Efficiency is dangerous and slowing down makes life better • Hacker News Board

Re: Efficiency is dangerous and slowing down makes life better

#291

Post by lexicality » Sat, Aug 22, 2020, 1:00 PM UTC

bschne wrote: post #2

I've found "efficiency as the opposite of stability" a very powerful concept to think about - even though it's fairly simple, it seems to be almost a fundamental law. Whether it's about the economy at large, your own household, a supply chain, what have you - as soon as you optimize for efficiency by removing friction, you take all the slack/damping out of the system and become instantly more liable to catastrophic f…

It's funny, I've been sucked into factorio after the 1.0 release announcement and this really rings true.

If you make a production line that has perfect throughput with no buffers then you get fantastic efficiency and productivity right up until a single train hits a biter and is delayed by 30s.

Then you spend 3 hours trying to deal with every machine being out of sync with every other machine with constant start/stops :(

Re: Efficiency is dangerous and slowing down makes life better

#292

Post by bobwernstein » Sat, Aug 22, 2020, 2:40 PM UTC

war1025 wrote: post #104

Taek wrote: post #46

This comment thread is making me realize we don't have a good word to distinguish between efficiency as in "we only have 7 hospital beds because that's all we need on 99% of each day" and efficiency as in "we replaced steps X,Y,Z with just step X', because we found that X' could accomplish everything that XYZ could accomplish but it's faster, more accurate, and cheaper". One makes a tradeoff by reducing overheads and…

This seems like it sort of ties in with NN Taleb's idea of "Anti-fragile" [1]. Perhaps also Chesterton's fence [2]. Maybe also the whole premature-optimization thing [3]. And of course the too-clever-by-half coyotes [4]. Really maybe it just comes down to "be wary of making changes that reduce resiliency." I was hoping to come up with something cohesive with this comment, but really I guess I just agree with what you…

the way you wrote this comment, with the sources fully visable at the bottom, amazing. I will start doing this myself.

Re: Efficiency is dangerous and slowing down makes life better

#293

Post by akendo » Sat, Aug 22, 2020, 2:53 PM UTC

clairity wrote: post #162

civilized wrote: post #146

Earlier quoted context omitted.

> efficiency as the opposite of stability In statistics, there is a slight variant of this thesis that is true in a precise formal sense: the tradeoff between efficiency and "robustness" (stability in a non-ideal situation). For example, if you have a population sample, the most efficient way to estimate the population mean from your sample is the sample mean. But if some of the data are corrupted, you're better off…

i stumbled on "stability' too, because it's a static quality. rather than robustness, i prefer to use the term resilience, a dynamic quality, since efficiency is also a dynamic quality. you can trade efficiency for resilience and vice versa (as the parent poster switched to later). edit: i should add that i don't entirely agree with the thesis of the article, which exhorts us to slow down, thereby trading efficiency…

Do you have any reference or source of this statement? Just asking out of curiosity.

Re: Efficiency is dangerous and slowing down makes life better

#294

Post by adwn » Sat, Aug 22, 2020, 3:11 PM UTC

war1025 wrote: post #104

Taek wrote: post #46

This comment thread is making me realize we don't have a good word to distinguish between efficiency as in "we only have 7 hospital beds because that's all we need on 99% of each day" and efficiency as in "we replaced steps X,Y,Z with just step X', because we found that X' could accomplish everything that XYZ could accomplish but it's faster, more accurate, and cheaper". One makes a tradeoff by reducing overheads and…

This seems like it sort of ties in with NN Taleb's idea of "Anti-fragile" [1]. Perhaps also Chesterton's fence [2]. Maybe also the whole premature-optimization thing [3]. And of course the too-clever-by-half coyotes [4]. Really maybe it just comes down to "be wary of making changes that reduce resiliency." I was hoping to come up with something cohesive with this comment, but really I guess I just agree with what you…

> [4] https://www.epsilontheory.com/too-clever-by-half/

Thank you for this link. I'm halfway through that article and will probably read every single one on that website.

Re: Efficiency is dangerous and slowing down makes life better

#295

Post by goodsector » Sat, Aug 22, 2020, 3:16 PM UTC

civilized wrote: post #146

bschne wrote: post #2

I've found "efficiency as the opposite of stability" a very powerful concept to think about - even though it's fairly simple, it seems to be almost a fundamental law. Whether it's about the economy at large, your own household, a supply chain, what have you - as soon as you optimize for efficiency by removing friction, you take all the slack/damping out of the system and become instantly more liable to catastrophic f…

> efficiency as the opposite of stability In statistics, there is a slight variant of this thesis that is true in a precise formal sense: the tradeoff between efficiency and "robustness" (stability in a non-ideal situation). For example, if you have a population sample, the most efficient way to estimate the population mean from your sample is the sample mean. But if some of the data are corrupted, you're better off…

I may be wrong, but it seems to me that 20th century (theoretical) statistics research overemphasized efficiency at the expense of robustness. My guess is that this has to do with the (over-)mathematization of statistics in the past century, as opposed to a more empirical/engineering viewpoint. Efficiency typically only holds under extremely narrow (and often impossible to check) assumptions, which is great for mathematicians proving theorems and creating theories of efficiency. On the other hand, robustness is ideally about unknown unknowns and weak assumptions, which is hard to deal with mathematically.

It seems already the 21st century is seeing a more balanced emphasis on theory vs. real world applications though.

Re: Efficiency is dangerous and slowing down makes life better

#296

Post by sriku » Sat, Aug 22, 2020, 3:24 PM UTC

bschne wrote: post #2

I've found "efficiency as the opposite of stability" a very powerful concept to think about - even though it's fairly simple, it seems to be almost a fundamental law. Whether it's about the economy at large, your own household, a supply chain, what have you - as soon as you optimize for efficiency by removing friction, you take all the slack/damping out of the system and become instantly more liable to catastrophic f…

Another way I like to state this is "bottlenecks are beautiful design devices".

Re: Efficiency is dangerous and slowing down makes life better

#297

Post by chalst » Sat, Aug 22, 2020, 4:27 PM UTC

goodsector wrote: post #295

civilized wrote: post #146

Earlier quoted context omitted.

> efficiency as the opposite of stability In statistics, there is a slight variant of this thesis that is true in a precise formal sense: the tradeoff between efficiency and "robustness" (stability in a non-ideal situation). For example, if you have a population sample, the most efficient way to estimate the population mean from your sample is the sample mean. But if some of the data are corrupted, you're better off…

I may be wrong, but it seems to me that 20th century (theoretical) statistics research overemphasized efficiency at the expense of robustness. My guess is that this has to do with the (over-)mathematization of statistics in the past century, as opposed to a more empirical/engineering viewpoint. Efficiency typically only holds under extremely narrow (and often impossible to check) assumptions, which is great for mathe…

Not my impression.

The kind of outlier-culling technique suggested by civilized is not recommended these days because it adds unprincipled choice points to what Andrew Gelman calls the 'Garden of Forking Paths' [1, 2]. Thus they are bad for hypothesis testing, which tends to be what most statisticians care about.

Additionally the technique obscures the relationship between the variance of the sample and the population variance if we do not have reliable knowledge of the population distribution; likewise for the mean if the mean is not close to the mode. These problems can be quite dramatic for long-tailed distributions.

[1]: http://www.stat.columbia.edu/~gelman/research/unpublished/p_... [2]: https://statmodeling.stat.columbia.edu/2016/09/30/why-the-ga...

Re: Efficiency is dangerous and slowing down makes life better

#298

Post by themeiguoren » Sat, Aug 22, 2020, 4:46 PM UTC

jjoonathan wrote: post #54

bschne wrote: post #2

I've found "efficiency as the opposite of stability" a very powerful concept to think about - even though it's fairly simple, it seems to be almost a fundamental law. Whether it's about the economy at large, your own household, a supply chain, what have you - as soon as you optimize for efficiency by removing friction, you take all the slack/damping out of the system and become instantly more liable to catastrophic f…

It is a fundamental law. Systems theory has the concepts of "gain margin" and "phase margin" -- how much you can amplify feedback or delay feedback, respectively, before your self-adjusting feedback mechanism fails to find equilibrium and turns into an oscillator. Even though most non-engineering systems don't fit the mathematical theory, the idea that only a finite amount of gain + delay is available, and that the t…

Yeah, as a controls engineer I’ve worked on systems where the main requirement was not gain or phase margin, but a root sum square of the two. Nichols plots drive home the idea that it’s not really the margin at two discrete points that matter, but the close approach to the critical point.

Re: Efficiency is dangerous and slowing down makes life better

#299

Post by xwdv » Sat, Aug 22, 2020, 5:42 PM UTC

Florin_Andrei wrote: post #249

xwdv wrote: post #198

Earlier quoted context omitted.

This isn’t really too correct. A lot of things are neither stable nor efficient.

> A lot of things are neither stable nor efficient. Yes, and those are the shitty things. But when you want to improve them, make them non-shitty, you're facing a choice: stability, or high performance - pick one.

Life is short and we’re not building generational ships so let’s just go fast and efficient and if it crashes it crashes, but by then the lion’s share of profits will have been made and maybe even spent.

Re: Efficiency is dangerous and slowing down makes life better

#300

Post by civilized » Sat, Aug 22, 2020, 5:46 PM UTC

chalst wrote: post #297

goodsector wrote: post #295

Earlier quoted context omitted.

I may be wrong, but it seems to me that 20th century (theoretical) statistics research overemphasized efficiency at the expense of robustness. My guess is that this has to do with the (over-)mathematization of statistics in the past century, as opposed to a more empirical/engineering viewpoint. Efficiency typically only holds under extremely narrow (and often impossible to check) assumptions, which is great for mathe…

Not my impression. The kind of outlier-culling technique suggested by civilized is not recommended these days because it adds unprincipled choice points to what Andrew Gelman calls the 'Garden of Forking Paths' [1, 2]. Thus they are bad for hypothesis testing, which tends to be what most statisticians care about. Additionally the technique obscures the relationship between the variance of the sample and the populatio…

This post seems to be conflating a few different things:

1. Trimming for mean estimation, which removes extreme values in an algorithmic fashion

2. Subjective removal of outliers based on researcher judgment (this is the garden of forking paths Gelman talks about)

3. Estimating other distributional properties, such as the variance, with trimmed estimators

These are all different things and come with different theoretical and practical risks and benefits. Trimmed means are perfectly good statistical tools, although they have their limitations like anything else.