This passage from @debcha'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.
@mrtazz@debcha I haven't gotten to this point yet but enjoying it a lot so far. And boy this encapsulates something I've struggled to articulate before...
@mrtazz@luis_in_brief Thank you both! — and yes, this is simultaneously a familiar phenomenon across fields and one that bears giving voice to when we can, not least because it is nearly always humans who provide the necessary slack and resilience to systems, and the cost they pay to do so is often disregarded because it doesn’t show up on a balance sheet.
@debcha@mrtazz yeah. I am dealing with that a lot in the open source software space right now, where there is an ongoing decade-long drive to make this inherently inefficient, very human chunk of valuable infrastructure more amenable to control and efficiency. I often talk about it in terms of James C. Scott's Seeing Like A State, but I suspect your book will also provide a lot of useful shortcuts as I finish it.
@debcha@mrtazz People down the river from you just valued open source infrastructure at $8T and yet we have very little language to discuss it, which is ... a real mess waiting to happen (already happening). https://social.coop/
@luis_in_brief Thank you for reading it! Coincidentally, I’m on a train en route to New York to speak with a group of researchers who’ve received funding to investigate different aspects of open-source digital infrastructure, so this is a perfectly timed vote of confidence, thank you. (And I specifically acknowledge Seeing Like A State at the end of my book because I never cited it directly but it really shaped how I think about these systems.)
@mrtazz@debcha I'm an engineer in the infrastructure world.
This argument is an eloquent "they don't build them like they used to." Survivorship bias, redundancy in retrospect.
I've read the engineering magazines from before the 1906 San Francisco quake. Huge debates how how to best build with concrete. But not afterwards. We celebrate Julia Morgan's clock tower, not the failed brick buildings.
Even after the 1906 quake, they didn't improve building codes. The rich got better buildings, but it took another 70 years before California got serious about quake safety for normal houses.
The decision not to install enough lifeboats on the Titanic was absolutely an "efficiency" decision. To keep allowing cheap brick construction in earthquake areas after 1906.
Heck, the first law code in human history included building quality, about building collapses. Code of Hammerabi.
Humans will human and go cheap. What survives got lucky. There is no magic past where things were always better built.
@sewblue@mrtazz The argument of this section (and the book as a whole) is literally the opposite of ‘they don’t build them like they used to’: this is how we’ve been building things for a long while now, and we’re going to need to do better with our infrastructure and systems in the future if we want them to function at all (not least because of anthropogenic climate instability).
@debcha@mrtazz I work on that infastructure. I don't agree. Based on what I've read here at least.
In my industry we are on a 4-500 year replacement cycle on 100 years-ish worth of infrastructure. Not building anything is a far bigger risk than not going far enough when we do.
A 700 year replacement cycle isn't out off normal.
Eventually it will all start breaking at once and you will end up like Flint Michigan or Jackson Mississippi.
Redundancy delays when you get there. It isn't a fix.
That said the exact industry matters. Telecom is awful about investing in infastructure, primarily because their rate structure and regulatory environment allow it. Water, gas, electric and public works are different, because they often have an "obligation to serve" which changes investment strategies.
@sewblue@debcha@mrtazz I'm now trying to work out what infrastructure has a replacement cycle measured in centuries. I'm guessing that's dykes... In .nl. I can't think of any other infrastructure that has been around long enough in roughly the same state now as it was even 100 years ago. Let alone in 400 years time...
@sewblue@mrtazz I’m familiar with survivorship bias, with the mathematics of replacement of water pipes in the US, with the issues in Flint and Jackson, etc. I suspect that our positions are not dissimilar and that we have a shared commitment to these systems, but I don’t think this medium is serving our discussion well and I’ve (obviously!) made my arguments at length elsewhere, so I’ll refer you to that if you’re interested. Thanks!
All modern human-technical systems (i.e. ALL our systems) operate in failure mode except in extremely unusual "average" conditions and they ALL then require heroic human efforts to recover (or operate at all) when conditions deviate from average.
@debcha@Tallish_Tom@mrtazz for those not aware of Cook's work he was an anaesthetist, and was drawing on his experiences in healthcare. Basically despite everything improviement methodology advocates say, health and other human systems aren't factories, so drawing lessons from production line optimization is fraught, and may have adverse consequences on patient/user/customer outcomes.
@mrtazz@Tallish_Tom@epistatacadam In her 1989 CBC Massey Lecture series, “The Real World of Technology” (the audio is still online, and there is also a book), Prof Ursula Franklin describes this as a societal fixation on a ‘production’ model, vs a ‘growth’ model — her example was higher education but it’s broadly applicable, and it’s been a touchstone for me since I first came across them as an engineering student.
This gets me so angry.
I more and more are convince that the "production model" or Taylorism doesn't even work for production (and never has).
It is very similar to many economic models, which feel to me like wishful thinking. If the outcomes don't match the one predicted by the model, you did it wrong and you need to follow the model more closely.
Add comment