Shameless plug: Adrian is great, and he’s a language / pronunciation / accents geek at least as much as I am into engineering distributed systems :-)

Enjoy him faking accents, quite well to my taste, esp. Russian: https://youtu.be/O6IXQGNVhSU

Try it have guess yours for fun: https://start.boldvoice.com/accent-guesser

Adrian: https://accentamazing.com

Here's something deep to kick off 2025.

When it comes to countries, there are generally three forces at play: the Politician, the Oligarch, and the People.

The Politician controls laws and regulations. The Oligarch controls wealth. The People hold the power of pitchforks, to be wielded if and when they are pushed past the breaking point.

Unless an imminent threat to the status quo arises, in the short to mid-term, both the Politician and the Oligarch generally prefer a weaker country. This counterintuitive and controversial model seems to explain reality quite well.

The Politician thrives on corruption in its many forms. The weaker the country, the stronger their control through laws and regulations. Similarly, the Oligarch thrives on cheap labor; a weaker country means more people willing to work for less.

In the absence of competition or conflicts — such as an armed conflict, the [AI] arms race, or the Cold War — the Politician and the Oligarch often enter a symbiotic relationship. I hesitate to call this a conspiracy; it’s better understood as a game-theoretic Nash equilibrium, a natural point of convergence regardless of whether the parties can communicate or trust one another.

One might combine the Politician and the Oligarch under the term Elite. Personally, I find this oversimplification unappealing, but the big picture remains the same.

In the long run, to model the present and predict the future from a game theory perspective, we must describe the bifurcation points. These points are determined by the threats the Politician and the Oligarch aim to mitigate. There are two key types of threats: external and internal.

An external threat emerges when a country becomes so weak that further decline poses an existential risk. An internal threat arises when the People become fed up.

The "best" way for a country controlled by the Politician and the Oligarch to mitigate external threats is twofold. It is to weaken potential adversaries, and to convince the People that fighting external threats is their patriotic duty.

(The "second-best" way is to strengthen the country itself. However, this is more costly: the Politician risks sacrificing popular support, and the Oligarch risks sacrificing profits.)

To deflect internal threats, the "best" strategy is to ensure the People never reach the breaking point. This involves “sophisticated” use of carrots (propaganda and censorship) and sticks (instilling fear of losing what little they have).

(The "second-best" strategy is to give the People what they want, though this is costly and not always aligned with the short-term interests of the Politician or the Oligarch.)

Now, let’s examine this model further. I don't view it as absolute truth, but it is simple, and it may well be good enough for making predictions.

For the system to be sustainable long-term, the Politician and the Oligarch must correctly assess both external and internal risks. In the past—up until about a decade ago this was a manageable task. However, the explosion of technology, especially social media, has dramatically heightened the stakes.

Today, the Politician and the Oligarch are more likely to overplay their hand, underestimating the risks. Corrections, when they occur, are then overdue and often catastrophic. We’re witnessing this globally — arguably starting in 2019, though many didn’t notice until the chaos became impossible to ignore.

What’s the best play for the People?

First, to internalize that the People are the primary stakeholders in a strong country — far more so than the Politician or the Oligarch. Then, the people make this demand for national strength loud and clear. Two effective methods are voting based on results, not promises, and expressing dissent in ways that are hard to ignore as propaganda and censorship intensify.

What lies ahead? Can the Politicians and Oligarchs maintain their hold? Likely, yes. In the recent years alone, they have amassed wealth and wield powerful media machines. They can regroup and strategically concede minor points. This seems to be what we’re observing.

On a 5–10 year scale, we might see minor tax reductions in targeted areas. However, patriotism tends to surge in the face of clear, visible danger. If minor concessions fail, the unrest of 2020 could repeat itself.

On a 10–20 year scale, a new equilibrium will likely emerge. It may be more transparent, with clearer contracts between the People and the Politician and the Oligarch regarding trade-offs between benefits and lifestyle.

Ideally, the People need better ways to weaken the bond between the Politician and the Oligarch. Historically, monarchies have occasionally succeeded in this. Perhaps this century will bring more self-organized and anti-fragile systems.

Countries like Singapore, Hong Kong, and Dubai may offer glimpses into future safe and prosperous societies. I hope more regions can chart their own unique paths toward similar success.

I remain optimistic for a Golden Age — not just for the United States but globally. A pendulum swing may well over-correct on the way back, that is more beneficial for the People. If nations like El Salvador or Argentina can replicate Hong Kong’s progress in half the time, we may be living in a vastly different world by 2035.

Happy New Year!

F*cking nanny state.

When my dima dot ai did not open in California I assumed it's "to protect our safety".

Perhaps someone jealous reported my plain old static HTML web page. Shit happens.

It's a bummer my short link of zoom dot dima dot ai did not work. We can live with this though.

But not opening com.docker.socket because it contains malware — that's quite insane, isn't it?

To continue on the topic of entitlement and various attitudes towards it.

So I'm finding myself in this line. I'm fairly certain I can cut it. Probably 99+% certain, since my status lets me go right after first class passengers.

I'm a nice guy though. (c)

I politely ask one person if they are in a hurry, or perhaps they would kindly let me through. They said they are. Fine, I ask the next person. Same story. The third person. She says "sorry, I want to go first". I politely share that they were not called yet, while I am allowed to pass already.

She begins a verbal fight.

I am really, really trying to be a nice guy. I mean, bear in mind, I was fortunate enough to mostly not be exposed to the culture of anti-entitlement for a while. In Asia, for instance, they let me into most lounges, while in the US it’s only if the fight is long enough, and even then you may have to pay. Sigh. But I digress.

At some point this lady says: “What makes you believe you are entitled to go before me?” Well, I answer this question, in my preferred Asperger way, trying to be as unbiased and detached as possible. She exclaims that I would not just be cutting her, but also the entire line. Some people from further in the line support her.

As politely as I could, although matching her already heated accusations in tone, I suggested: "Well, why don't we walk to the counter together and check my privilege?"

She explodes. Sincerely, I could not have possibly imagined this would be a trigger point. Well, somehow it was.

Seriously. My assumption was that since "Check your privilege!" means to ask someone to not be rude, then, logically, "Let's check my privilege" would be playing exactly into the modern (?) agenda. Since, of course, I am not asserting that I am privileged, and, in my favorite Socratic way, I am perfectly willing to be proven wrong. And I am merely suggesting that the person with sufficient authority — the airline staff in this case — could act as an arbiter, and as the bearer or the Truth.

Well. Lesson learned.

Perhaps from now on I should only be referring to any privilege by calling it a P-word. Since somehow even the mere idea of someone potentially having a privilege and being aware of such a possibility is already triggering enough.

How did civilization even get this far in this backwards corner of the world?

Because instead of framing regulation as "those guys who make and enforce laws are trying to take even more liberties from us", the way a proposed regulation enters the conversation is "we know now that people, left to their own devices, tend to do bad things, to themselves and to the others, and we are looking for ways to mitigate them".

Also, libertarian as I tend to lean, it helps to think of a hypothetical regulation as if it would help someone I care for, or could care for, to not cross the line.

∼ ∼ ∼

Trivial illustrative examples.

One: Lottery.

On the one hand, I am not aware of any "requirement" to play lotteries in any way. Let's leave aside the philosophical conversation that, from the game theory perspective, one's career or relationships can be viewed as gambles. Practically speaking, no one is forcing anyone else to buy lottery tickets. So, from a fiscally-libertarian point of view, state-run lotteries are among the cleanest ways to collect "stupidity taxes".

Heck, as someone who considers buying lottery tickets highly irrational and highly irresponsible, I would be happy if the state lowers our tax rates, and instead raises capital from its citizens and residents via various lotteries, i.e. only from the people willingly participating in such a form of capitalistic democracy. Right?

Well. Wrong. Because lotteries are addictive, and it generally is wrong to assume people are taking part in them consciously. This bug of the human nature is well researched and well documented. Therefore, a society that is built around avoiding this bug, not exploiting it, is a more stable society.

Besides, imagine you have a friend or a relative, who is otherwise a brilliant human being, but who suffers from lottery addiction. Would you heartlessly view them as a lost soul who can't assume responsibility for their actions and thus should pay the price until they learn? Or would you rather want to be part of a society that simply treats this person as suffering from a disorder that is worth being treated? I would sure hope the latter is the case.

Two: Healthy habits.

People like junk food, and it is a big business by itself. If anything, the capitalist entities involved in making money via junk food would go places to make us buy more and more of it.

Following the same logic as above, we can conclude that it generally is better to live in a society that is taking some measures against people getting too addicted to junk food. In this particular case, we don't even need much regulation which would restrict those unhealthy foods; just introducing better incentives could help. If anything, lower health insurance costs for those who exercise regularly, plus some media campaigns pro healthy eating habits, may well be the measures good enough to get the job done.

∼ ∼ ∼

The bottom line is that over the past few years I managed to both soften my stance on various regulations and become less agitated in those conversations by simply adjusting my frame of reference from "how dare you take away our inalienable right to do something?" to "what makes you think this particular artificially introduced constraint is the best way to counter this particular human deficiency?"

Personally, this altered experience of discussing the topic is something I wholeheartedly recommend, and I wonder if such a change in perspective is worth perpetuating on a broader scale.

☝️I wrote this four years ago. Am proudly a “recovering libertarian” since, for bars and other forms of armchair conversations.

If you believe ChatGPT will very soon replace software developers, think again.

This is a very nasty “feature”. Of the OpenAI flagship product itself, no less. And yet they cannot fix it. For well over a year now.

Mic drop.

Heh, I like how it labels the keys almost right, but struggles with words on the screen.

I wrote this in 2017. Wow.

...

A human’s knowledge of “outcome Y is better than outcome X” is worthless compared to the humans’ agreement on “for problem P, formulation B is better than formulation A”.

If you used ML/AI, happened to not like the outcome X, and are ready to justify Y is a better outcome, don’t argue about Y being better than X.

Rather, if X is a machine-discovered solution to problem postulated as A, find an alternate formulation, B, that yields Y or something closer to Y than to X, and then argue B, not A, is the right problem to solve.

Serious question.

It's perfectly fine for companies to occasionally give away gift cards to employees, right? Say, where was some internal beta-testing, and someone who discovered the most bugs, or was faster, is entitled to a small gift.

As in, last time I checked, no one has ever complained about these occasional gift cards technically being tax free. As long as they are nowhere near the replacement for compensation, and as long as they constitute a small fraction of money paid out.

~ ~ ~

Now, imagine a company of some ~50 people that is using a corporate service with the "decentralized hosting" option.

Such as your work messenger, collaborating editing frameworks, code hosting / review / testing tool, or metrics collection services that do dashboards and alerting. Nothing out of the ordinary, all the standard shenanigans, which most companies are already paying for regardless.

And when I say the hosting is decentralized I mean every team member can run a small part of this service on their own device.

To keep things simple, imagine all it takes is opening the Web browser on some internal web page and signing into your corporate account from that page. The rest if PFM (pure freaking magic). One can literally "start participating in the bonus program" by keeping their browser open.

~ ~ ~

The more often one's browser is open and ready to serve the team's needs, the more "bounties" can its owner collect.

If the company is a Web3 company, incentivizing people is trivial: just have their browsers earn tokens, set up a bridge to convert these tokens into the company's main token, and have people effectively have their "token bonus" tied to their "in-browser mining activity".

This is very much the ESPP (employee stock purchase program), but one's degree of participation in it is conditional on them continuously keeping their browser open on a particular page.

Of course, one could also dedicate their home machine, or an unused old phone, to running this company's "internal services".

For a more advanced thought experiment, consider making it a possibility that someone who works for A can "collect bonuses bonuses" by providing their machine's spare resources for B, while someone who works for B can "collect bonuses" from A.

~ ~ ~

To be clear, we're talking about some ~$10 per month "total bonuses collected" for a given company.

Even if some employee (or a founder, or the head of DevOps) decides to go wild and adsorb 99+% of the "compute power" of this "company's decentralized cluster", the grand total of what the company spends on these "internal services" is under ~$1,000 per month.

And, for reliability reasons, the math would not let one person claim more than a ~quarter of these tokens. Since we need the overall setup to pass the bus test.

~ ~ ~

So, we're now talking about a 100% decentralized stack to run the company's processes. Obviously, some data concentration should be in place, since the data must not be lost — this can be a single box + single object store hosted on a major cloud provider (i.e. pay ~$25 per month to AWS). Or this can be a few devices with their own local CPU and storage. The founders would simply keep one device each, and perhaps hand out a few to other people.

Encryption technologies are very much good enough these days to ensure no data is tempered with, as well as to ensure no data is leaked, unless the leaker a) was explicitly granted access to some data, and b) has had their password credentials leaked as well.

Even a potentially malicious bad actor can not do anything beyond what they could have leaked otherwise. So, this setup is even more secure than 99+% of modern-day tech companies.

~ ~ ~

Going back to gift cards though, which is what this post is about.

Is there anything potentially wrong with the setup where the task of running the bi-weekly payroll also includes buying gift cards to the "top performers", according to some internal dashboard of who these top performers are?

Because if no, I just can't see this not being the long-term future.

This ☝️

I have a thought that might look crazy at first, but hear me out.

Thought: Rationality is literally the weakest force when it comes to human motivation.

And by motivation here I mean both self-motivation (hard work and discipline) and external motivation (marketing and persuasion).

Here's an even crazier thought. Thinking rationality is somehow big of a force is a false belief many otherwise smart people are instilled with in their formative years. Believing that rationality is what ultimately works is literally a major trauma inflicted onto an undeveloped brain, and we spend decades afterwards to cure ourselves from this trauma.

One of my favorite sayings of all times is: the reality prevails. (Thanks Peter!) However, an intellectual brain tends to misinterpret this wisdom by getting it completely backwards.

Here is how I understood rationality from when I remember myself all the way to early forties at least. There's the real world and there's a model of the world. All models are wrong, but some are useful. With enough smart people, enough motivation and resources, and plenty of feedback loops, useful models appear. Knowing a useful model one can strategize within it. Then, whoever has the "most useful" (i.e. the "least wrong") model plus enough reasoning skills can navigate the reality more effectively.

That's why we have water and electricity and bridges and airplanes and the Internet.

Well, what a wonderful model. Except it does not work when applied to an individual.

Think of something mundane yet fixable that routinely bugs you and/or someone close to you. Why isn't that fixed? Isn't it a definitive step for the better life? Doesn't one benefit from having that fixed?

It's only rational to say: "It bugs me. It can be fixed. If it bugs me one more time I'll go ahead and fix it. This would be a worthy investment into my life: time, well-being, prosperity, happiness."

Moreover, I am very positive quite a few otherwise rational and intelligent people are reasoning along these lines about something that bugs them. Maybe not every day, but several times per month for sure.

And yet how likely is it that that "something" does get fixed? Well, far more likely than unlikely.

The logic that "I need to eat or I die of starvation, and therefore I will find a way to eat" somehow does not map to "I need to get that [blank] fixed."

The only reasonable (rational!) explanation is that understanding, or even internalizing that something is worth doing is nowhere near as strong of a human motivator as, well, other things.

How do I deal with this myself, by the way? Mostly with two mind tricks:

One: Don't be too hard on myself.

A lot of balls will be dropped. The human civilization is fairly well designed for people who routinely drop many, or even most balls. As long as the important stuff is in order, and as long as I'm nowhere near the bottoms of my peer group when it comes to dropping important balls, that's definitely livable.

Two: Leverage identity.

That's the oldest trick in the book, and it works. Instead of thinking "I should do X [or I will suffer in the future]" or "I should do X [or I would label myself as a procrastinator who doesn't respect myself]", reframe this as "I should do X because I am the person near whom X is done".

No ego, no emotions, nothing to be proud of when X is done, nothing to be ashamed of if X is not done. Just slowly work my life towards the state where if I'm participating in something for a long enough time, X, and various Xes, do get taken care of.

I'm the force of nature that makes Xes happen, and in my finite lifespan, I ultimately am quite comfortable to program my self so that a nonzero and even somewhat increasing share of my conscious being is spent on perpetuating the ritual of having the Xes around me taken care of.

A great post by Jos, short & sweet, and not too nerdy.

TL;DR takeaways as understood by me:

1) Most often to move forward we need to [consciously] rewire our [collective] brains.
2) Investing into having some [shared] "better understanding" of the problem is just a waste of time.

Kind of trivial, but I wish someone was there to explain this to a much younger me. And I wish I had the skills to explain this to my younger teammates. There appears to be a generational pattern. Oh, wait, the Greeks have noted this one already; case closed.

If you think AI is taking over the world, think again — your understanding of what the world is may well be quite biased.

(I.e. out of sight out of mind — this is real.)

This is from Anthopic, about Claude.

TL;DR: By "penetration rate", software development is #1, design and arts #2, education #3. Other use cases, including legal — far less noticeable disruption. Can even say it's "barely noticeable".

I.e. some zero-"AI" "code completion tool" for legal emails may well have had more impact on the field over the past few years compared to all the modern-day's "AI" hype.

I just had an obvious yet long-reaching revelation.

It was always the case that the future belongs to technologies that are easy to teach & make people use.

(This is how stuff like nginx / Redis / node.js / Mongo / Kafka came into play.)

But the inflection point is that teaching others to use some technology is no longer about making something appealing to humans.

Now it is about making it trivial enough to that AI can "comprehend" it. Because the main channel how something proliferates into the minds of developers is if and when an AI suggests some tool, with ready to copy-and-paste-and-run code snippets.

In other words: If you are working on a new technology today, and you need developer adoption to succeed, ditch any and all presentation materials. You are no longer hosting "events" and "bootcamps" to reach out to your audience.

Instead, to get to developer adoption, focus on making sure AI models are effective enough at suggesting solutions that use your code!

The more trivial one-sentence prompts result in the output having your library, the better off you are.

"AI-first SEO for software components" is the future. Scratch that. "AI-first SEO for software components" in the present.

Pragmatic advice: If you are working on evangelizing some technology, come up with a ~hundred reference use cases for where it shines, and begin tracking top AI models' outputs for these prompt, to collect the data on how many of them a) mention your product, and b) produce correct, ready-to-run code.

The world is healing. Even Sabina is speaking up.