> The problem isn’t that enterprise software doesn’t work - it is that you didn’t make it work for you. The only measure that counts, hidden behind all those layers of abstraction, is the final outcome - decision advantage. This is often viewed as an implementation detail (the second 80%), and implementation is often outsourced to experts (COVID should have also taught us all something about “experts“). The industry of consultants who themselves have never run businesses but tell cargo cult companies how to execute implementation further obfuscates the lack of productivity gains.
This article is all over the place. It's written in a very confident and attention-grabbing style but lacks clarity and concreteness - especially so for an article criticizing abstraction!
The quote above is the closest I could find to an interesting thesis. I would restate the argument as:
1. SaaS companies optimize for revenue
2. The buyers of SaaS are undiscerning and will buy convoluted SaaS products even if the software doesn't actually improve worker productivity
3. Therefore economic growth has slowed because so much talent goes into SaaS which builds products that no one needs.
It's an interesting argument and I'm sure there's some truth to it. But surely the article would be better if it focused on supplying evidence for the incentive mismatch in #2 and showing that the magnitude is large enough to cause #3, rather than throwing out ambiguous and condescending claims about Covid, consultants, cargo cults. The core concept of "decision advantage" is not even defined, and seems to be borrowed from military terminology.
The core premise of this article is convincing: A lot of our software solutions are running to stand still. Occasionally I'll see a video from long, long ago (e.g. https://www.youtube.com/watch?v=ED3jbNCCjok from an episode aired in 1983) and I'll marvel at how little we've actually changed.
Having said that, it would be awesome if everyone stopped cargo culting on the backs of cargo cult analogies. The number of blog entries and articles describing South Pacific cargo cults, and using it to describe any and all follow-the-leader, copy-paste behaviours is...well it's painfully ironic.
Seems the words "rent seeking" should appear more in this discussion. "B2B SaaS" companies may well be cargo cults; but the blessing they're dancing to secure, the goal they seek, is "passive income".
> Satya goes on to argue that AI has the potential to restore productivity growth.
And Satya has a big investment he needs paid off. Had Satya's statements come from an economist or other it might be a little more believable.
> It is symptomatic of an engineer who doesn’t understand the underlying problem and tries to pattern match to get at the solution.
Pattern matching is absolutely the correct thing to do for most situations because it is rare that anything new is necessary. Almost all of engineering is a reflection of business processes and business has the same mode: a new business process is most of the time pieces of old processes rearranged.
> But when the abstraction isn’t tethered to empirical results or anything measurable and falsifiable, the abstraction is just an obfuscation, a cargo cult ritual.
That's not a problem with the abstraction that's a problem with observability. Abstractions are insanely useful both pragmatically and otherwise. O11y is often the last thing considered, if it's even implemented at all.
I'll also go on to say that it's interesting that the stagnation problem is framed in such a way. Software is a tool, not a product. It's going to take a product or class of products to fundamentally change the direction of growth and innovation. For all we know AI might even be it, contrary to the author's contrary.
I’ve been thinking about this a bit lately because I’m taking an economics class, but the pursuit of maximal profit doesn’t make as much sense to me as it used to. Specifically, what actually happens, in economic terms, to profit? Nothing good, it seems.
So maybe the “cargo cult” is the ruthless pursuit of profit, as opposed to regular reinvestment in the firm. Companies are incentivized to optimize productivity rather than grow it, as every dollar saved is a dollar to take out of the economic cycle and put into the investor’s pocket.
Didn’t corporate tax rates plummet appropriately at the same rate that computer technology exploded? That seems to me to be important, to the point that I’m not sure we’ve got the right culprit here.
Tech over-promised and under-delivered. The article presents a compelling argument that resonates deeply with the current state of the U.S. economy and its technological focus. Tyler Cowen's notion of the Great Stagnation, alongside the critique of modern tech's superficial advancements, underscores a critical misstep in our economic development strategy. The emphasis on software and tech, while neglecting the foundational sectors of manufacturing and engineering, has led us into a productivity paradox where technological advancements do not translate into real economic growth or improved living standards for the average citizen.
The analogy of cargo cult companies vividly illustrates the pitfalls of prioritizing abstractions and financial metrics over tangible solutions and innovations. This approach has not only stifled genuine innovation but has also diverted attention and resources away from sectors that are crucial for sustainable economic growth and competitiveness on the global stage.
> Satya Nadella at Davos in January 2024 said; “ ...inflation adjusted, there is no economic growth in the world, I would say and that's a pretty disappointing state. In fact, the developed world may have negative economic growth...PCs were the last time, when actual, economic growth came about, right? So, the last time it showed up in productivity stats were when PCs became ubiquitous.”
It feels good to hear someone up high say the quiet part I’ve suspected may be true out loud.
Starts with a rant against "cargo cult software" and ends with an endorsement of cult-of-personality management-by-CEO.
Perhaps the two are more closely related than OP realises?
I think a lot of this exogenous growth stuff is just mysticism for people eager for a secular religion.
https://www.employamerica.org/blog/it-wasnt-ai-how-fiscal-su... the endogenous growth story isn't as "fun" but much simpler: until we 'run out of workers', there is no reason to make anything more productive, and tech will be, at best, messing around with useless vanity stuff.
Our society isnt really structured to do things that require many years of research so were either making things incrementally better or just going after low hanging fruit. We need different incentives than profit.
Computers made everyone their own secretaries then gave them terrible tools to do it. So now people earning 5-10x the salary of secretary can spend 60% of their time maladministrating their work.
"Computers"
It’s not that productivity has stagnated. It is that much of our produce is now intangible. But we do put out terabytes of memes and Instagram pics and TikTok’s everyday. For good or bad that’s what people like nowadays, and that is where money goes to. If we count entertainment and bingeing as legitimate economic utilities, we are in the midst of the craziest economic boom ever.
Seems like quite a stretch to suggest broad, long-term economic stagnation is due to ineffective use of software.
It also doesn't really make sense that "great leaders" is a solution.
There's also a final recommendation to "exclusively work backwards from decision advantage" which might be good advice depending on what they actually mean (well, not the "exclusively" part, but I don't think that's meant literally), but not really to do with great leaders.
I think this must be a marketing piece by a management consultant (it's packed full of tropes and cliches)... and could kindof appeal to some C*O who is confounded by the technology their company uses and this could let them imagine themselves as a "great leader" by making more efficient use of software by "working backwards from decision advantage."
> (COVID should have also taught us all something about “experts“)
That wearing a mask and getting vaccinated was good advice?
That the internet is full of malicious people trying to discredit experts giving sensible advice just for their own profit and fame?
Maybe I'm remembering the pandemic differently than the author.
> To say the quiet part out loud: the OODA loop is all about the human. Technology serves humans. Instead, we have cargo cult companies such that the mere acquisition of architecturally-ordained technology is perceived to be the solution.
The first time I came across this basic thesis was back in maybe 2000ish? Or a few years later. There was a great article about an author who was whiteboarding out a solution, and the architect wrote this phrase, "Configuration data will be stored in XML" (the title of the article).
> This has always been true: The steam engine didn’t matter until it was put into a ship and locomotive
Sure, but had steam engines not been put over coal mines they wouldn’t have been refined to the point of propelling themselves [1]. It didn’t matter to contemporaneous productivity; it definitely mattered to it in the future.
[1] https://acoup.blog/2022/08/26/collections-why-no-roman-indus...
Sorry, this is too ambitious and overstated. A thesis about a four decade long, all reaching economics phenomenon needs much more research and evidence to be taken with any seriousness.
I'm with the author on the critical eschewing of abstraction, but I don't think that leads so clearly to the strong leader form. It suggests we need to be wiser technologists - understanding why we are making abstractions and what the downsides are and not simply ignoring dangers that are hard to address within our chosen abstraction. Organizationally that should mean more, not less independence for people at every level, who understand the boundaries where systems break down.
First question, where can we go from where we are? What do people desire that they can not have? I am not going to suggest that there is nothing left to be invented but I find it quite hard to come up with things that would have a significant impact. The one big thing that comes to mind is health - cancer treatment, organ replacement, and so on.
> Technology serves humans. Instead, we have cargo cult companies such that the mere acquisition of architecturally-ordained technology is perceived to be the solution
Sounds right on the money
And of course it has to be the latest bells and whistles without questioning why is it even there. Or how it can serve the company and not the way around.
>PCs were the last time, when actual, economic growth came about, right?
Uhh... smartphones? Shale oil? EVs? AI? Those alone are at a significantly higher (inflation adjusted) market cap than the entire S&P 500 in the 90s.
I don't buy this argument at all. We have seen massive real growth in the last 20 years.
Errr SaaS caused the great stagnation. Or non text UIs. Or consulting firms. I’m confused!
And this is just factually incorrect.
> Intel’s road to ruin was when the CFO became the CEO and managed the business by the numbers and abstractions.
I really could have used a concrete example here. Author seems like they have a very clear idea of what they're talking about but kind of failed to communicate the vision.
> Satya goes on to argue that AI has the potential to restore productivity growth.
And I would argue investment in infrastructure (I know, hand-wavy term) would do better.
The Great Stagnation is obviously a consequence of worker protection and environmental protection laws. Give me an exemption to both and I will turn this economy into a rocketship.
Everyone knows this is it. But everyone likes worker protections and environmental regulations so we pretend it’s not. But we know.
Remove those rules and TSMC’s AZ plant will be done yesterday and SpaceX’s Starship will fly more often.
We can dance around it but this is reality. All this other stuff is just window dressing.
This theme of progress has been on my mind for decades.
My conclusion, while simplistic, points to an unfortunate mismatch between (1) our personal/subjective expectations to “progress”, (2) business management dogma, and (3) the shift from “enabling-progress” to “distribution-progress”.
1. When I mean by subjective expectations, is that as the article refers to, many of us (techies), carry expectations to what progress must be like, the flying cars, spaceships and robot friends. These are images put into our minds, and some of them are based on pure fantasy and others, on former scientific trends, and their extrapolations. Others again, from former and current influencers, using them to manipulate us. An initial step to better grasp whether we are stagnating, is to look at the “possibility-space” between what is fundamentally possible within the principles of known science, and see where we are dragging our feet. If we are missing progress in obvious areas, it may come down to economics, but most likely is is due to the lack of transitional/incremental investment opportunities with a sufficiently small risk/capital step size to allow for progress (on this i (2) ). We are not likely to get anti-matter worm-drives as the next iteration from SpaceX’s Starship, the step-size is simply too risky even if the physical principles are understood.
2. Since the mid-70s, the general business trend has been to take all business decisions based on a NPV/ROI, hereby assuming that we have sufficient knowledge about the outcome of an effort to estimate whether it is worth doing. Likewise, the frameworks of risk management are seldom able to manage uncertainty in imagination (i.e. the “unknown-unknowns”), forcing a decision bias towards preferring investments with known risks, that can be managed. Combining these management practices makes it basically impossible to allocate capital towards fundamental R&D, with no certain outcomes, and where the ROI can have an asymmetric upside, by opening up a completely new paradigm for business and progress. On top of this, “Venture” Capital, is not and have never been free from this type of management, the ventures are known (whaling, SoMe, or SaaS) the risk taken is that of successfully capturing market. They are not the vessels of progress, but distribution. We could also put our faith in the academic institutions, and government labs to bring about the future. But they too, have largely been funded according to the same model, but with a different incentive structure (think grants, citations, recognition) that also values controlled risk and predictable outcomes. As an aside, I remember one of the most productive professors at my old institute also said to apply for research funding for the already concluded project, and then use the money for the next project. That way he always had a perfect success rate, and could match the research project to budget, time and scope.
3. When I mean with the two types of progress, is that we perceive the invention of telephony as progress, in the systemic sense, now “it is”, but having a telephone in every home, that availability is certainly also progress, as it makes it useful “I have it”, and doubly so for network-effect products, where “you’ll have it too” matters, like the phone.
Of these types of progress, “enabling” naturally most precede “distribution”, but as a society we can gain great wealth from the latter, ignoring the first for quite a long time, before we realize something is “missing”. I think that is what has happened from the factors in (3) for the last 4-5 decades. Computers and IT have driven a massive distribution-progress, blinding us to the fact that enabling-progress was dwindling. And now, when the economy can’t hide the fact much longer, we start to discover that lack of “progress” in the subjective sense.
So, if it's so bad, the more productive companies will take over - like Costco.
I denigrate the term SaaS. Do any other industries have this weird acronym for what they innovate on? It can even be called nBA if acronyms had to be used for everything( "no Brick Apps")! Imagine how SaaS sounds for somebody not in tech! And how arrogant of Satya to just completely wipe out smartphones as an economic growth engine , his mind is stuck in the 70s! I would say technology has always followed from institutions with heavy funding, who had the time and money to do this kind of innovations. If there is no incentive for those institutions to do those innovations, say, for example DARPA invents a neuralink like machine for helping the soldiers be at peak performance, but turned out it did not really help but hurt them more, they would not do so , and so, the landscape for the tech industry to imitate and develop a prototype off that tech is non-existent. Elon musk tried to change the path away from relying on those institutions, but it is just not practical to do so in terms of time and money for a publicly funded company in the capitalistic RoI hungry stock market. One would argue that even the funding that institutions get is from tax money, but once that money is gone, it is gone, they aint asking for interest and principal on that money! May be it is time for the stock market to adapt and create such an environment where interested public can give away some portion of their money , like crowd funding, and they choose which individual be eligible for using that money to advance innovation, similar to democracy but this time, with peer reviews from experts in the field! This would work for the economic growth as a whole because it is not individualistic but everyone is working towards the goal, similar to what John Nash says in the beautiful mind, the best result comes in a group when everyone does what's best for himself and the group! Adam smith was wrong!
Not sure how apt the cargo cult metaphor is in this case, but the article seems to strike in the general target area of our current malaise.
Interesting observation about a complete lack of economic growth (as corrected for inflation), but I'm not convinced that the Service as a Service industry is to blame; I think it's more of a symptom than the cause.
In this (some would say terminal) stage of capitalism in the West, we no longer focus on creating value -- the main objective seems to be extracting it instead. If every organization's goal is to turn their capital into more capital, what they actually do to achieve it is incidental. One could argue that a company of five thousand paper pushers milling around performing ineffective rituals is at least not causing any harm on balance (as compared to, say, arms manufacturers). Nevertheless, the current trend is to reap everything and sow nothing - somehow without the foresight to acknowledge the absolutely desolate desert this is creating.
One observation that caught my attention was the mention of the OODA loop. Access to information is a prerequisite for power - even if you don't have leadership privileges, being well informed empowers individual decision-making. All this corporate obfuscation disempowers people at all layers of the hierarchy. It makes for really ineffective organizations - sadly, this is absolutely acceptable now, because they don't need to be effective at anything other than temporarily grouping resources as an entity in the big consolidation shell game.
This has a lot to do with the scale of an organization, I think once a company grows beyond double digit headcount, things go downhill [1]
1: https://news.ycombinator.com/item?id=39172505 (I guess I like to harp on this subject)
I also have an ECE degree and came to similar conclusions in 2001 after the Dot Bomb and 9/11. I've been awake to the eventualities we're facing today around becoming slaves to technology (that widens wealth inequality) for basically my whole career.
But I disagree on one point: abstractions are not the problem. The lack of abstraction is what leads to opinionated frameworks like Ruby on Rails that obfuscate what's going on through magic like syntactic sugar without formally documenting the underlying abstractions. I work mostly in Laravel, which suffers from similar issues, but thankfully admits its flaws because it's written shamelessly in PHP. Frontend frameworks seem to mostly require the developer to drink the kool-aid because they don't even mention the determinism issues inherent to Javascript's async model and the various syntax hacks to emulate classes that diverge from its original conceptual simplicity. The same thing happened to PHP 5 when design-by-committee pass-by-reference classes were added, which broke its copy-on-write semantics and made it just like every other object-oriented frankenstein language.
No, the answer is to flip systems of control on their head and start talking to people at the bottom.
For example, Apple may never fix its external display support. I can't rely on my MacBook to wake up monitors, so sometimes it takes several minutes of plug and pray and even having to restart to get them to turn on. Or the other day, I discovered that mouse keys can't be enabled if the trackpad clicker is broken and tap to click is disabled. When you push command-option-(fn?)-f5 (a horrific keyboard shortcut), it brings up the modal to enable mouse keys, and you can tab through the options in an attempt to click the checkbox, but pressing return just closes the window. So a mouse is required to bypass the mouse. An astonishing oversight that was never fixed FOR YEARS.
At modern tech companies, there's nobody at the wheel, and there are no adults in the room. Yet FAANG has almost $1 trillion PER COMPANY. We get so mad at the government for squandering its trillions of our tax dollars on maintaining colonialism around the world, but meanwhile tech companies have vacuumed up all available capital so that we subsist on driving Uber to make rent.
I mean, I'm comfortable calling this. I think it's over, and we're into clear and present danger scenarios. I'm just so tired.
want to see an explanation that perfectly aligns? https://wtfhappenedin1971.com/
Abstractions obscure connections between hidden information.
I think Tech has nothing to do with it, it's just neoliberalism and financialization of the economy.
Basically the idea that your problems are your own, and the society (or government) has no responsibility to help people who are struggling for some reason.
This coupled with the ability to put debt burden onto people, one way or another, which enabled lot of rent seeking in the economy.
It start with us, the people
Today the 2 most popular articles are about javascript tools to generate a color palette..
I mean, aren't there more interesting things to talk about?
I submitted this, it got ignored: [1] (Chinese scientists make breakthrough in BCI-assisted rehabilitation trial, 'showing higher safety than Musk's Telepathy')
Transhumanism, gene selection/editing should be on everyone's mind, it'll shape the next civilization
The "great stagnation" is a political choice, it leads to interesting stuff happening elsewhere, while your group is left eating the junk food, it's an interesting paradox, almost like a bubble or a jail
at my place of work, we have a lot of files (in the business sense, not the computer sense), and we need to manage them. we used to use this piece of software that we paid another company to host and manage, and it was a little clunky-looking because it was all written in Java, but it worked great. the software did everything we wanted it to, from importing scanned documents, to allowing us to make forms that can flow through workflows we defined. it was great, except that it required the use of Internet Explorer and Java, but we stuck with it... until the company who made the software was acquired by another company, which is making a similar kind of software, except browser-based... and a billion times more terrible.
this new software can be self-hosted or hosted on the company's servers, and we opted to try the self-hosted option out. it's been a complete nightmare. users have to wait 30 seconds or more after clicking "Submit" on a form before it gets submitted. form flows regularly get hung up or vanish completely.
right now, the software is using 24.2GB of RAM in IIS, and another 26.8GB for SQL Server, despite being used by only a few dozen users simultaneously—there is absolutely no excuse for this, in anything resembling a sane universe... but this is what we're stuck with for the time being. it's horrible. and the company who makes the software suggests we would get better performance if we went with their hosted option instead of self-hosting—as if that could somehow magically make this piece of utter shit run any better at all whatsoever.
but this is where things are right now with software: engineers who have no idea how to engineer things are building layers and layers of abstractions atop each other without giving a single shit about how well it all runs, or how many resources it consumes, or whether the entire functionality of the software you're building could be expressed several order of magnitude more simply by merely stepping back and reevaluating intended functionality and expressing it in code more directly than a billion layers of C# abstractions heaped on top of each other.
the funny (but sad) part is, many users, including some of my coworkers, think this is all completely normal, that such simple functionality must be difficult enough for modern computers to run, because their expectations have been set so low by contemporary software. people just don't care if a form takes 30 seconds to submit—maybe if it takes a full minute, then something might be wrong, but 30 seconds? to submit a form to an intranet web server? perfectly acceptable, why would that be a problem, they say.
I wish I could be more optimistic about the future of software but at this point it looks like we're kinda just fucked—nobody cares about making good software anymore, and most people have no idea how good, fast, and useful software could be if it was just written a slightly different way than we're taught in college.
[dead]
The Costco screenshot is really interesting. The terminal apps of old were truly works of art and _incredibly_ fast. A non-technical worker wouldn't take long to understand the system and the keys/shortcuts to do something quickly. I remember having to sit with a few folks when we were looking at modernizing an app, watching them with a few key strokes process a record or something and thinking "we'll never match this speed".
Web or even desktop apps these days really pale in comparison.