define in a rigorous fashion. People are more comfortable with simple things, like atoms and subatomic particles and rigid objects, which behave as a simple unit following simple laws. That makes biology a "softer" science -- not because it's easier, but because it's more difficult.

Hard skills are easy to measure: can you write the code that does the thing, quickly? But soft skills like "can you figure out what needs to be done?" and "can you communicate with clients and coworkers?" and "can you write clear instructions?" are hard to measure.

We have a notion that only things that can be measured are real. But that skips over the fact that some things are important, but difficult or impossible to measure. It's poor reasoning to assume that only things that have already been given rigorous definitions are relevant, but it's popular among people who don't wish to practice soft skills.

It would be great if we could appreciate that science, and software, are human activities done by humans for humans, requiring a mix of skills both easy and hard to measure.

Maybe "hard/soft" is a bad word couple because "hard" also sounds like "difficult", implying it's higher-value. Maybe it's better to speak of "firm and fuzzy sciences".

It's hard science because the rules are Hard Lines. Clearcut, sharp, easily communicated.

It's soft science because it's squishy. The rules are soft and malleable, depending on the context of the question.

> But quantum mechanics is founded on the idea that nothing in it can be guaranteed and it's all just statistics and probability -- yet we consider QM a "hard science".

Quantum mechanics describes the probabilities exactly, it includes every single way the particle can behave. There is no statistics here, it just gives an exact probability distribution. This is not like psychology at all where we don't have any exact descriptions of how humans can behave in different situations.

Or in another way, the results of physics are "hard" (unyielding) since we have yet found any experiment where the particles don't behave according to the distributions we get from quantum physics. Those distributions are therefore as exact as anything humans can know can be exact.

Psychological experiments aren't that exact, the psychological experiments run produces different distributions all the time since they don't have hard requirements on the input, humans aren't exact, humans are very fuzzy data wise so until we can specify a human precisely psychology will remain a soft science.

Psychology is more like a set of disciplines. Some hard, some soft. Some real and some woo. Lots of great stuff to learn from cognitive psychology, especially for the AI-interested in this forum. Evolutionary psych, though, is a bunch of post facto "just so stories," with my apologies to Kiplng for the comparison.

And no, we can't stop seeming to clam no true Scotsman. That's because there are things masquerading as legitimate. And we're on HN. So any time someone points out the divide, those accusations are inevitable.

> psychology is based in neurology

It isn't though. It's not like math, built on axiomatic foundations. Psychology predates the discovery of neurons.

(Math isn't really like that either.)

Reply on news.ycombinator.com