Other than very principaled os quacks everyone uses scihub and LibGen, no os needed
> They create all sorts of abominations like genetically modified beer and animals that glow like jellyfish.
While gmo beer is nice, I should note that bioluminescence is the most bannal and boring thing in lab biology. People do it because you essentially get "for dummies" guide about making anything and everything bioluminescencent. That is because it's critical for many experiments so we're really good at it. But it's equivalent to writing an http.server in programming, everybody can do it in the 20s, even grandma.
---
On the whole though, it seems that the obvious difference between the two stories is that the requirements in the former are fuzzier.
It may well have cost DARPA less to draw up entirely new nuke designs, or a whole new WMD program... but it might have proved problematic to change in other ways
Not to say I don't agree with the general flow of ideas here, but I think the spec fuzziness angle was a bit unexplored... though I'm not sure it makes for as good a parable.
I think of this as 'playing' the system. You need to poke enough handles on a system so you can get faster feedback. Then keep poking and listening to what returns. Second and higher order interactions ensue etc. Learn to hear the system: it'll tell you what it wants to do.
> Turns out it's really bloody hard to replicate an organisation.
The farther I go in my career, the more wary I am of jobs that promise a unique culture. I'm wary even if I agree with how it's unique or fit for the company.
In these cases, the risks outweigh the benefits. At best, I learn to work in new ways which make me successful here but not anywhere else, and make me less valuable in the future. Or I don't and I fail.
Or just go somewhere there's less of a "unique approach".
Well, there was the small issue that Toyota lied about the miracles of the Toyota Production System (later the Toyota Management System) for almost 30 years. There was even a small industry devoted to spreading the religion (e.g. Productivity Press) that the founders didn't actually use (e.g. all the inventory that Toyota claimed it didn't keep was kept at the huge network of mom-and-pop garage shops in the region around there plants, many of which had appalling health and safety conditions). Toyota made 10s of millions from GM alone offering "TPS training" in a system they didn't actually use.
My suspicion is that the DoE had the full process already and didn't want to junk it for a new one, that's why they had to try their best to recreate the missing foam element. On the other hand the PHDs came up with their own process so there was much more latitude to freely experiment.
Could be that a specific pathway is harder to explicate vs trying to hit a general target. I feel though that the foam story is more indicative of general difficulty with replicating most org structures though ...
Yea I totally agree with your overall post, just a nitpick on the foam one as I've seen many teams scrambling around looking for the "1% bug that makes our process work only 99%".
Just to supplement to your point on losing one person making the organization go down, from my experience sometimes it's the "soft knowledge" e.g. telling people: issue 1 go to person A, issues 2 and 3 go to person B, issues 4 and 5 can be ignored that's not written down and causes the whole thing to go haywire. One would think that it would be trivial to record this but somehow it's often not done...
Interesting article and reminds of this article https://danwang.co/how-technology-grows/ that also talks about process knowledge that is often hard to articulate, forget replicating.
I think we need to understand what is the base rate at which any group of smart people can generate great outcomes. While the freshly minted Phds did successfully build a nuke ( not at all an ordinary feat), would this same set of people been able to produce a similarly stellar outcome in their next assignment. One could argue, there is an element of randomness at play, that would make great outcomes hard to come by, even if we were able to master the art of replicating organizational culture. Perhaps, teams end up building bench strength with each new project, and eventually produce something remarkable that is much greater than the sum of their previous experiences/output.
Good point, and great article! I think its less about the randomness here and the possible reversion to the mean, than the fact that smart folks with higher levels of autonomy in their chosen path will tend to produce innovative outcomes, even when the odds are stacked against them. To industrialise it though requires a team, and we start getting into process knowledge.
Agree that over a sufficient period of time, smart( or reasonably intelligent) people will produce great outcomes.
I think this is true even in academia.. A lot of the well known university incubated companies, e.g. nicira, databricks etc. came into existence from a bunch of grad students working on slightly different areas sitting together in a lab..and the existing valley ecosystem that sort of had institutionalised the process of taking ideas from a research paper to billion dollar companies.
Some nitpicking:
> knowledge from open sourced papers
Other than very principaled os quacks everyone uses scihub and LibGen, no os needed
> They create all sorts of abominations like genetically modified beer and animals that glow like jellyfish.
While gmo beer is nice, I should note that bioluminescence is the most bannal and boring thing in lab biology. People do it because you essentially get "for dummies" guide about making anything and everything bioluminescencent. That is because it's critical for many experiments so we're really good at it. But it's equivalent to writing an http.server in programming, everybody can do it in the 20s, even grandma.
---
On the whole though, it seems that the obvious difference between the two stories is that the requirements in the former are fuzzier.
It may well have cost DARPA less to draw up entirely new nuke designs, or a whole new WMD program... but it might have proved problematic to change in other ways
Not to say I don't agree with the general flow of ideas here, but I think the spec fuzziness angle was a bit unexplored... though I'm not sure it makes for as good a parable.
Good point! And good point ...
>I taught my son how to ride his bike by explaining how it works broadly, letting his play with it, and encouraging when he did the right things
..reminded me of this: https://www.dougengelbart.org/content/view/236/158/
Unsure if that is of any help now or not ;)
Will try with the second one :-) Though I didn't use the training eheel either..
I think of this as 'playing' the system. You need to poke enough handles on a system so you can get faster feedback. Then keep poking and listening to what returns. Second and higher order interactions ensue etc. Learn to hear the system: it'll tell you what it wants to do.
> Turns out it's really bloody hard to replicate an organisation.
The farther I go in my career, the more wary I am of jobs that promise a unique culture. I'm wary even if I agree with how it's unique or fit for the company.
In these cases, the risks outweigh the benefits. At best, I learn to work in new ways which make me successful here but not anywhere else, and make me less valuable in the future. Or I don't and I fail.
Or just go somewhere there's less of a "unique approach".
The same thing happened at the NUMMI[0] plant. Despite Toyota sharing all its secrets, GM couldn't replicate the production process
0: https://www.thisamericanlife.org/561/transcript
Well, there was the small issue that Toyota lied about the miracles of the Toyota Production System (later the Toyota Management System) for almost 30 years. There was even a small industry devoted to spreading the religion (e.g. Productivity Press) that the founders didn't actually use (e.g. all the inventory that Toyota claimed it didn't keep was kept at the huge network of mom-and-pop garage shops in the region around there plants, many of which had appalling health and safety conditions). Toyota made 10s of millions from GM alone offering "TPS training" in a system they didn't actually use.
Really interesting, thank you! Its a great example.
My suspicion is that the DoE had the full process already and didn't want to junk it for a new one, that's why they had to try their best to recreate the missing foam element. On the other hand the PHDs came up with their own process so there was much more latitude to freely experiment.
Could be that a specific pathway is harder to explicate vs trying to hit a general target. I feel though that the foam story is more indicative of general difficulty with replicating most org structures though ...
Yea I totally agree with your overall post, just a nitpick on the foam one as I've seen many teams scrambling around looking for the "1% bug that makes our process work only 99%".
Just to supplement to your point on losing one person making the organization go down, from my experience sometimes it's the "soft knowledge" e.g. telling people: issue 1 go to person A, issues 2 and 3 go to person B, issues 4 and 5 can be ignored that's not written down and causes the whole thing to go haywire. One would think that it would be trivial to record this but somehow it's often not done...
And that "soft" knowledge is indeed key. It's frustratingly intangible, seemingly trivial, yet incredibly powerful.
Constructing a nuke is easy. Accumulating sufficient fissile material without drawing unwanted attention is costly. Several millions USD.
Interesting article and reminds of this article https://danwang.co/how-technology-grows/ that also talks about process knowledge that is often hard to articulate, forget replicating.
I think we need to understand what is the base rate at which any group of smart people can generate great outcomes. While the freshly minted Phds did successfully build a nuke ( not at all an ordinary feat), would this same set of people been able to produce a similarly stellar outcome in their next assignment. One could argue, there is an element of randomness at play, that would make great outcomes hard to come by, even if we were able to master the art of replicating organizational culture. Perhaps, teams end up building bench strength with each new project, and eventually produce something remarkable that is much greater than the sum of their previous experiences/output.
Good point, and great article! I think its less about the randomness here and the possible reversion to the mean, than the fact that smart folks with higher levels of autonomy in their chosen path will tend to produce innovative outcomes, even when the odds are stacked against them. To industrialise it though requires a team, and we start getting into process knowledge.
Agree that over a sufficient period of time, smart( or reasonably intelligent) people will produce great outcomes.
I think this is true even in academia.. A lot of the well known university incubated companies, e.g. nicira, databricks etc. came into existence from a bunch of grad students working on slightly different areas sitting together in a lab..and the existing valley ecosystem that sort of had institutionalised the process of taking ideas from a research paper to billion dollar companies.