Daniel Wagner
Esteban Küber 
Efficiency is the removal of redundancy. Redundancy is a necessary element of resilient systems. The unbounded search for efficiency has one result: brittleness.
During the COVID lockdowns, and to a lesser degree when the Evergiven got stuck, we saw to how much the Just In Time invisible chains of production and distribution around us affect our lives. As climate change significantly disrupt farming throughout the world, we'll continue to see similar effects.
I'm all for global chains of distribution: specialization is real and you won't ever grow bananas in Alaska (even if Iceland showed it is technically possible last century), manufacture a computer chip in Paris or produce wine in Greenland, but we as consumers have to accept and understand that having pineapple out of season anywhere in the planet is not reasonable at the prices we're used to paying. In some French super markets I've seen signs on the produce next to the price with the country of origin and helpful information of when the growing season is. I found that as an excellent nudge for the almost entirely fictional homo economicus. I'd like us to surface that information to everyone for everything. Maybe that way people would understand just how connected we are.
But that interconnectedness and search for "efficiency" is not only in financial or production systems, it exists in software, hardware, any kind of industry, and in governments. When we "engineer out" expertise out of government to the private sector, government then lacks that expertise entirely. When we engineer away redundancy from software, we end up with software that can fail in catastrophic ways.
One thing I keep seeing is the same lessons being learned across different disciplines: pilots and doctors learning about the importance of checklists, road and industrial machine engineers learning about safe by default design, industrial and software UX designers learning about how to best make machines and humans talk to each other. We need more cross pollination. Across industries. Across borders. Across people. That's how we build a better future. And for that we need to listen.
Brennan Lee Mulligan and Matt Mercer talk "Ma" (間)
Eleanor Saitta@ekuber
Are you familiar with the Resilience Engineering Association? If not, I have a real treat for you.
Are you familiar with the Resilience Engineering Association? If not, I have a real treat for you.
@dymaxion i am not!
@ekuber
It's a cross-and multi-disciplinary group looking at how high-reliability organizations can, do, and should work, how systems fall, and how the human parts of those systems help them not fail completely.
https://www.resilience-engineering-association.org/
Good starting places in the literature are Dekker's "A Field Guide to 'Human Error'", the Woods paper "The Theory of Graceful Extensibility: Basic Rules that Govern Adaptive Systems", and (from a different end of the community) Leveson's Engineering a Safer World.
NaClKnight 🧂⚔️🥊
MidgePhoto@ekuber
If I'd written a(nother) book it would have been called "Failure Modes". They repeat.
If I'd written a(nother) book it would have been called "Failure Modes". They repeat.
Kinene⭐🐻@ekuber Don't forget to include those in the arts. We are able to think beyond the various professional boxes to literally see bigger pictures.
@c_merriweather "limitations are good for innovation" is a good one that covers art and engineering
datenwolf@c_merriweather @ekuber I keep telling my peers and students, that good engineering starts with well defined constraints.
Ype Kingma@ekuber
Even within the field of distributed computing "achieving high reliability is expensive".
Leslie Lamport, 1984, "Using Time Instead of Timeout for Fault-Tolerant Distributed Systems"
wroof
Rene Gat
MostlyTato@ekuber
Aim for the best, expect the worst and plan for it. Its a generic concept that works in all areas.
Its also an understanding that optimism and pessimism are false constructs when practical realism enters the chat.
Aim for the best, expect the worst and plan for it. Its a generic concept that works in all areas.
Its also an understanding that optimism and pessimism are false constructs when practical realism enters the chat.
haui ☭
🇵🇸@ekuber
It reminds me of people trying to figure out gravity before it was common knowledge. They ascribed the fact that stuff falls down to god or any other outlandish idea.
The same happens if you ask a liberally educated person to explain cooperation.
They use individualism, idealism, utopianism, reinvent the wheel.
We have tools for this for over 100 years. It is called planned economy, socialism and materialism. Its dead easy but the knowledge has been forcefully replaced.
Sylvain Wallez@ekuber Yep, I like this French regulation that requires displaying all this information. We always make a deliberate choice of buying seasonal products that are produced locally, even if they're more expensive.
@swallez so many small things that add no cost but have a material impact. I always think of the legislation forcing plastic bottle caps to remain attached by default. Such a small thing, such a large impact on both loose trash and recyclability. There are so many small changes we can do that no one would bat an eye at that would improve all our lives. So many others that people would have kneejerk reactions towards that would be beneficial to all.
Alsy@ekuber True! And for me, for things other than food, this also starts with questioning myself "do I really need it?". Not always easy as a geek that loves home automation, but even there I try to stay frugal. I also repair a lot of things. We changed our dishwasher recently after 35 years and many repairs, basically my entire adult life 😅
gabriele renzi@ekuber i think it's an EU regulation that you have to indicate the origin of fruits and vegetables. This is certainly the case in Italy too, they generally indicate regional production too. And it does impact sales somewhat, but people buy out of season fruit from abroad just the same, cause it's a cultural issue.
Richard W. Woodley ELBOWS UP 🇨🇦🌹🚴♂️📷 🗺️
@the5thColumnist I said Paris, as in the 20 arrondissements, not France in general
What is it about Paris that makes that unlikely.
@the5thColumnist chip manufacturing is done in specialized factories, supplied by an entire chain of production for industrial machinery, expertise and needs space. Paris lacks all three. It's the same reason you're not going to have a farm in London, nor a dry-dock in Milan. The conditions aren't there to support it, and even if you changed that your have to significantly destroy what's already there to allow for it.
Are you being purposely obtuse?
Are you being purposely obtuse?
No. I just was not aware of the geography of Paris, other than it being the national capital.
Janis@ekuber I'm sorry, all I can think is, "well, of course Epstein had a few effectively non-English-speaking friends... and employees." It's that nasty habit of American narcissistic thinking. i don't wonder where I got it.
From a different angle, the Dalai Lama also has some lucid perspectives on interconnectedness that do address living without proverbial pineapples.
Jan Lehnardt 
@ekuber i was just bringing this up in the context of all but essential-for-rail-service personnel on train stations. It’s around-water-freezing cold here and there is lots of snow and ice. It’d take a station keeper an hour tops to clean things and make usage safe, but no, we optimised that position away and now we play slip-and-slide with serious injuries for the third week running. Great system.
Graydon@ekuber Preach!
By and large, the motivation for efficiency to maximize the amount that can be extracted from that step of the process; the brittleness and fragility of the machine upon which everyone's lives depend has been put in so people get rich as much as possible. (It's not even "get rich", it's "as much as possible" or maybe "as soon as possible" if you can tell those apart.)
If we want to live through the time of angry weather, we're going to agree that nobody is or gets rich.
HTTP 1.1/418 Teapot@ekuber This is a wonderfully concise phrasing of something I regularly wave my arms and froth about when talking to people about this stuff. Thank you!
Daniel@ekuber @janl "how infrastructure works" has a great paragraph about this too https://chaos.social/@mrtazz/111844063360502168
Daniel (@[email protected])
Attached: 1 image This passage from @[email protected]'s "How Infrastructure works" is such a truth that often gets forgotten or ignored on the hunt for profit. It's a very familiar and recurring theme in resilience engineering texts and research. And it also rings true for me in this current trend of continuous layoffs that take more and more slack and capacity out of tech systems being maintained (in addition to the human cost) as remaining humans need to do more work in the same amount of time.
Alex Hoffmann@ekuber that is so nicely framed. @wendynather
sean watters
Weekend EditorThere's a book on this theme by Tom DeMarco, _Slack_. The general idea is that complete optimization causes fragility; to be robust requires slack to handle emergencies, and to innovate.
A bit old now -- from 2001 -- but we still haven't learned the lesson, so...
Harald WernerDie allzu hart sind, brechen.
Die allzu spitz sind, stechen
Und brechen ab sogleich,
Und brechen ab sogleich.
(Wolf Biermann)
kcp
Seb
FediThing 
JohnMashey@ekuber
Agreed. Years ago, a group of us visited the NUMMI car factory, joint venture of GM and Toyota. They started with talk about place & methods, emphasizing importance of Just in Time logistics… and were quite embarrassed when, during our visit, the production line stopped because some truck coming from L.A. got delayed.
https://en.wikipedia.org/wiki/NUMMI
The brittleness of hyper-efficient systems leading to societal collapse was one of the themes of A Deepness in the Sky:
Wandering Star@ekuber This reminds me of an argument I had with management of the network operations center where I worked. I made a beautiful point about "pickle factory staffing" versus "firehouse staffing" but it took them months to see my point.
Jayne
ティージェーグレェThat's a different perspective of how I perceive efficient systems as a decades long sysadmin sort.
To me: downtime is inefficient.
Redundancy: improves system availability and uptime, thus maximizing resource utilization.
Figuring out the balance is a challenge, perhaps even an art form.
I've been pretty good at it on occasion!
To me: SPOFs are horribly inefficient and to be avoided and mitigated whenever possible.
Let's take a simple example, of a NAS (Network Attached Storage). Presumably, comprised of a RAID (Redundant Array of Inexpensive [well, originally that was the acronym] Disks).
A RAID0? Hella fast, horrible, a single disk in the array will take the entire thing down, so for each disk in a RAID0? It adds a multiplicative rate of failure.
RAID1? Redundant, but meh.
RAID5? Now we're getting somewhere! Has some decent trade offs. Can lose a disk and still rebuild, but has some better advantages for a bit more storage and not horrible speed sacrifices.
RAID6/(NetApp would call this: RAID-DP, [Open]ZFS RAIDz2): gettin a bit more sophisticated: can lose two disks in an array simultaneously and still recover. Circa 2006? I encountered failures such as that when drive densities were around 72GB and a 2TB NetApp array cost around $60,000. Still: being able to recover from that kind of failure (and stay up and have data available)? Total life saver of a feature. The BER (Bit Error Rate) of disks, tends to go up, as disks get denser, and guess what the storage industry does? Makes denser disks cheaper and more widely available as part of their ongoing business model/design.
OpenZFS RAIDz3 (able to endure 3 simultaneous disk failures) is about the minimum I would tolerate for well over a decade; even better: a CARP (Common Address Redundancy Protocol) HA (High Availability) pair of OpenZFS RAIDz3 systems (kind of imagine creating a RAID1 of two RAIDz3s; but the abstraction layer is more at the TCP/IP layer, so the NAS is available, even if one goes totally offline for some reason).
There are of course, more elaborate realms of redundancy (I haven't even talked about off site stuff yet for DR[Disaster Recovery]) and generally speaking: the more impactful downtime is? The more critical such designs become. Circa 2002-2006 I was a Network Security Specialist at a company that had a "burn rate" of $100,000/minute for each minute of down time. My annual salary? Was like $75,000/year; so I felt the stress there pretty keenly if things went down for any appreciable period of time.
Heck, even before that I helped an employer go from "two nines" to "three nines" of uptime (despite being a much smaller company with a much lower burn rate, and in an industry where "five nines" was considered industry standard) but these days? It seems as if AWS doesn't even manage three nines, it's dumb founding; but I guess Jeff BeeZoos empire is too big to fail or something? I don't get it. Must be a cushy job if you can get it (when he isn't laying off 16,000 people at a time).
To me: downtime is inefficient.
Redundancy: improves system availability and uptime, thus maximizing resource utilization.
Figuring out the balance is a challenge, perhaps even an art form.
I've been pretty good at it on occasion!
To me: SPOFs are horribly inefficient and to be avoided and mitigated whenever possible.
Let's take a simple example, of a NAS (Network Attached Storage). Presumably, comprised of a RAID (Redundant Array of Inexpensive [well, originally that was the acronym] Disks).
A RAID0? Hella fast, horrible, a single disk in the array will take the entire thing down, so for each disk in a RAID0? It adds a multiplicative rate of failure.
RAID1? Redundant, but meh.
RAID5? Now we're getting somewhere! Has some decent trade offs. Can lose a disk and still rebuild, but has some better advantages for a bit more storage and not horrible speed sacrifices.
RAID6/(NetApp would call this: RAID-DP, [Open]ZFS RAIDz2): gettin a bit more sophisticated: can lose two disks in an array simultaneously and still recover. Circa 2006? I encountered failures such as that when drive densities were around 72GB and a 2TB NetApp array cost around $60,000. Still: being able to recover from that kind of failure (and stay up and have data available)? Total life saver of a feature. The BER (Bit Error Rate) of disks, tends to go up, as disks get denser, and guess what the storage industry does? Makes denser disks cheaper and more widely available as part of their ongoing business model/design.
OpenZFS RAIDz3 (able to endure 3 simultaneous disk failures) is about the minimum I would tolerate for well over a decade; even better: a CARP (Common Address Redundancy Protocol) HA (High Availability) pair of OpenZFS RAIDz3 systems (kind of imagine creating a RAID1 of two RAIDz3s; but the abstraction layer is more at the TCP/IP layer, so the NAS is available, even if one goes totally offline for some reason).
There are of course, more elaborate realms of redundancy (I haven't even talked about off site stuff yet for DR[Disaster Recovery]) and generally speaking: the more impactful downtime is? The more critical such designs become. Circa 2002-2006 I was a Network Security Specialist at a company that had a "burn rate" of $100,000/minute for each minute of down time. My annual salary? Was like $75,000/year; so I felt the stress there pretty keenly if things went down for any appreciable period of time.
Heck, even before that I helped an employer go from "two nines" to "three nines" of uptime (despite being a much smaller company with a much lower burn rate, and in an industry where "five nines" was considered industry standard) but these days? It seems as if AWS doesn't even manage three nines, it's dumb founding; but I guess Jeff BeeZoos empire is too big to fail or something? I don't get it. Must be a cushy job if you can get it (when he isn't laying off 16,000 people at a time).
Christian Berger DECT 2763@ekuber Yes this is true, but there's a new kind of efficiency hindering process that provides no resilience against crises. It's things like Billionaires who, increasingly, not only waste valuable resources directly, but also influence policy making into wasting even more resources. It's what got us into car-centric worlds, despite of cars being highly inefficient and fragile systems.
Andy Linton 🇮🇪 
✅Perhaps there's a lesson here for Wellington Water, the local councils and nNZ government.
Ölbaum
Noel Kelly