s09e29: New clouds; Spreadsheets; Half-lives; Snow Crashing chapters 17 and 18
0.0 Context setting
Monday May 17, 2021. It’s not hot and sunny anymore. It’s kind of cold and cloudy. But! I’m going to be two weeks vaxxed on Thursday and a new life will await me in the non-masked colonies.
In this episode:
- Three things that caught my attention (clouds, spreadsheets and half-lives)
- Snow Crashing: chapters 17 and 18
On with the show:
1.0 Some things that caught my attention
“New cloud formation just dropped”
Not that kind of cloud. Via Deb Chachra on Twitter, the Washington Post’s Capital Weather Gang published a photo taken by James Montanus of “newest official cloud type” Asperitas over Lake Ontario.
Caught my attention because: it’s beautiful, I learned about the World Meteorological Organization’s International Cloud Atlas, and photography. Also makes me think of weather forecasting, which is very computer.
Spreadsheets
A few things to do with spreadsheets. Apart from Excel probably being one of the most-used programming IDEs in the world and an example of how people can just understand something like MapReduce1 if they just do it (being: when you have a bunch of data and then you do the same operation to it, like say a whole column of one number and then you want to apply a formula to it, and then you do the draggy-handle-thing and it all dynamically updates whenever you change a value in that first column), here’s another thing about spreadsheets that will probably horrify some of you and also hopefully engender some curiosity.
It is this: A case for Spreadsheet Wireframes by Clair Rock, and unfortunately on Medium which means maybe you need to subscribe to read it? I don’t know.
Look, really. The whole thing is about using Google Sheets (a collaborative working environment! That allows multiple users to work together in real-time!) to do IA, design and content creation seamlessly. Go read it.
Caught my attention because: Really. SPREADSHEETS. This is interesting because a) it’s horrifying; b) beautiful; c) usable; d) also still horrifying; d) yet another reason to use that Jeff Goldblum quote about life from Jurassic Park, but mangle it to “spreadsheets find a way”.
Organizational half-life and technical debt
A super interesting thread from Will Demaine at Monzo about engineering organization half-life. I will try to summarize:
- You have to make decisions about technology all the time. Sometimes you will need to make a decision about how to do something for which there is not a commodity or generally accepted approach.
- This means you could choose the “wrong thing”.
- “Wrong thing” might be something that totally works and is fine, but is now idiosyncratic and not how everyone else does it. Which is still fine because it works, but is increasingly out-of-phase with “how everyone else does it”.
- By “out of phase”, I mean: you’re increasingly going down a specialized path for something that works but also has increasing maintenance cost.
- Demaine’s example here is where Monzo “uses a custom RPC framework called Typhon [but] other, off-the-shelf, arguably better systems exist now which didn’t exist when Typhon was built”.
- In this particular instance, Typhon does a normal thing. It’s not a competitive thing. It’s not a secret sauce thing. So Demaine says: “If you gain no advantage from your custom solution, you’re effectively continually taking on debt by not swapping it out. The longer you wait to adopt, the more your paths diverge and the harder the migration becomes.” [Tweet]
- So Demaine points out: you wouldn’t do it this way if you were starting now. But also remember: you didn’t have that information back then. Going custom was reasonable!
- Demaine has a non-Monzo example: “Uber initially settled on Thrift and Mesos, which subsequently got drowned by gRPC and Kubernetes respectively!” [Tweet]
Now the interesting bit. All this prior is just your regular “this is how we make software and sometimes we have to make a decision about what software to use when it’s early, or whether to write our own. Then we have to make a decision about whether to keep using it.”
So far so regular technical debt.
Demaine’s interesting way of looking at this is to use your organization’s turnover/staff attrition as information to help you make the decision about what to do about that technical debt.
So, he says: normally you might make a decision about what to do with technical debt (e.g. your custom implementation that now has a commodity/standard implementation) based on your current technical capability and organization. But… what if you made that based on your future capability?
Demaine says if you’re growing, then soon the number of people who know and are familiar with the in-house implementation are going to be outnumbered by the people who don’t and are newer. So: how long will that take? Demaine again:
With 250 employees, a 30% yearly growth rate and 10% attrition, it only takes slightly over 2 years for 50% of the engineers in the org to have zero context on the original decision.
The numbers for 100 engineers, 20% growth and 10% attrition, that comes out to 3 years.
So! Now you have another input to “what should we do about this tech debt, housekeeping or potential migration”: how long will it take for it to become mythological? Again: super interesting.
Caught my attention because: Really, thinking about the human factor of organizational knowledge and learning (whether through growth or just attrition) and how that factors into planning software is super interesting and this feels like it can lead to better (if not more informed or better considered) decisions about planning.
2.0 Snow Crashing, chapters 17 and 18
Chapters 17 and 18 are our first real introduction to Nova Sicila through the stupendous fuckup of Jason Breckinridge, baby first-time owner of a Nova Sicilia franchise.
Jason is the franchise owner of Nova Sicilia #5328 in the Valley. Jason’s fuckup is a classic case of being over-eager to impress the boss with some personal service and in doing so, not reading the instructions properly.
But I’m less here for the commentary on race relations (the Mafia is big on the melting pot and respecting its members’ cultural identity—and yet also happy to portray other franchulates like the Narcolombians as uzi-toting killers, rather Trump-style).
I’m not here for the values of a legalized/decriminalized Mafia operation (big on personal service, of course), either.
Or for the Mafia approach to business competition (Jason’s business school thesis was titled “The Interaction of the Ethnographic, Financial and Paramilitary Dimensions of Competition in Certain Markets”).
No, I’m here for how the Mafia uses tech. Here’s what we learn:
The Mafia runs a bank (therefore: most other franchulates should operate banks, too, right?) called the Goombata Point Bank. It also runs a bit like airmiles or a rewards program, so business as usual for a 1992 extrapolation. Opening an account gets you 3,333 points, becoming a citizen gets you 500 points and when you hit 10,000 Goombata Points you get a free cruise to Sicily. We’re just a tiny bit away from some nascent gamification of Mafia activity!
The important thing to know about Goombata Point Bank is that the balance of accounts “is stored in the big computer in Brooklyn” which again is a reminder that:
- There are no decentralized ledgers in Snow Crash, which means that we can’t blame “Snow Crash is a business plan” for cryptocurrencies; and
- There aren’t really decentralized computers in general, you can pretty much assume that “the big computer in Brooklyn” is your regulation-standard mainframe and that it’s probably dealing with a bunch of technical debt.
That’s banking. The second is that this intro to Nova Sicilia is one of only four occurances of the word cellular in the entire book (the others being two references to Hiro’s computer having a cellular network link and one reference to the Raft making satellite-uplinked network access available over a Raft-wide cellular network).
So, “cellular”. There’s a big brouhaha over Jason’s thesis in that it calls out the bad performance of a failing Narcolombia franchise. Somehow, the owner of that franchise found out (how, exactly? It kind of feels like Jason’s thesis would’ve been printed out and submitted on paper) and decided to take umbrage through the form of a firebombing and drive-by shooting. But this was averted! Because the local Nova Sicila franchulate
got wind of these attacks before they happened, probably by intercepting signal intelligence from Mr. Cortazar’s fleet of poorly secure cellular phones and CB radios.
So it’s nice to see the Bad OpSec called out. Now I’m imagining the existence of a Nova Sicilia NSA providing SIGINT2 (I assume the Mafia is good at HUMINT3) with everything from IMSI catchers4 to a nice cache of zero-day vulnerabilities. Also, they use CB radios!
The net result of this use of SIGINT is that Jason is grateful to Nova Sicilia for preventing his family from being a) bombed and b) shot full of holes, and ends up with a smart Teracotta blazer, secure in the assurance from Mr. Caruso that with “a three-point grade average… you’re gonna kick butt, son!”
Now Nova Sicilia, being a complex, sprawling organization probably has an ERP system5 which would do all the back-office stuff of running a company. The helpful Wikipedia pseudo-donut says that ERP systems covers things like:
- procurement
- production (“product lifecycle management”)
- distribution or supply chain management (I imagine quite important to the Mafia)
- accounting (ditto, really)
- human resources (really ditto)
- corporate performance and governance (I mean yes, I suppose there is a focus on performance and how that performance is governed)
- business intelligence (see SIGINT above)
- enterprise asset management (gotta keep track of all those… cars?), and
- customer relationship management (really, is ERP software just made for mafia organizations?)
Nova Sicilia’s system is called Turfnet. We learn that Turfnet includes GIS because it has to keep track of all that territory:
Turfnet, the multiple listing service the Nova Sicilia used to keep track of what it called “opportunity zones”
The listing service part of Turfnet works in real-time I guess, because Caruso is able to pound away at the keyboard and wheel a monitor around
to show Jason a map of the L.A> area blazing with red splotches that represented unclaimed Turf sectors.
So, you know. A Zillow or whatever for Mafia franchise locations.
These days the Mafia might be concerned with being usurped by a franchise that has up-to-date franchise information rather than what sounds like an old, in-house application. I bet Mr. Lee’s Greater Hong Kong (Jason’s other choice to join “because I’m real interested in high tech”).
Fortunately in 2021, there are franchise territory mapping exists as a SaaS! SalesRabbit for example says that “territory management is a big part of successfully operating a sales team”, Geopointe for Salesforce will “streamline territory management and design… then use geographic and data insight to fine-tune and manage them” and, well, there’s more.
The other thing Turfnet does is provide a list of jobs-to-be-done. Jason’s job as a manager is “to portion work out to local contractors”, so his regular morning routine involves tidying up his lot, then :
safely inside, Jason signs onto Turfnet. A job list scrolls automatically onto the screen. All jason has to do is find contractors to handle all of these jobs before he goes home that night, or else he has to take care of them himself. One way or another, they have to get done.
On this particular morning
[Jason] signs onto Turfnet. A list of jobs scrolls up as usual, not a very long list. Interfranchise activity is way down today, as all the local managers gird, polish and inspect for the possible arrival of Uncle Enzo. But one of the jobs scrolls up in red letters, a priority job.
So, Turfnet also includes case management and workflow management. A bit of ServiceNow, then. The case management part (combined with the customer relationship part, I guess) probably features when Jason does collections from delinquent borrowers. Got to keep all those notes to make sure he can
emphasize that his organization takes a personal, one-to-one, hands-on, micromanaged approach to debt-related issues.
For delivery jobs in Turfnet, Jason portions them out to Kouriers, and it’s not made clear whether Jason (or Nova Sicilia in general) have a supplier agreement with RadiKS (Radikal Kourier Systems), or with individual Kourier… Kontractors (sorry). One would assume that with an organization like the Mafia and an emphasis on the personal, they like to make sure they can request a particular Kourier (which is what turns out to be the case here) on an individual contractor basis. So complicated! No wonder they have an enterprise that needs its resources planned.
I wonder what kind of business intelligence Turfnet provides. There’s the observation that “interfranchise activity is way down today”, but maybe that’s just inferred because of the lower than usual number of jobs.
Turfnet also probably includes some sort of messaging system, perhaps some sort of “electronic mail” that explains how Jason knows Uncle Enzo, the head of the Nova Sicilia, will be in town today.
These jobs have a bunch of metadata attached to them: their priority and a field for more detailed instructions or notes.
How do we know? Because of Jason’s Big Mistake. See, there’s a high priority job (not a Code H, which helpfully shows that the Mafia have their own Job Codes, which is… wonderful), and Jason Didn’t Read The Instructions, which leads us to our last bit of Mafia use of technology.
(The instructions are for Jason to make sure that Y.T. will deliver the documents. The point wasn’t the documents, the point was to engineer a meeting between Y.T. and Uncle Enzo).
Jason’s big mistake is to glom on to the important document delivery and to do it himself. We know this because:
- Jason has printed out the job request (there are printers! Of course there are: how do you fax things if there aren’t printers? And secondly, do people still have problems with printers in Snow Crash?) because he hands a hard copy to the guard; and
- The job instruction is referred to as “the fucking job sheet”, which is the thing that Jason did not read. Having Y.T. make the delivery was something “the job sheet explicitly tells you not to do.”
Which, my friends, is my final piece of concluding evidence that the Mafia is a legacy institution still dealing with a transition from a paper-based bureaucracy.
Anyway: the drive to Compton is an opportunity for Stephenson to inject quite a bit of social, racial commentary.
One mention of technology we have is that Narcolombia, a rival franchulate, has owners who’re tempted to skim every now and then, which is a bad idea because doing so likely results in grievous performance measures being enacted. Those grievous performance measures normally happen after “flying justice squads… fax your records back to the notoriously picky computer in Medellin”, which is just another example of a) faxing (again, one of only four mentions of faxing), and b) centralized computing.
Anyway. Jason drives through Compton. It’s not nice and he doesn’t enjoy it. The last piece of technology we see the Mafia use is a retina scanner, when Jason approaches the checkpoint for the franchulate that Uncle Enzo is visiting. (The checkpoint is very well protected with a portable claymore mine):
A top-ecehlon Mafia guardsman nails him with the retinal scanner. None of this ID card nonsense. They know who he is in a microsecond.
A microsecond is fast! It takes me 41 milliseconds to ping Google from here, but a microsecond is 1,000 milliseconds. I mean, I’m assuming that all the retinal data that’s needed is cached locally and that the data captured by the scan just now isn’t, for example, compared against master records at that big computer back in Brooklyn. I mean, I don’t think TouchID works that fast, nor FaceID. Hell, even just letting me know I’ve typed the wrong password into my laptop isn’t that fast.
… and that’s the end of chapters 17 and 18: Jason has made a Big Mistake and now we want to know Uncle Enzo wants to meet Y.T.
That’s it for today.
How have you been?
Best,
Dan
-
Signals intelligence: NSA: What We Do: Signals Intelligence, and signals intelligence features as, well, a key point in Stephenson’s later book, Cryptonomicon. ↩
-
Human intelligence: covered in the Office of the Director of National Intelligence’s helpful article, “What is Intelligence?” which to be honest is a very deep question. ↩
-
An International mobile subscriber identity-catcher, which you should definitely not make using this guide from Vice from 2018, “With $20 of Gear from Amazon, Nearly Anyone Can Make This IMSI-Catcher in 30 Minutes” ↩
-
Enterprise Resource Planning, Wikipedia ↩