There's a very interesting video about the "1.5 language problem" in Julia [0]. The point being that when you write high-performance Julia it ends up looking nothing like "standard" Julia.
It seems like it's just extremely difficult to give fine-grained control over the metal while having an easy, ergonomic language that lets you just get on with your tasks.
"Collaborate by Contract" (CBC) is my framework; it's new, and I'm publishing it piece by piece. It's not an "industry standard" yet, but it's more than theory: it's how I've learned to get execution discipline in teams where vague goals and shifting priorities are the default. It doesn't need to be practiced by the entire org; as with any other type of agreement, you just need two willing participants.
At the simplest level:
A CBC agreement is a short written contract between leaders and reports. Ideally, person to person, but leader and team is fine too. It defines the objective, deliverables, dependencies, expectations, and outcomes (i.e., success criteria). Work doesn't start until all parties agree.
A journal of contracts is just the running log of those agreements. Think of it as the team/dept/org's public ledger, what was agreed, by whom, when, and why, and ideally the artifact captures the negotiations and tradeoffs made to arrive at the final agreement. Patterns show up quickly (who delivers, who misses, where scope creep hides). This makes performance reviews objective and enables meritocracy.
Okay, things change, and sometimes new info emerges: agreements aren't stone tablets. So, agreements might include predefined checkpoints or "if/then" clauses. Instead of pretending we never change, CBC forces us to renegotiate in daylight, with both sides explicit about the cost of changing.
What leaders commit to: clarity, timely decisions, and removing dependencies. If a leader misses their side, say they don't secure the promised resource or they blow their own deadline, that's a contract miss just like an IC failing to deliver code. Also, by signing the agreement, the leader is sayin,g "I agree with this plan/strategy." In CBC, credibility runs both ways.
It's early, and I'm still publishing examples and tooling. But the premise is simple: if you can't write it down, negotiate it, and sign it, it's not really a commitment, it's just vibes.
Also, no, it's not Waterfall, because it's more about documenting "expectations" and "outcomes" than about specifying particular work activities.
I'm curious about it and your thinking on how to track things over time and see what has surprised us since we got started. It is useful to note down every time you (or your team) sets an expectation with someone (or another team) and then make sure you don't forget about that. It's also useful to be deliberate when setting expectations.
Having a public journal could well work for noting down when expectations are set and whenever there is a meeting of minds. I've found when tracking things like this that the amount of data can quickly grow to the point where you can no longer quickly and easily reason about it. The success seems to live and die on the data visualization or UI/UX.
Ok, I'll bite. From the article I can't really figure out what collaborating by contract (CBC) is, how it works in practice or how to introduce it to an organization.
A search in Google for "Collaborate by contract" gives three results, all from the same person, all in the last few weeks. Including this new article it's 1776 words in total on CBC. It doesn't seem to be real or something that has been tried out in an organization. It appears to be Al Newkirk's idea for a system that could work, but has not been tried.
Specifically, I'd like to see an example of a contract and who agrees to it; what the journal of contracts looks like; what happens when after an agreement everyone learns something that they didn't know when the agreements were made; what are the leaders committing to and what happens when they fail to deliver that?
Many teams have working agreements; and companies have employee handbooks.
I don’t know if you ever read these in detail, but they’re generally in one direction. Other than dating/relationships (manager and direct reports, etc) and some generally applicable guidelines, it’s favorable to management.
One thing I make very clear to my direct reports is that I expect them to hold me to account when I fail to do something or hinder the team; even going above me if needed.
But this is ad-hoc. It’s not consistent across the board, and I see managers who are active hindrances to their team or their mission.
This is also the norm in many companies, and it’s a problem.
Computers can already do a lot of things that no human can though. They can reliably find the best chess or go move better than a human.
It's conceivable (though not likely) that given training enough training in symbolic mathematics and some experimental data, an LLM-style AI could figure out a neat reconciliation of the two theories. I wouldn't say that makes it AGI though. You could achieve that unification with an AI that was limted to mathematics rather than being something that can function in many domains like a human can.
Wouldn't this unification need to be backed by empirical data? Let's say the AI discovers the two theories can be unified using let's say some configuration 8 spatial dimensions and 2 time dimensions. Neat trick, but how do we know the world actually has those dimensions?
Roaming a labyrinth and savaging young Athenians might seem like a positive change in the short term, but ultimately it’s probably just as unfulfilling as corporate IT.
The mentoring was amazing when I started, but unfortunately the company was bought/sold and things unfortunately went downhill from there (corporate profits rarely coincide with providing actual help and is more designed to make paperwork/justify spending the precious numbers. I was great at the actual job, sucked at the medical paperwork)
I am thrilled I got to help some kids in need of help/understanding/acceptance at least and seeing the joy/results first hand is something I will never forget
Do you have any more universal stack ranking / ruling out re: career paths? Bonus points for more epic phrasing: it didn't really hide that the subjective opinion is objective fact, but, I think there's a better chance I'm fooled into living someone else's life if you do it again.
IMO that's the fundamental difference between statistics and ML. The culture of stats is about fitting a model and interpreting the fit, while the culture of ML is to treat the model as a black box.
That's one of the reasons that multicollinearity is seen as a big deal by statisticians, but ML practitioners couldn't give a hoot.
You are describing the difference between academic mathematician statisticians and "applied/engineering/actuarial/business" people who use statistics. The "black box" culture goes back to before ML and before both computing Machines M and statistical Learning (iterative models)
I suspect that the "black box" philosophy for statistics/ML is actually bad if you don't have a quick way of verifying the predictions. For instance, using PCA as a "black box" is perfectly fine if you're using it to de-noise readings from a camera or other instrument, because a human being can quickly tell if the de-noising is working correctly or not. But if you're using PCA to make novel discoveries, where you don't have an independent way of checking those discoveries, then it might be outright essential to have a deep definition-theorem-proof style understanding of PCA. What do people think of this hunch?
The point about PCA applies to population genetics and psychometrics (IQ). Some conclusions have been derived using PCA that appear to be supported by little else, and these have come under question.
You make a good point, though the difference between ML and statistics isn't just about interpreting and validating the model. It's about the "novel discoveries" part aka Doing Science.
Statistical modeling is done primarily in service of scientific discovery--for the purpose of making an inference (population estimate from a sample) or a comparison to test a hypothesis derived from a theoretical causal model of a real-world process before viewing data. The parameters of a model are interpreted because they represent an estimate of a treatment effect of some intervention.
Methods like PCA can be part of that modeling process either way, but analyzing and fitting models to data to mine it for patterns without an a priori hypothesis is not science.
Only perfect multicollinearity (correlation of 1.0 or -1.0) is a problem at the linear algebra level when fitting a statistical model.
But theoretically speaking, in a scientific context, why would you want to fit an explanatory model that includes multiple highly (but not perfectly) correlated independent variables?
It shouldn't be an accident. Usually it's because you've intentionally taken multiple proxy measurements of the same theoretical latent variable and you want to reduce measurement error. So that becomes a part of your measurement and modeling strategy.
Simultaneously correct and devastating. Where I work, senior leadership don’t have a technical background and so don’t know when something is “good enough“.
The result is that most of our products are dogshit and are force fed to users within the company and to its clients.
That’s fair, similar point. Reasonable people can both appreciate the irony and feel bad for the victim. Few people IRL have the sociopathic “they had it coming so I just want to watch them bleed out” attitude that we see so much of online.
IMO your logic is all wrong. Comparative Advantage ist just applied "opportunity cost" of time. Humans and resources are unique, everyone has their theoretically "optimal" use of time in terms of economic output.
The invisible hand of the market will let you know what aspect of your output is most valuable for others.
The benefit of this invisible hand is that the "economy" as a whole does not need to know how good they are at producing everything. People just need to know if what they are producing now is more valuable than the next best alternative. Everything else will be sorted out with market forces.
In university lectures we were given the famous argument about olive oil from Greece and that it would never make sense to do our own olive oil because we both lack the natural resources (unique soil + sunshine) which allow olive trees to grow easily and we'd also have much better yields growing other things on the fields.
So to me, both opportunity cost and comparative advantage are really basic building blocks of economic understanding and I'm a bit dumbfounded that someone wouldn't understand these concepts.
It is good that you paid attention to economics 101 but we don't live in the 19th century anymore and economic theory has progressed a bit since Ricardo.
We don't have pure free market economies. Neither in China nor in the USA nor anywhere else. The see big monopolistic companies dominating most markets. We see an closer interlink between state and private corporations.
Even just with the currency manipulation that China engages in, things get screwed a lot. Or the special status the US has with the dollar. Real world is more complicated.
But even if we assume free markets, you misunderstood what the previous poster said. The problem with Ricardo's comparative advantages is that is assumes fixed advantages. It is like optimizing for a local optimum. You might be super inefficient in producing X because you have never done it but if you actually invested in learning how to produce X you might discover that you are really good at it and the comparative advantages would go in your favor.
I do still believe that trading with each others can lead to more net wealth in most cases and obviously full autarky is not realistic these days but like anything in economics, it shouldn't be taken as a dogma.
Absolutely agree. It's ridiculous that low wage labor is considered a "comparative advantage". It's an advantage to capital owners perhaps, but certainly not to workers. And like you said, advantages are not static.
In my opinion it's intrinsically valuable to have a diverse regional economy. Culture and economy are fundamentally inseparable, imagine a society where everyone is doing the same thing because of "comparative advantage" making them 10% more efficient than the other country... What poverty!
This aestheticization of factory jobs is something I've noticed to be driving the New Right's worldview. It's not dissimilar to and no less dangerous than the aesthetic fixation on the agrarian economy of Mao and Pol Pot.
Frankly, no, sweatshops are not important to the cultural fabric of a country.
The US has problems with housing affordability, with medical costs, and with service sector costs emerging from Baumol's cost disease, which are all things that will get worse with tariffs, ranging from higher construction costs, to higher pharmaceutical prices, to less service employees making the cost disease worse.
It's also untrue that comparative advantage only benefits capital. Consumers are hurt by higher prices and less job opportunities driving down demand on the labor market. This worldview of a zero sum contest between capital and labor is a populist fiction.
Manufacturing doesn’t have to equate to sweat shops. It’s hard to take your argument seriously when your judgement is undermined by such fallacy.
We have problems with housing affordability because asset values inflate inverse to the devaluation of the dollar. The dollar is deflating because a service economy is not as sustainable as a manufacturing economy. This is particularly pronounced when we all see the labor value of intelligent workers decreasing at a precipitous rate due to AI.
Tech bros who are frustrated with their job fantasizing about doing "real work".
An entire generation has grown up without assembly lines so it is easy to mystify it.
People in Vietnam don't enjoy making Nikes but it is better than what came before: subsistence farming.
But the Vietnamese factory worker trying to send their kids to university too.
Lots of people remember the 80s and 90s being better times with quality manufacturing employment without romanticizing the past. To this day multiples of the “information” sector are employed in US manufacturing.
We can agree unions should be stronger, but union jobs in America cannot compete with nonunion much cheaper labor in other countries. If you have free trade and zero Republicans the same thing happens. If the jobs go away the union doesn’t matter. That’s why the unions consistently lobbied against NAFTA, the WTO, etc.
I’m actually not even sure what specific labor law changes you could blame that on. Clinton was running the show in the 90s, and I don’t recall any big union busting under Bush, whatever else might be said of him.
> We can agree unions should be stronger, but union jobs in America cannot compete with nonunion much cheaper labor in other countries.
I mean, they can, if you put up trade barriers or introduce capital controls. It's not a coincidence that after capital controls were removed, basically any manufacturing that could, fled America. And I (and my family) in Ireland were massive, massive beneficiaries of this!
Like, you can definitely make the argument that globalisation has benefited the world overall, while being bad for a bunch of people in the developed countries. And it's not a bad argument.
But unfortunately for all of the people who think globalisation is great, the votes of all the people who disagree count just as much as yours, and it looks like they're willing to vote for anyone who even hints at promising to fix this.
> Clinton was running the show in the 90s,
He introduced NAFTA, which made it profitable for much US manufacturing to move to Canada/Mexico. Bush let China into the WTO (or was that Clinton too?).
Perhaps this is the inevitable cycle of prosperity? We see this in so many facets now as generations progress - your comment reminds me of antivax social media people who haven't ever seen anyone more sick than a cold or tech bros thinking a trade job would be better since it might magically be "more rewarding" (I'm guilty of this!) with no regard for how much privilege is inherent in sitting at a desk all day and getting paid to think.
Like the stereotypical kid who grew up rich not understanding the value of hard work maybe the inevitable result of easy and safe living is a blind spot so big we're doomed to fall back down as a society and start over again and again.
Sure, anyone not agreeing perfectly with the current system of global trade is part of the "new right"... Another way to look at it: globalisation weakens democratic control over the economy and undermines unions. Is that not a problem in your opinion?
Globalisation Also creates markets for the more advanced goods and services to be sold.
If we are going to wade into the deep waters of international trade, then you can’t look only at america or American workers without getting blind sided constantly.
At the depth you are talking - globalization has created more nations than anything else.
The undermining of democracy came with increased deregulation and increased lobbying and wealth concentration.
That's a strawman. What I was doing was pointing out the appeal to the aesthetics of work and associated buzzwords ("capital"), noting the absence of any actual economy policy that will deliver tangible benefits to existing people. It's the same old populist shtick that we've seen in countless fascist and communist regimes where certain modes of work are fetishized and life is regimented around that prescription by a central authority, in the pursuit of a subjective notion of pure work. The giveaway is the attempted justification of an economic policy in service of a nebulous "cultural" impact.
> It's the same old populist shtick that we've seen in countless fascist and communist regimes where certain modes of work are fetishized and life is regimented around that prescription by a central authority
> Frankly, no, sweatshops are not important to the cultural fabric of a country.
IMO industrial policy is the way to mitigate risks of war or extortion vis-a-vis a specific trading partner. Only once this safety criteria is fulfilled, politicians can think about tackling other issues with industrial policy - and unfortunately these further initiatives often fail or have unintended second-order effects (e.g. we want Intel chip factory in Germany).
My understanding is that due to human nature wars mostly start due to religious or extremist views of individuals leading a nation. Such a risk of your trade partner invading you because they don't like your skin color can be hardly formalized in an economic theory (maybe there exists one already, idk).
So role of industrial policy would be to ensure that a certain balance is kept with regards to creating dependencies to other nations, which could be abused in case of war.
Famous negative examples of failed industrial policy for Germany would be the dependence on gas mostly from russia and the dependence on oil mostly from middle east.
Another example would be the agricultural subsidies to ensure all citizens can be fed even when other nations would not export any food. A current example in Germany would be the production of "German steel" using fossil energy instead of production of CO2-neutral swedish steel. As Germany is part of EU, this is a conflicting view: We can't on one side ask for more trade and integration of supply chains between democratic EU countries, but on the other side assume that Sweden will deny steel exports to us when we'd need it.
Producing steel in Germany with fossil energy instead of doing it in Sweden with hydroelectric power is both more expensive and has more negative externalities (CO2 emissions due to use of fossil fuels). Therefore such industrial policy reduces welfare that would otherwise be available for German people.
> And surely in order to leverage comparative advantage, an economy would need to know how good they would be at producing every possible good.
Comparative advantage is an emergent property of trade that occurs naturally, it is the default state of being and can only be undermined by government policy.
You benefit from comparative advantage when you buy bread from the bakery instead of spending 2 hours a day baking your own bread.
Imagine how much poorer you'd be if the government put a large tax on you buying bread to force you to bake it yourself, in the name of self-sufficiency.
That's what's happening with these blanket tariffs, instead of targeting only critical defense manufacturing, Trump also wants t-shirt sweatshops to magically come back to the US despite only 4% unemployment. It's rank foolishness.
The alternative to comparative advantage is that there exist countries where it's economically optimal for them to produce every single possible good with finite resources taking into account the opportunity cost of producing one good over another. Or to put it another way, in a world where comparative advantage doesn't exist, the country in question must have the same economic outcome for any good they produce, and that seems ludicrously implausible to me.
Comparative advantage makes sense, with a national security overlay. That’s where I’ve landed anyway, and is a very simple explanation for all the more complex perspectives out there.
It's even simpler than that. Ecuador doesn't even need to be better than the US at growing bananas, they just need to be better at growing bananas than the US is at developing software relative to their banana growing abilities.
My favorite example is from an economics class quite a few years ago now. Michael Jordon is super efficient at making money playing basketball (told you it was a while ago). But he's also pretty good at mowing his lawn, since he's tall and athletic. But since he's way better at playing basketball, it makes sense for him to focus on basketball and paying some kid to mow his lawn, even though the kid is way less efficient at mowing lawns.
The US is way more advanced than Ecuador, and could presumably develop some hyper efficient banana greenhouse using genetic engineering and AI or whatever. But Ecuador is still pretty good at growing bananas and the US is much better at developing software, so buying bananas from Ecuador and putting the AI greenhouse resources into developing software instead makes way more sense.
> In order to leverage comparative advantage, an economy would need to know how good they would be at producing every possible good.
Maybe I'm not getting what you're saying, but I don't think so. The point of comparative advantage is that even if country A is better at making guns and butter than B, A is better off only making guns or butter and trading to B for the other.
A key assumption being: "Factors of production are fully employed in both the countries. ... The theory assumes full employment. However, every economy has an existence of underemployment."
Another key assumption is "The labor cost determines the price of the two commodities. ... The theory only considers labor costs and neglects all non-labor costs involved in the production of the commodities."
One assumption not listed there is an implicit assumption as in much of economics of infinite demand for anything and no law of diminishing-to-negative returns when considering the environmental and psychological costs of consumption.
So, if you have unemployment in the producer country like China (meaning, there is no reason for them to limit their production) along with a significant capital investment in production infrastructure (like in the Shenzhen region for electronics), and you have limited demand in the consumer country like the USA (meaning, only so much can be sold there at any specific time), then the country which can produce stuff more cheaply will just flood the market of the other country for all goods in question -- even if the consumer country could in theory produce one of the goods at higher costs (or lower quality). Of course, there may eventually be macroeconomic issues like balance of trade issues and countries unable to pay for more goods (which the USA has avoided to date because the US dollar is the refactor global currency backed by the USA's global policing role for decades as a defacto empire). But even if labor in the consumer country like the USA is free, given realistically a lot of cost related to equipment and energy (and increasingly AI and robotics) and more nebulous things like supply chain integration and a can-do attitude, the consumer country may not be able to compete on price and quality of finished products from the more materially productive economy.
Tangential, but "Humans Need Not Apply" makes a good argument when they suggest that horses are essentially obsolete in modern industry (in the same way people may be soon). It's not that you sometimes use horses to any great degree in modern manufacturing (whereas before they pulled carts and turned machines) -- it is that for almost any industrial task horses are more trouble than they are worth now in terms of cost and reliability compared to electric motors or diesel engines and so on.
An economic theory like "Comparative Advantage" that entirely emphasizes labor costs is increasingly obsolete if human labor is less and less a major factor of production. The theory assumes a country will always have people doing something productive, but that is like saying we should bring horses back into factories when robots are generally more reliable. If people are not skillful with access to tools and capital and don't have a can-do attitude, then they will just suffer economically (unless protected somehow) No doubt there are special cases where horses are still useful in production or transport like how mules were used recently to get supplies into hurricane damaged North Carolina, but they are rare as long as the modern industrial system and its surrounding infrastructure functions well. Similarly, there may still be human roles in production, but they will continue to diminish. In 2010, I put together some options for dealing with this situation, available here:
https://pdfernhout.net/beyond-a-jobless-recovery-knol.html
It seems like it's just extremely difficult to give fine-grained control over the metal while having an easy, ergonomic language that lets you just get on with your tasks.
[0] https://www.youtube.com/watch?v=RUJFd-rEa0k