A great word to taboo in hardware tech is "performance".

I once witnessed a very heated argument among a group of programmers discussing some code on a microcontroller. Half of them thought the performance was abysmal. The other half thought the performance was great. To everybody it was obvious to see what the performance was, why were the other half being so bone-headed about it?

Turned out the first group were trained mechanical engineers, the second were trained software engineers. For the first group, "performance" meant how closely the motion of the device attached to the microcontroller tracked the optimal path. I.e. the performance of the function approximation that the microcontroller code was doing. To the second group, it meant how fast the code ran. This was completely irrelevant to the first group, since as long as it managed to run 1000 times per second (which it easily did), nobody cared how many nanoseconds there were to spare each cycle. Why are you even talking about that? The performance is shit, we need to improve it and who cares if it takes a few more CPU cycles.

Applause! I hate the word "performance" and its evil spawn, "performant".

Do you mean speed? Efficiency? Accuracy? Say what you mean.

This is my excuse for why I have no idea what "performance optimization" entails or means. Now to convince interviewers..

In a context of an interview it means how performance of an algorithm scales with respect to the input length. This is measured in Big O notation. Understanding that is seldom very helpful on the job, but it's very important for interviews, so make sure you learn it.

That's not specific enough. Memory usage, elapsed time, and energy usage are all functions of the input size, and each is a different optimization problem. Instead of talking about performance optimization, you should be talking about memory optimization or speed optimization.

Hahaha I appreciate the candor. Frankfurt's On Bullshit has never been more appropriate than at this time in my life and in history.

I wish it was a common practice to ban certain words from difficult conversations. But whenever I suggest it people just talk past me as though I've said some weird thing.

There's facts and opinions, but also definitions, and too often disagreements boil down to a different definition of a single word. Ban that word and say what you mean and the disagreement may suddenly be gone.

Edit: I wrote this before reading the second paragraph. :) I'll leave it as an alternate way of saying the same.

Let me ask, has anyone successfully "tabooed" certain words, or do people not understand the purpose of doing so?

I don't use the "taboo" terminology, but I frequently ask people to operationalise their definitions in arguments. ("What specific steps would I go through to test whether the system is scalable or not?")

The first time I ask, people ignore it and/or don't understand it.

The second time I ask, people get uncomfortable because generally they haven't thought about what they mean to that level of detail, and they try to weasel out of it.

The third time I ask, it diverges. Some people give me a reasonable definition. Some people say "sorry that's not possible" and depending on the subject, I pause the argument at that point to do something else.

> The first time I ask, people ignore it and/or don't understand it.

What if this also happens the second, third and Nth time you try?

> The second time I ask, people get uncomfortable because generally they haven't thought about what they mean to that level of detail, and they try to weasel out of it.

Or maybe people decide to stay here, believing their subjective views to be good enough answer for such a question. Then what?

This is where it has paid for me to practise interpersona skills. Once I'm outside of the common situations every situation is unique and I have no general advice.

I don’t think the parent commenter’s intent was that the words be universally banned, but instead context-specific words banned for a specific conversation. It’s just trying to remove implicitly differing definitions.

Typically I just disengage myself from the topic (or the drama/tension that is brewing up) and start asking what people mean when they use a certain word.

We've somehow started arguing about optimisation, for example. Time to take a step back and ask what "optimisation" actually means.

In fact, me saying "let's step back a bit" is pretty much my go-to phrase for saying we've gone too far down a rabbit hole and we should come back up for air.

I don't think I would have much success trying to make certain words taboo. Much better to create a shared understanding of what something means instead of trying to avoid it.

I didn't know this was a thing that had a term either! I don't talk politics anymore except with folks that I explicitly know want to talk politics, but I frequently suggest that people shouldn't use a whole litany of terms in those discussions: right/left-wing, capitalist, socialist, liberal, conservative, communist, libertarian, fascist...so many people use these terms as short-hand for "stuff I really/don't like", and then project values and policies onto them.

Throw them out, talk about the issues, talk about your shared values, etc, and you'll quickly find that you agree with people a lot more than you realized.

I didn't want to mention politics, but "Critical Race Theory" is a prime example and should really be tabooed as a temporary exercise to aid understanding. So many discussions about it without people having a consistent understanding of what it means. I think they've even written laws regarding it without defining what it means. But, sadly, politics is rarely about understanding each other or reaching consensus.

> But, sadly, politics is rarely about understanding each other or reaching consensus

This made a recollection (possibly a lucid dream) about a country who has replaced an (implicit or explicit) adversarial system with one that starts by getting the parties to co-author a document on what they agree upon.

I think this is what our politics (organisational and the representative system alike) should try to implement

Totally agree. Many of these terms mean totally different things to different people. And sometimes contradictory things to the same person. Get rid of the labels and talk issues instead.

And I guess the same is true in the examples of tech discussions where people couldn't agree because "scaling" or "performance" meant different things to different people, because by the nature of their work, they focus on different issues.

So often I see engineers get sucked into arguing something based on a buzzword that they either believe the system they are building needs to conform to, they think other people (outside the company) care about, or because it agrees with their pre-conceived design notions.

Nuking words interrupts this, I often ban the words "scalable", "Automated", "platform", and "system". It forces the conversation back to specifics such as "Ability to handle X requests per second supporting Y users", "Process requires x, y, z steps rather than a,b,c steps", "service supports use cases A, B, C"

Banning those words also makes the conversation easier to follow for people who don't know the jargon. Jargon can be a quick shortcut when everybody is talking the same jargon and is clear about what every word means, but it can become a massive barrier to discussion if one of those criteria is not met.

I have never heard of this approach before, but it sounds flipping great. I will now look for arguments that I can start and then later solve with this method. Winners all around!

> first being specific about what we agree upon.

Or even just trying to define the goal you are trying to accomplish. I've seen many disagreements on software projects because people's definition are completely different. The arguments are just a symptom of a deeper problem that no amount of careful discussions are going to fix.

It is very valuable to be able to spot differences in definitions, so you can address them before moving on. If you feel like you are talking past eachother, that might be the issue causing it.

I have been accused of being dim for not understanding a common word. Or alternatively of intentionally playing dumb.

Then I layout out six reasonable definitions for the word that could fit this context. Then I am accused of making other people feel dumb. Unless it's my wife, who simply claims, "You know what I meant." :)

This might not be the right place to ask..

How to find value from LessWrong? I see people derive value from it, but when I visit it, it’s not obvious

Read "The sequences" [0], may be the Harry Potter and method of rationality book.

For lesswrong (and the Rationalist community in general), I've found that their "broad stroke idea" is what brings the most value. The more in-depth discussions bringing those ideas to specific topics is intriguing to read, but probably much less valuable for daily life.

[0]: https://www.lesswrong.com/rationality

The site has a "curated" feed with articles the moderators deem to be useful.

What value you'll derive from it varies (especially if you're not into AI alignment) but there are a few gems every few weeks.

Tangentially related: https://en.wikipedia.org/wiki/Aumann%27s_agreement_theorem

This seems quite impractical for the real world, since it assumes perfect knowledge, perfect communication and, above all, rationality.

I've often remarked at how common it is to hear two people talking about completely different things using the same words, thinking they're both understanding each other. Communication could be something that we think we're better at than maybe we actually are.

That's why type systems are important in languages.

Of relevance, the adjective "clean": https://www.steveonstuff.com/2022/01/27/no-such-thing-as-cle...

The one thing that was drummed into my head in grad school was to define your terms. I often ask students "How much is 'most?'" as in "most people think X." Technically 50.1% would be "most" but when I say "most" I generally think closer to 75% or more. So people could think they are disagreeing over whether "most people think x" is true but they are really disagreeing over the definition of "most."

Communication is improved if, rather than redefining terms in common use, you just use the terms in common use that already mean what you want. So if you mean “at least 75%”, just say “at least 75%” rather than “most”, especially if you are negating it in a condition where negating it with the common definition would not be true.

>Many intense arguments can be diffused by first being specific about what we agree upon.

Exactly. Whenever I have the inclination to get into an argument about politics, I try to start by finding a common premise, in order to find where the problems or disagreements start. And eventually, I find that we have similar goals where you wouldn't expect sometimes, but are coming at it from completely different sets of facts that guide us to different conclusions.

This is a great idea. Not just for tech discussions; I think some political discussions would also be much improved if we could taboo certain words. Some people get too hung up on certain -isms without explaining what specific policies they're talking about.

People should do this with the word "decentralized".

Without properly-defined terms, where the participants agree on definitions, it's impossible to have a rational discussion grounded in logic.

"You keep using that word. I do not think it means what you think it means."

Or just ask "what do you mean by X".

This is one of those things "rationalists" do that doesn't always make sense. It is possible to tell someone to "taboo their words" to obstruct someone, in order to make it hard for them to discuss something whose meaning you know very well. It's also possible to tell someone to taboo their words because you want to nitpick an imperfect definition--regardless of whether the flaws in thde definition are relevant to the current argument.

I hereby taboo the words "taboo" and "rationalist"