Robert Ritz

Google's Weather API doesn't want me to use it for weather.

Mon, 10 Aug 2026 12:12:58 +0800

Personally I’ve never wanted to make a weather app, but then I got an email from Google saying that their new AI based weather forecast model, which is supposed to be cutting edge, is now available for unlimited free experimental preview.

I’ve always found the precipitation forecasts to be quite poor in Mongolia. Since 2011 Mongolia’s National Agency for Meteorology and Environmental Monitoring (NAMEM) have used a [Cray](National Agency for Meteorology and Environmental Monitoring) super computer for forecasting. It works, but gets hourly precipitation wrong quite often. They are directionally correct, but they usually don’t help me in deciding to take an umbrella or not on a given day.

So when I got the email, I thought, why don’t I see how good it is? The best way to do that is to throw up a simple web app and test it over the next few weeks. So I go over to the console, enable the API, get a key, then ask Codex to whip it up.

It refuses….

The rationale is that 21.1 of the Google Maps Platform Service Specific Terms specifically prohibits this…

Weather API 21.1 Restrictions. Customers may not use Google Maps Content retrieved from Weather API to recreate a Google service or product (e.g. use data retrieved from Weather API in a weather app or weather model whose primary purpose is to provide weather information).

Codex helpfully nudged me into something that would tell me if I should take an umbrella or not. An app that tells me to take an umbrella or not is a weather app, albeit one that gives a binary signal rather than a precipitation percentage.

So if I make an app where the primary purpose is giving the news but I want to throw weather at the top that is fine. But if I want to make an app to test out this new weather API that is not, according to their own conditions. So to test out a weather API I need to make an app where the purpose isn’t weather. Lovely…

Update: I explained the silliness of this refusal and said that I’m just testing the API. It went ahead and built it. weather.robertritz.com is live and I’ll see if this API is any good.

Serious fuel shortage in Mongolia

Thu, 06 Aug 2026 12:34:21 +0800

This shortage is showing the strength of third neighbors and the weakness of direct ones

For the past few weeks, there has been a shortage of fuel in Mongolia. This is mostly due to the Russian invasion of Ukraine and Ukraine’s serious hammering of Russian refineries and pipelines. An additional challenge coming up will be the annual winter maintenance at the refinery in Irkutsk, where most of Mongolia’s fuel comes from. That usually shuts the refinery down for two to three weeks and causes shortages each year during the back-to-school rush.

Here are Mongolia’s fuel imports from 2021 to 2026. I wrote about a shortage back in 2021, and that one seemed to show an emerging pattern. At that time, a brief reduction in imports in early summer caused a cascade that resulted in a shortage a few months later. This happens for two reasons:

Mongolia has a limited amount of fuel storage capacity (about 45 days at most) for AI-92, the most common fuel for passenger vehicles.
Supply from Russia is limited, even before accounting for the war in Ukraine.

Mongolia is building more storage tanks, but this will only put a Band-Aid on the real problem: supply. About 95% of all fuel in Mongolia is imported from Russia, mostly from one refinery. If that isn’t a supply risk, I don’t know what is.

Source: [Data.mn](https://data.mn/en/data/mrpam-petroleum-imports)

Supply shocks from the Russian side are a yearly occurrence, and the war in Ukraine has only exacerbated them. When I wrote about the shortage in 2021, I noted that the government should have had a pretty good idea, months ahead of time, that there would be a shortage. Yet there was seemingly little public activity by the government to address the issue.

Third-neighbor strength

2026 is no different. New fuel supplies from South Korea and China have been secured, but they will have to arrive by southern rail. This route uses a different rail gauge, adding cost and time to the import process. In addition, fuel from South Korea must first be transported by sea, offloaded at a port, and then taken by train to Mongolia.

South Korea has pledged to supply 50,000 tons of fuel products monthly. China pledged an emergency supply of 6,000 tons, presumably as a one-time shipment.

Notably, Chinese oil imports have gone down significantly during the Iranian conflict. This is something of a mystery, given that China’s fuel storage, much of which is outdoors and visible from the sky, has not drawn down. Some experts in this space believe China may have a “secret stash” of fuel created in preparation for a potential conflict over Taiwan.

Benzin.mn

I’m astounded by the number of direct-name domains in Mongolia: oboi.mn, meaning wallpaper; unegui.mn, meaning free, although it is used for classified ads; data.mn, my own domain, meaning data; and medee.mn, meaning news. There are a few others that I can’t remember right now.

Well, now there is benzin.mn. Benzin is the Mongolian word commonly used for gasoline, although technically it means benzene, which is not gasoline. The site is a crowdsourced map showing which stations have gasoline.

This morning, I checked and only four stations in Ulaanbaatar were reported to have fuel. This is on top of an odd-even license plate restriction and a maximum purchase of 50,000 MNT.

Oil, oil everywhere, and not a drop for my Prius

Ironically, Mongolia has a lot of oil. It exports basically everything it produces. Production appears to be declining, something I didn’t realize until I looked at the chart this morning.

Source: [Data.mn](https://data.mn/en/data/mrpam-petroleum-production)

Nevertheless, in 2016, India decided to invest $1 billion in an oil refinery in Mongolia. At the time, this was the estimate of the fuel it would be able to produce:

…the oil refinery will have a processing capacity of 1.5 million tons of oil. As estimated, the oil refinery will produce 560 thousand tons of gasoline, 670 thousand tons of diesel fuel, and 107 thousand tons of liquefied gas per year.

By comparison, Mongolia imported 828,000 tons of AI-92 in 2025. This would therefore be a significant step toward reducing Mongolia’s dependence on foreign fuel.

There are a few problems though. Oil production declined from 4.6 million barrels in 2021 to 3.7 million barrels in 2025. Doing the math, 1.5 million tons is roughly equivalent to 30,000 barrels per day. If the refinery operates year-round, which is fairly standard apart from short maintenance shutdowns. That amounts to 10.95 million barrels per year. This means Mongolia’s current oil production would utilize only about 34% of the refinery’s expected capacity and therefore produce less gasoline that it is capable of.

It’s possible that there is a lot of oil-production capacity that could quickly be brought online and that I’m not aware of. The existing numbers don’t look good, though. Part of me wonders whether India’s investment in Mongolia was more of a snub at China than an effort to actually help Mongolia.

Winter prospects

It’s hard to say whether this fuel shortage will continue through the winter. The promised emergency supply from China will do little to help. However, the 20,000 tons of AI-92 (out of the 50,000 tons total) promised by South Korea will put a meaningful dent in the average 72,000 tons of AI-92 imported each month.

In the short term, at least through September, I expect the odd-even rule to continue. The real risk in all of this is diesel fuel. If that runs low, food trucks can’t get to market, and trains can’t transport food or coal.

With COP17 coming up, desertification looks like a small problem.

Lecture 2: The GPU Economy

Wed, 05 Aug 2026 22:32:02 +0800

In this second lecture, held on April 9, 2026, Brad Gerstner of Altimeter Capital and Sunny Madra of Nvidia (previously Groq) discuss the economics of GPUs and inference. Anything that is my personal opinion will be denoted with italics.

Course Page
Watch the Lecture

The main question for the lecture is fairly simple: how does the cost of inference keep falling when both model size and demand keep going up?

The GPU economy

Gerstner starts with a long-term view of technology. His slides show GDP per capita roughly doubling every 25 years, while technology’s share of GDP grows from about 5% in 1998 to an estimated 15% in 2030.

He also compares investment returns over the previous 15 years:

Technology companies: about 15% per year
Non-technology companies: about 6% per year

His point is that compute sits at the root of these changes. Most of the gap between tech and non-tech appeared after 2020. I have a hard time seeing this as a compute story and not a market run up that has little to do with value….but whatevs

Gerstner then introduces Madra, describing him as a founder who keeps getting acquired by larger companies. Madra was president of Groq before joining Nvidia.

How Groq works

Groq founder Jonathan Ross previously worked at Google, where he helped develop the company’s Tensor Processing Unit (TPU).

Madra describes Groq’s chip as a deterministic dataflow system. A compiler decides where the calculations will happen before the program runs. This is useful for large language models because generating tokens requires quite a bit of math.

The amount of work needed to generate one token depends partly on the number of parameters in the model and the amount of context it must process.

Prefill and decode are different problems

Madra separates inference into two parts:

Prefill processes the prompt and its context.
Decode generates the response one token at a time.

These operations put different demands on the hardware. Madra says it could make sense to use different machines for each one.

Groq’s design has a large amount of SRAM, which is fast memory located directly on the chip. Nvidia has a wider collection of chips and systems designed to work together. Madra also discusses NVLink Fusion, which allows custom processors to connect to Nvidia’s systems.

In the example he gives, combining the two approaches can generate 2.5 times as many tokens using the same amount of electricity.

Will inference keep getting cheaper?

Gerstner asks what will push the unit cost of inference down. Madra gives three answers:

Improvements in the supply chain
Work by hardware and systems engineers
Power

Lithography, the process used to manufacture increasingly small features on chips, is beginning to reach physical limits. One response is to make the chips larger. Cerebras produces wafer-scale chips that Madra compares in size to pizza boxes.

Hardware development is also happening alongside model development. Better hardware makes larger models possible, while larger models create demand for more hardware.

Madra says models with as many as 10 trillion parameters are coming. Even a 50-fold increase in capacity might not be enough over five years if models continue to get bigger.

There are already tens of trillions of tokens being generated each year. This was in April. Appetite for tokens has increased like they predicted, if not more so.

Cost goes down, usage goes up

Gerstner says OpenAI initially had negative gross margins because inference was expensive. As the cost of inference came down, the models became more useful and people became more willing to pay for them.

The models are also beginning to take actions rather than only answering questions. That increases the number of tokens used for each task.

Madra says there are two things that people outside the industry may not see yet. First, today’s public models were not trained on the newest hardware. Second, companies are still figuring out how to get more work from the models they already have.

A system that can take a problem and work on it continuously will use far more tokens than a chatbot waiting for someone to ask a question. Both alluded to the fact that there isn’t enough compute, and probably won’t be for some time. Cost goes down….great, but I want 1,000X the number of tokens.

Is AI a bubble?

Gerstner points to Anthropic’s rapid revenue growth as evidence that willingness to pay is increasing along with model capability.

Madra argues that the current models and chips do not show where demand will settle. Newer models will be trained on newer hardware (Blackwell at the time of this lecture), and companies will get better at putting those models to work.

This does not answer whether every company or infrastructure investment will earn a good return. It does explain why falling inference costs do not necessarily mean lower total spending.

Q&A

How should people prepare for this?

Gerstner says the Industrial Revolution disrupted many existing jobs, but people found new ways to create value.

He gives a more immediate example from Altimeter: he would not hire someone who does not use Excel. His argument is that some forms of IQ may become easier to obtain, making EQ and problem-solving more valuable.

Madra uses mathematics as another example. If AI systems begin making new discoveries, the people and organizations that know how to use those discoveries could have an advantage.

They both gave pretty crappy answers. But I think the answer is….we don’t know. And that’s ok to give as an answer, but they don’t do that.

What is going on with Apple?

Gerstner says Apple’s AI strategy is risky because language models are not yet good enough to do everything locally on a device. Privacy is also a major consideration for the company.

The case for Apple is its existing base of devices. If smaller models improve enough, products such as Siri could become much more useful without sending every request to a data center.

Power remains a problem. Madra says an 8-billion-parameter model running on a phone can drain the battery in about 30 minutes.

I personally feel that Apple is doing a great job holding off. Apple has always succeeded by thoughtfully doing something, not slapping new crap everywhere…ahem Samsung.

How should AI CEOs talk about AGI?

Gerstner criticizes fear-based arguments intended to produce regulatory capture (cough…Dario….cough). At the same time, he says people should not put their heads in the sand.

He takes Dario Amodei and Sam Altman seriously when they describe how quickly the models are improving. The rate of change, in his view, is becoming fairly parabolic.

What is Nvidia’s long-term business model?

A student presents two possibilities. Nvidia could protect its margins and accept slower revenue growth, or lower its margins and maintain a larger share of the market.

Gerstner says Nvidia has a strong product schedule and much of its capacity is already booked for the next eight quarters. He also points out that people once said Nvidia could never become a $1 trillion company.

Competitors such as Groq, Cerebras, and Google’s TPU can succeed without Nvidia failing. Gerstner describes this as the good part of capitalism.

This student can’t see the forest for the trees. He sees the GPU market as a zero sum game. Nvidia has never worked like that. Silly premise.

Course notes

That is it for this week. I will keep posting my course notes and comments on this page.

My 10 year old coded his first Minecraft mod....sort of

Wed, 05 Aug 2026 19:01:16 +0800

My son used Claude Code for a while to help him fix mods and do some other small things, mostly gaming related, on his home computer. Since I switched to Codex a few weeks ago I convinced him to switch as well. It was easy since he uses perhaps once a day to a few times a week. He is 10 so he can type reasonably, but he prefers Wispr Flow to dictate (I set him up with my unused account since I use a local app on my Mac now).

Last night he was looking for a mod for Minecraft, which is focused on quickly switching between keybind profiles by hitting a hotkey. He found one, but it was buggy and didn’t work well. So I said….just make it yourself. He was super skeptical at first, but I said Codex with Terra could easily do it.

I sat him down with and helped him with the first prompt and showed him how to phrase it so it would actually make it and not just brainstorm endlessly. He set it to work and it took two turns to get it to launch and work properly, and another 30 minutes or so of work to polish it into something he really really likes. He is over the moon at what he did and he thinks it is amazing. He specifically directed the model to ensure it complies with server rules and doesn’t give him an unfair advantage (this impressed me).

As a 10 year old he is somewhat exasperated at the huge amount of AI generated content out there and he generally considers it a menace. But this got to him and showed him the real power. Now I’m convincing him to polish it a bit more then release it on Curse Forge, a popular modding platform, so others can benefit from it.

If he would have had to learn enough Java to make this work on his own, it would have taken him two weeks of work, and he would have given up after a few hours of learning. Part of me feels like learning how to code would have been better, because it would have taught him how to learn a new skill. Another part of me thinks that is stupid, and that I don’t even code anymore. What matters is the process, asking the right questions, and ensuring you make something that is quality.

So my son “coded” his first Minecraft mod.

US visa bond program is now permanent

Tue, 04 Aug 2026 13:54:21 +0800

The visa bond program for the United States started impacting Mongolia in April 2026. This program requires B1/2 visa applicants to post a bond of $10,000-$20,000 dollars and limits visa terms from three months to a year. This was a pilot program, and it was just made permanent.

It’s annoying for us, to say the least. At least for now, when my wife’s 10 year visa expires, she will need to go through this process. Since she was 18 or 19 she has gone to the US many, many times. In fact she lived there for 8 years, which is when I met her. We were even married there.

We go to the US to visit family, and to do business, every year. One of the largest jewelry trade shows in the US is in Las Vegas each year and she attends. In fact most of the purchases for her jewelry business in Mongolia come from US suppliers.

I understand the State Department’s reasoning for this decision. I’d argue that this program, as it is currently structured, will probably cost the State Department more money. For all those thousands of Mongolians who travel to the US regularly, and who don’t break the rules, they will now have to go to the US Embassy in Ulaanbaatar once a year (in the best case) and do a visa interview. It is going to overload the already stressed system they have.

It would make much more sense to be able to waive this program for those with a long and obviously strong visa history.

Back in 2015 when my wife and I were married, she had to get a visa to go to the US. We made no secret of the plan to get married in Texas, and we planned to go on our honeymoon and then come right back to Mongolia. The consular officer basically said that she wasn’t allowed to get married in the US, which isn’t correct for obvious reasons. But the reason they said this is because once you are married to a US citizen and in the US you can file for an adjustment of status and then stay. That wasn’t our intention of course, but for a consular officer it’s certainly a bit of a weird thing to hear from a visa applicant in a country where most people would love to immigrate to the US.

Anyways, I tell this story to point out that my wife has had ample opportunity over the past 20 years to immigrate legally. She didn’t. So this program makes little sense other than to create more hassle for us and the consular officers at the embassy.

Funny enough, all immigrant visa issuance from Mongolia are paused anyways. So even if we wanted to move our family to the US we couldn’t…

Wild Strawberry Jam

Sun, 02 Aug 2026 09:20:22 +0800

One of the very nice things in Mongolian summer is the wild strawberries! Last night we made jam with about 3 liters of strawberries.

Lecture 1: Economics of the AI Supercycle

Wed, 29 Jul 2026 14:25:24 +0800

In this first lecture instructor, held on April 2, 2026, Apoorv Agrawal (currently Altimeter Capital, prev Palantir) goes over the format of the course and the purpose of the course with some primer concepts for the class. Anything that is my personal opinion will be denoted with italics.

Course Page Materials for this week:

Format of the course

The course is primary guest speakers covering different areas of the current AI supercycle. Speakers are in areas such as GPUs, SaaS, infrastructure, etc.

Why take this course?

We are at the start of a massive supercycle. Agrawal makes the contention that this cycle will be larger than cloud, mobile, and the internet.

Where is the $ in AI today?

Agrawal asks some questions from the class then continues the rest of the time with a discussion.

Q: Is this CAPEX generating revenue?

A lot of money is going to CAPEX right now (capital expenditure), the so called 5 layer cake Jensen Huang has referred to (energy, chips, power, interconnect, memory). This CAPEX is amortized (at least for GPUs) over about 5-6 years. Is it worth it?

Google: $100/user/year
Meta: $70
OpenAI: $10

How do you get to >$10 per user per year? Agrawal makes the point that knowledge work isn’t the answer, advertising is.

How this revolution is different

Software gets 80-90% gross margins. AI services at billions in revenue still isn’t profitable. Average margins for AI companies are much lower, around 30%.

Amazon took 8 years to build out AWS, and during that time people were asking if Amazon would go bankrupt. This is probably because Amazon had famously near 0 net profit for many years post IPO, instead focusing on reinvesting to increase revenue.

In cloud most value accrues in apps, but in AI its at the semiconductor layer. Agrawal states that gross margins for semis is 75% (0-30% for AI apps layer). This is the inverse of the cloud period. Also, semis are built for 5-6 year depreciation cycles, but apps generate revenue now. Mobile supercycle had the same inflated CAPEX early in the cycle.

Don’t underestimate Google

A student made a point that Google is one of the only complete AI stacks today, they have TPUs, apps, existing customer base, etc.

Agrawal makes the point to pay attention to earnings calls. CAPEX has an equilibrium (limited to free cashflow and cash reserves and also the expected revenue gains from additional CAPEX), so earnings calls will signal a change in this equilibrium.

This lecture was in April 2026 and this is somewhat prescient but perhaps underestimated the frothiness of the AI CAPEX splurge. Google just had its very first period of negative free cashflow since 2004.

Agrawal says:

Google won the internet supercycle
Apple won mobile
Meta won social
Oligopoly in cloud

Because of Google’s power in chips, apps, and existing customer base it might be likely that Google wins here.

Training and inference are unpredictable

Inference is dependent on when humans are awake, so this causes the usage to be unpredictable. This was how AWS got started (by using excess capacity since Amazon didn’t have traffic at night), so I feel like this isn’t as big of a problem as he states it is.

Where are all the ASICs?

ASICs are application specific integrated circuits. Groq (acquired by Nvidia in Dec 2025) and Cerebras (Jan 2026 deal with OpenAI) are two that come to mind.

There is 300 billion in revenue to go around. There wasn’t much discussion here around this topic, perhaps more in a later lecture.

Where will value accrue in AI?

2024: 90B vs 2026: 435B in revenue. But the shape of the revenue hasn’t changed much.

Yet profits remain in the semis later.

AI Apps - Weekly active users

ChatGPT has most users, then Gemini. Consumer apps are dwarfing ChatGPT, even with all of the hype.

ChatGPT is closer to the niche app category (like Spotify and X) than it is to Netflix or YouTube.

Is knowledge work where most people are?

In ChatGPT you have to go and ask a question. This is very active and most people don’t prefer to engage in apps like this. To increase revenue Agrawal makes the point that these companies should use ads rather than subscriptions, and he thinks this is where it is heading.

Intent is much cleared in AI chat apps, so sure revenue per ad will be higher, but the inventory will be much lower as well.

Course notes

That’s it for this week. I’ll keep posting my course notes and comments on this page.

Timeback cache shenanigans

Wed, 29 Jul 2026 12:22:47 +0800

My son is doing GT Anywhere, which is an offshoot of Alpha School. The Timeback platform kept crashing on him for two of the lessons. I suspected a browser cache issue because of how it loads (via an embedded Electron browser in the app).

Deleted the cache in the container and it worked. Had some help from Codex to tell me where to look.

I truly think Timeback, Alpha, and GT School are the way forward in education (it’s working great for my son). I expected there would be some rough edges with something so new.

Even with all of these rough edges my son loves it. After class his current favorite thing to do is play table tennis.

Broken machine day

Mon, 27 Jul 2026 13:36:13 +0800

Last Friday was broken machine day. Two of our six-sided hole machines, used for inserting holes and slots into furniture pieces, broke down.

For the first machine one of the spindles wouldn’t raise or lower. These are controlled electronically from the computer via an pneumatic system. We suspected either a computer issue, electrical, or a solenoid valve which controls the pneumatic air flow.

The offending spindle.

After a lot of troubleshooting I was convinced it was the solenoid valve. Our factory manager climbed up and checked on it.

Solenoid valves have these little buttons that allow you to activate them manually. This way you can see if the valve is working without an electrical trigger.

They have these little red LEDs telling you when it is being triggered. This is a “two headed” (no idea if that is what they are actually called) which allows airflow two ways. Often these are just open closed situations but this one allows airflow to move the spindle up, and another to move it down.

The electrical part was working, but the solenoid valve itself was borked. We heard it clicking but no movement.

We bought a new one at a local supply store but it was the wrong electrical type! These come in AC or DC varieties. The one we bought was for AC 220V but we needed it to work with DC 20V. Well it turns out we can just take the electrical ends from the broken valve and put it on the new valve. I’m going to research later to see why this is.

Anyways after a few hours we got that fixed. On to the next machine!

The next machine gave an error code, which is usually very helpful. The sensor on the lower spindle wasn’t responding, which usually mean the sensor is broken. These are little sensor that determine if a pneumatic cylinder is in the up or down position (or extended or collapsed depending on the orientation).

We replaced the sensor with a spare we have, but it still didn’t work! So time to trace wires.

We ended up finding out where one sensor wire was split into two, and one of the connections came off. This was nicely spliced too, but I think wear over repeated use made it come loose. We spliced it again and taped it up. I need to remember to bring my wire shrink insulation to the factory for the future.

Oh well it’s all good now! We’ve gotten quite good at troubleshooting these machines over the past year or so. We keep backup parts because a new part that isn’t available locally (like sensors) would take about 10 days to get here from China.

I’m quite surprised at how the old troubleshooting skills I learned from building PCs back in the day, or by working at an IT outsourcing company nearly 20 years ago helps with this stuff. It’s pretty much all the same.

AI requires a builder mindset to get value

Mon, 27 Jul 2026 08:41:38 +0800

Only 3% of American households are paying for an AI subscription. If LLMs make it so much faster and easier to get work done, why isn’t everyone using them?

This X post from Sam Altman sums it up nicely:

chatgpt work is remarkable, and “work” undersells it.

from my phone i sent:

“use all my chat history to figure out ideas for a long weekend trip with 8 friends, plan the best three options, make a full-stack site where the 9 of us can coordinate on what we would want to do in each place and decide where to go, and then after we get to group agreement make reservations. draft an email in my gmail i can send out to my friends when the site is ready.”

it…just worked.

I’m sure it did work. I’ve made so many personal utilities that are either single-use (i.e., throwaway) or that I use every week. I’m struck by the prompt.

Sam had to think through his plans, the range of options he wanted, the output he wanted, whether he needed coordination, and how he wanted to get the results. In other words, he thought like a builder. This is a mindset that, in my experience, is pretty uncommon.

When judging whether a product or service is techie, builder-oriented, or non-mainstream, I usually check with my wife. She is super smart, high-agency, and can generally look at something and immediately tell whether it will help her in life or work. I’ve been trying to teach her how to use ChatGPT Work to get things done for the past week or so. She has used ChatGPT for quick answers, brainstorming, and similar queries for some time now. But she hasn’t really used it to produce work yet.

I convinced her to try making a series of presentations for a course she is teaching. The process has been relatively bumpy, I must say. First, her prompt was too vague, and the LLM dutifully made everything in one go, with predictably middling results. Then I recommended that she start with one slide deck and, once she was happy with the result, ask ChatGPT to use it as a guide for the others. That worked better, but it took 30 minutes to make a single slide deck because of all the checks. It worked, but she was pretty underwhelmed. In her mind, she could have done a better job herself in maybe 45 minutes, so why wait 30? After making it herself, she would also understand the thought process behind the slides and be better able to present them. Fair point.

I think a better, chunkier, and more specific prompt at the beginning would have gotten her 90% of the way there, with perhaps 10 minutes spent writing out what she specifically wanted. But then I realized that she didn’t exactly know what she wanted until she started doing it. Well, I should say that she knew what she wanted, but she didn’t know how to articulate it as a series of instructions an LLM would understand.

Without a lot of coaching, I just can’t see more than a small percentage of people thinking this way. It’s weird. When they want to pay their bills, most people don’t think about the mechanics of finding the bills, paying them through a payment system, and recording that each bill was paid. Builders think about those things.

Until LLMs can do that builder-style thinking for ordinary users, I don’t see my wife, or people like her, choosing to turn to one most of the time.

LLMs are bad at blog posts

Thu, 23 Jul 2026 10:12:37 +0800

Excellent post today from wakamoleguy about how LLms are bad at writing blog posts. Quick and simple argument about information theory that I believe is correct, and more importantly, this great snippet:

So let’s go back to the beginning: you have an idea, but writing is hard! Write anyways; write poorly. Use an LLM for research and feedback. Do not simply hand it your idea and expect it to generate something enjoyable or effective to read. And while you’re at it, lean into the imperfections, because even if they aren’t information-theoretical depictions of your ideas, they help make your writing interesting, unexpected, and yes, surprising. And that is what makes it fun to read.

I’m working on a longer form data driven blog post over on my nascent Substack publication, and I’m finding it hard to write like I used to. I’m writing then asking the LLM to give feedback. It often wants to make my data points overly specific, which I find annoying. Nevertheless it finds typos and areas I didn’t make all that much sense, so it’s helpful! This way I don’t have to bug a colleague or my wife to read it first!

Get to writing!

America's ice cream consumption

Fri, 17 Jul 2026 18:35:39 +0800

Neil Howe and Christian Ford write about ice cream consumption:

Over the last 50 years, US ice cream consumption has steadily declined. In 1975, the average American consumed 18.2 pounds of ice cream. In 2025, that figure fell to only 12.0 pounds. That’s a -34.1% decline.

This is quite sad to me. One of my fondest memories with my grandparents is eating dinner and having a scoop of vanilla ice cream afterwards (usually with chocolate syrup on top). My grandfather would have a scoop as well. Some doctors even believe it helps prevent Type II diabetes.

Wed, 15 Jul 2026 20:43:33 +0800

It’s my father in laws birthday. He enjoys a card game called 13 (which he is very good at).

You add up the number of cards you have left at the end of the round. Once you reach 25 you are out. I held my own and was second to leave the game out of four. It’s not really my game but it’s good fun.

Long running task in Codex

Wed, 15 Jul 2026 12:49:46 +0800

20 hour in and I’m getting apprehensive as to the utility of this run. Lots of code changes and testing. The process is a well defined one, at least conceptually. It is still on step 1 of 5. The first step being scraping government websites in a country (not the US) that ostensibly has a lot of open data, but practically tries to restrict scraping by any means necessary.

Codex is dutifully writing, running, and updating numerous scrapers to deal with the random 403s that shouldn’t 403, data inconsistencies on government websites, and connecting records via trial and error.

But I heard a long time ago never to attribute to malice what is more likely simply incompetence.

Tue, 14 Jul 2026 19:56:05 +0800

Fascinating point from antirez:

I believe many programmers at this point have less impact they could have because they look at the code

I couldn’t agree more. The thing should do what it is supposed to do. The person directing the LLM agent is responsible for checking that manually or autonomously. But looking at the code is a suboptimal use of time.

Tue, 14 Jul 2026 18:57:21 +0800

This is our fourth time coming to this part of Bulgan Aimag in Mongolia. This little Soum center called Khutag-Undur has a nice hill next to it. Mongolian summer is a brief respite from a harsh environment.

Tue, 14 Jul 2026 18:04:55 +0800

From Numeric Citizen Blog:

I don’t know if I should trust the news, but it seems that opposition to data centers and artificial intelligence is growing.

I get the same feeling. The sense I feel is that people genuinely think AI companies are trying to take peoples jobs, and that data centers are causing increases in electricity prices as a result.

Mon, 13 Jul 2026 19:20:55 +0800

Not a bad way to end the day.

Sat, 11 Jul 2026 11:35:14 +0800

Traveling in Mongolia for Naadam and testing out GPT5.6 from the room.

Fixing stuff day

Mon, 06 Jul 2026 16:09:54 +0800

Today is fixing stuff day.

For one of the hats I wear I’m a university administrator. Over the past few years our admissions system has become something of a hydra. We have four different deployed applications for our admissions system and one for our student information system. I’m going to put them all together into a big mono repo today so it’s easier to manage and maintain them.

For my other hat I’m ostensibly the CEO of a furniture company, but because no one else on the staff knows how to terminate ethernet, that’s my job today. Needed to move a security camera a few meters one way in the factory. I made a standard straight through cable and it didn’t work! Powered up but no data to the NVR. So I looked at the source cable and I see it’s split! The installer decided to use 1 ethernet cable for two cameras (IP cameras use only pins 1, 2, 3, and 6 typically) so he wouldn’t need to run another cable (yay for laziness). Rather than re-terminate both ends of the extension I just cut one end and only put in the wires for 1, 2, 3, and 6 pins. Then it worked!

It doesn’t matter your title, if you are good at technical stuff it will always end up being your job.

Fable and GPT 5.5

Sat, 04 Jul 2026 10:08:20 +0800

Fable seems to be a great model in my testing. Currently I’m building a Mac desktop app, and I’m working hard to be faithful to the look that Apple has for their apps. Fable seems to be pretty good at it. But so does GPT 5.5 as long as I’m clear. Maybe my problems aren’t hard enough.

Fable has a tendency to use jargon a lot more in a dense, PhD student sort of way. It’s able to elegantly explain things though if asked. It shares the Opus tendency to jump in and get started even when something isn’t clear. I suppose that is good if you want to move fast but not great if you are a deliberate person.

Do I feel like Fable is a step change? Perhaps yes. But it still isn’t good enough for me to let it run on its own. Is the extra cost worth the extra intelligence, given that I still have to baby sit it for tasks? No, probably not.

GT Anywhere

Thu, 02 Jul 2026 11:46:54 +0800

My son started GT Anywhere yesterday, and offshoot of Alpha. After learning that by meeting his daily XP goals he gets points in their online store for students, he is extremely focused on completing lessons on the Timeback platform today.

He already has a savings account, a debit card, and cash in his Steam account. But this opportunity to get more points he could use to buy things (and compete on the leaderboard) is addictive for him. I’m hoping his motivation stays strong.

He also thinks the lessons are more fun than his old international school based on the IB system (he was in the PYP program).

Keep in mind it’s the summer and he would be happy spending his time flying his FPV drone, building custom game configs, or on Minecraft mods.

Mongolia Wants AI Data Centers. First It Needs Power.

Wed, 01 Jul 2026 10:26:26 +0800

Mongolian Prime Minister Uchral Nyam-Osor was recently on a panel titled “No Power, No AI” at the World Economic Forum in Dalian, China. He touted Mongolia as a place to invest in renewable energy. In the seemingly inevitable AI future, global power consumption will dramatically increase. I’m generally positive about this future, but I also believe governments (with the exception of China) aren’t really taking power demand seriously.

PM Uchral wants Mongolia to be a destination for digital infrastructure and renewable energy. In my view, this is partly an attempt to turn Mongolia’s biggest infrastructure weakness into an investment story: the country badly needs more power.

The PM’s pitch sounds good. Mongolia has land, sun, cold weather, and proximity to China. But AI data centers do not run on vibes. They need power, water, fiber, political certainty, and security. Mongolia has some of those things in theory. In practice, each one gets complicated pretty quickly.

These are comments made by PM Uchral, edited lightly for grammar and clarity:

If you want to work in Mongolia, you have a few advantages:

Mongolia is non-aligned. It’s a democratic country with a strong interest in becoming a trusted digital and energy partner.

Data sovereignty is one of Mongolia’s big advantages. Companies seek secure and strategically located data infrastructure.

Mongolia has significant land opportunity, with the government owning most of the land.

Mongolia is a landlocked country. Mongolia is a land “linked” country and is close to China. You can link to China.

After he said this, the moderator, Michael Wang from CGTN, asked a question:

Moderator: How fast do you think regulation and permitting might take? Uchral: It’s about the licensing?

Moderator: How fast is it? Uchral: It’s fast enough. We will try… My main goal is about red tape reduction.

Note: CATL Chairman Robin Zeng gave Uchral a round of applause for this.

I want to talk about each one of these supposed advantages and explain why they might not actually be advantages.

Non-alignment

Mongolia is non-aligned. It’s a democratic country with a strong interest in becoming a trusted digital and energy partner.

Mongolia is truly non-aligned. Its small size and location next to two serious powers, China and Russia, create an environment where Mongolia is everyone’s friend and no one’s enemy. This is great for economic ties. Mongolia’s famous Third Neighbor Policy aims to use states such as South Korea, Japan, and even the US and EU as a counterbalance to Mongolia’s neighbors.

Unfortunately, this desire to make everyone happy means Mongolia wants to keep its neighbors (and other third neighbors) happy. A few examples:

Mongolia’s main North/South rail corridor has been largely stagnant since 1991. It is still 50% owned by a Russian partner, and this partner refuses to allow expansion, connection to other rail lines, or real investment. An East/West rail corridor would significantly improve industry in remote Mongolian regions, but it is dead on arrival due to this situation.
In 2018, agents of the Turkish government abducted a director of schools associated with the Gülen movement in Ulaanbaatar. They succeeded in getting as far as the airport before being stopped by Mongolian authorities. The agents were allowed to leave, and while the school director was saved from abduction, the schools were later shut down at the request of the Turkish government.
The proposed 315 MW Egiin Gol hydropower plant has been effectively vetoed by Russia. Russia claims it would negatively affect water flowing to Lake Baikal.

These examples show that while Mongolia is indeed non-aligned, it is often at the whims of its neighbors, and sometimes its third neighbors. Mongolia certainly is interested in becoming an energy partner, probably because Mongolia’s electricity generation has lagged behind demand. For a data center investor, the issue is not whether Mongolia is friendly. It is whether Mongolia can protect a project when a neighbor, supplier, lender, or political ally decides it has a problem with it.

Source: Data.mn

Data sovereignty

Data sovereignty is one of Mongolia’s big advantages. Companies seek secure and strategically located data infrastructure.

Uchral’s premise here is that companies are seeking secure and strategically located data infrastructure. Perhaps he wants to make something like the digital equivalent of the “doomsday” seed vault in Svalbard. So what do companies want when it comes to data sovereignty? I would guess several things:

Reliable infrastructure to power data centers
Solid security to keep data secure
Affordable prices for renting servers or colocating servers
Easy ability to move data in and out

That’s about it. Here is how Mongolia stacks up:

Mongolia currently does not have enough electricity for its own consumption, and water in Ulaanbaatar, where nearly all skilled labor exists, is also a constraint. The OECD projects that Mongolia’s water demand could exceed current supply by 2040 or sooner, with scarcity likely to worsen in high-demand areas including Ulaanbaatar. The recent MCC-Mongolia Water Compact increased Ulaanbaatar’s available water supply by nearly 80%, which is good news, but it also shows the scale of the problem. Northern Mongolia has more water, but it lacks skilled workers. The large Oyu Tolgoi copper mine in southern Mongolia currently flies workers in and out of Ulaanbaatar on a shift schedule.
Mongolia’s only publicly listed Tier III data center, which guarantees reliability, security, and redundancy, is run by Khan Bank, Mongolia’s largest commercial bank. Mongolia’s government-owned National Data Center claims to have a Tier III rating but is not listed on the Uptime Institute’s website. Mongolia’s new national data center aims for a Tier IV rating, but will be located in the New Zuunmod area, which shares water resources with Ulaanbaatar.
Prices for publicly rentable servers in Mongolia appear at first glance to be in line with international pricing, until you look into the machine details and see that they are running servers based on chipsets that first appeared in 2010-2014.
According to the International Telecommunication Union (ITU), in 2023, Mongolia’s total bandwidth was 590 gigabits per second across all links (fiber and satellite). Utilization in 2023 was 66% in total, but during peak hours my understanding is that these links are nearly full. The projection was 750 Gbps in 2024, but even this wouldn’t be a lot for an AI data center. A gigawatt AI data center would need somewhere around 1 terabit per second (Tbps). One terabit per second is 1,000 gigabits per second.

This is not even mentioning the danger that Russia’s internet kill switch poses. It is unknown how a possible internet shutdown in Russia would affect the internet in Mongolia.

Significant land opportunity

Significant land opportunity, with the government owning most of the land.

Mongolia certainly has land, and PM Uchral is right that the state controls most of it. That makes land available in theory. Unfortunately, it does not make a gigawatt data center easy to build. A project of that scale needs power, water, fiber, roads, permits, local consent, and contracts that investors believe will still mean the same thing ten years later. Local consent from the herders who use the land is a real sticking point.

Mongolia can bring in outside expertise, and the Oyu Tolgoi mine is the obvious example. It also shows the risk. Oyu Tolgoi has gone through tax disputes, financing disputes, a 2022 “reset” with Rio Tinto, corruption investigations around earlier agreements, and continuing political pressure to renegotiate the terms. The mine is successful partly because the resource is enormous. A data center investor will have to ask whether power, water, and political patience are enormous enough too.

Oyu Tolgoi also shows how expensive “available land” can become once a project gets going. The mine spent millions of dollars on local development, including roads, schools, health facilities, power connections, water systems, and herder support programs. Under its 2015 Cooperation Agreement, Oyu Tolgoi contributes $5 million per year to the Gobi Oyu Development Support Fund for community projects in Umnugovi aimag. By 2019, that fund had invested $22 million in 179 projects, including schools, kindergartens, a health care center, a flood prevention dam, livestock disinfection, and other local programs.

Moderator: How fast do you think regulation and permitting might take? Uchral: It’s about the licensing?

Moderator: How fast is it? Uchral: It’s fast enough. We will try… My main goal is about red tape reduction.

The legal framework for this push into energy and data centers has not been finalized yet. Even when it is finished, I don’t believe there will be an answer to how long a project would take from the start of the permit process to breaking ground.

Mongolia is land “linked” to China

Mongolia is a landlocked country. Mongolia is a land “linked” country and is close to China. You can link to China.

This is perhaps Mongolia’s greatest advantage, and also its greatest risk. First, proximity to China means easy access to many of the technologies required for data centers, including networking, cooling systems, solar, and newer battery technologies like sodium-ion batteries from CATL. The main exception is state-of-the-art AI chips, particularly from Nvidia. Chinese chips are indeed catching up, but they aren’t there yet.

Chinese AI companies are currently competing by being more efficient via distillation, or by increasing the number of Chinese AI chips, which are less efficient, thereby increasing power use. In China, increased power use is not as much of a problem given that China’s power production is increasing 9.7% per year (compared to Mongolia’s average of 5% per year since 2011, with demand outstripping supply by 30% in 2024).

China also purchases nearly all of Mongolia’s current exports, which are mostly mining exports. Yet China would not be a realistic consumer of data center output in Mongolia. An optimist would say that Mongolia, with its massive solar resources (about 250 sunny days a year), is perfect for solar. Mongolia’s cold, previously a big negative for battery storage, isn’t as much of a problem with new sodium-ion battery technology, which retains 90% capacity down to temperatures of -40° C. Yet it is hard to understand why a Chinese company would build a data center in Mongolia, as opposed to at home, where it would benefit from low-interest government loans.

I’m generally positive about Mongolia’s ability to be energy independent in the future (it isn’t currently). One thing holding domestic energy production back is price. Mongolia’s current energy prices (256 MNT/kWh, or about $0.07 USD/kWh on average for homes) result in domestic energy production being unattractive for private producers. China also keeps energy prices low through a mixture of state investment and cross subsidies. CATL recently announced a battery storage facility planned for Mongolia in the range of 100-400 MWh. This represents 4-14% of Mongolia’s energy production shortfall in 2024. For comparison, for a 100 MW AI data center, Mongolia would need more than 2,000 MWh of battery storage for that data center alone.

This isn’t a problem for China, given massive state investment in Chinese energy production. Mongolia does not seem willing to put massive state investment into power generation. At current prices, according to my own back-of-the-napkin math, a solar PV + battery power plant has a payback period of 20+ years, beyond the life of the batteries themselves, making this proposition unprofitable.

If the end consumer of this data center capacity is not China, but someone else, there is always the geopolitical risk to consider. With all internet fiber lines leaving Mongolia running through either China or Russia, and with close physical access to Mongolia, running state-of-the-art AI data centers in Mongolia would be a risk to both intellectual property and, given potential export controls on frontier AI models, to the companies operating them.

Mongolia’s own National Data Center was hacked in 2018 by suspected state actors and was discovered by an outside research lab, which notified the Mongolian government.

Poor Mongolia, so far from the ocean, so close to China.

Mongolia may eventually become a serious energy exporter or data infrastructure hub. I would like to see that happen. But the current pitch skips over the hard part: Mongolia is still trying to solve the same power, water, and connectivity constraints that an AI data center would immediately make worse.

A ton of copper concentrate is a commodity. AI may be the new electricity, but companies and states care where this new power is generated. Mongolia has not yet proven its ability to be a good steward of investor rights or a free market.

Are Mongolian Courts Biased? 75,000 Criminal Cases Say Mostly No

Thu, 12 Mar 2026 08:00:00 +0800

But your employment status still matters more than it should.

International criminology research has consistently found that demographics influence criminal sentencing. In the United States, Sonja Starr’s 2012 study found that men receive 63% longer federal sentences than women, even after controlling for the offense. Studies from Russia, Poland, and elsewhere have documented similar patterns with gender, education, and socioeconomic status. The general finding across countries: who you are affects how you’re sentenced, even when it shouldn’t.

Mongolia publishes all of its court decisions online at shuukh.mn, a public database maintained by the judiciary. Every criminal case, with the full text of the decision, is there for anyone to read. I’ve lived in Mongolia since 2012 and have spent a lot of time digging into Mongolian data for this blog. So I wanted to test a simple question: does the same pattern of demographic bias show up in Mongolian criminal courts?

I collected and analyzed 75,323 decisions from Mongolia’s Criminal Court of First Instance, covering 2020 to early 2026. Let’s take a look at what the data says.

What’s in the data

Each case on shuukh.mn is a full court decision written in Mongolian. Here’s an example: a 41-year-old male herder from Tuv Province, primary education, family of 5, no prior record. He stole 700,000 MNT (about $200 USD) from an acquaintance, pled guilty, paid restitution, and was fined 500,000 MNT.

That single decision contains most of the variables I extracted: demographics (gender, age, education, employment status), the crime (Criminal Code article, category), and the outcome (sentence type, amount, aggravating and mitigating factors). Not every decision has every field. Age, for example, is missing from 42.7% of cases. But across 75,323 decisions, the coverage is good enough to work with.

Here are a few more cases to give you a sense of the range:

Case 84144: A 39-year-old employed woman in Nalaikh (Ulaanbaatar), secondary education, convicted of seriously injuring her spouse with a knife while intoxicated. 3 aggravating factors, 5 mitigating factors. Sentenced to 41 months imprisonment.
Case 44606: A 24-year-old self-employed man in Khan-Uul district, secondary education, convicted of a traffic offense that caused serious injury. No aggravating factors, 4 mitigating. Given a 12-month suspended sentence.

The defendants are overwhelmingly male (86%), with a median age of 34. Most have a secondary education.

What Mongolian criminal justice looks like

The picture is pretty different from what you’d see in the U.S. or Europe.

Fines are the dominant outcome. Out of 75,323 cases, 42,517 (56%) resulted in a fine. Imprisonment is a distant second at 11,859 cases (16%), followed by community service, suspended sentences, and probation. Mongolian courts are not sending most convicted defendants to prison.

Over half of all cases (53%) involve violent crime, which under Mongolia’s Criminal Code covers chapters 10 through 13 (assault, domestic violence, robbery, etc.). Property crime is the next largest category at 25%, followed by traffic offenses at 13%. Drug cases are relatively rare at just 2% of the total.

I converted all sentence types to a common scale of “month-equivalents” so I could compare across fines, imprisonment, and other sentence types. Fines were converted using the Criminal Code’s own formula (15,000 MNT per day, or about 450,000 MNT per month). This gives us a single severity measure that works across the full range of outcomes.

The employment surprise

In the international literature, gender and race are the usual suspects for sentencing bias. Mongolia doesn’t have the racial diversity that drives disparities in places like the U.S., so the question here is about gender, education, age, and employment.

Employment status turned out to be the standout finding. It’s the only demographic variable that survives rigorous statistical correction across all model specifications.

Unemployed defendants receive an average of 13.9 month-equivalents in sentence severity, compared to 6.8 for employed defendants. That gap of about 7 months in the raw data shrinks to 3.7 months after controlling for crime type, criminal history, and aggravating and mitigating factors, but it remains highly significant (p < 0.001). It’s consistent across every model I ran: OLS, multilevel, log-transformed, different sample definitions. Employment status is the one demographic variable that keeps showing up.

To put that 3.7-month effect in perspective: the average fine in the dataset is about 1.3 month-equivalents. So the employment penalty is roughly equivalent to moving from a typical fine to a sentence that’s nearly three times as severe. Employment status is not a legally prescribed sentencing factor in Mongolia’s Criminal Code. It shouldn’t matter. But it does.

Gender: more complicated than it looks

Gender is where a simple analysis would give you the wrong answer.

If you run a straightforward regression with all 29,847 complete cases, the gender coefficient is small and doesn’t reach statistical significance (p = 0.074). You might conclude that Mongolian courts treat men and women roughly the same. That would be partially right and partially wrong.

A two-stage model tells a different story. In stage one, looking at whether a defendant is imprisoned at all (N = 40,626), women are 41% less likely to be imprisoned than men with similar cases (odds ratio = 0.59). That’s a large effect. But in stage two, looking only at defendants who were actually imprisoned (N = 5,578), women receive sentences about 4 months longer than comparable men (p = 0.003).

This is a classic Simpson’s paradox. The overall average hides two opposite effects that cancel each other out. Women get lighter treatment at the imprisonment decision stage, but the women who do end up in prison tend to have committed more serious offenses (they had to clear a higher bar to get there), and they receive correspondingly longer sentences.

So gender does matter in Mongolian sentencing, but it operates through the imprisonment decision, not through sentence length.

What actually drives sentencing

The most reassuring finding from this analysis is about what happens when you decompose what’s actually driving sentencing variation. I ran a sequential regression, adding blocks of variables one at a time to see how much each group contributes.

Aggravating and mitigating circumstances (things like whether the defendant confessed, compensated the victim, or committed the crime while on probation) explain 16.2% of sentencing variance. Demographics explain 3.7%. Crime type adds 3.1%. Court effects and year effects are basically noise at 0.5% and 0.2%.

In other words, the legally prescribed sentencing factors explain more than four times as much variance as all demographic variables combined. Mongolian judges are, for the most part, following the law. The system isn’t perfect (that employment gap is real), but it’s not driven by demographics in the way that research from other countries might lead you to expect.

What the data doesn’t tell us

I should be upfront about the limitations. Age data is missing for 42.7% of cases, and that rate is getting worse over time (32.5% missing in 2020, 61.5% by 2025). More recent court decisions appear to have shorter biographical sections. The primary analysis uses complete cases only (29,847 of the 75,323 total), though the results hold up when I relax the age requirement and use a larger sample of 36,829.

The severity measure for non-imprisonment sentences (community service, suspended sentences, probation) is approximate. I used the fine-to-months conversion from the Criminal Code, but reasonable people could argue for different conversion rates. I ran sensitivity checks with both conservative and liberal conversion rates, and the key findings hold.

And of course, there are things the data simply can’t capture: quality of legal representation, the specifics of each case beyond what’s coded, and whether plea bargaining plays a role that isn’t visible in the final decision text.

What this tells us

Mongolia’s criminal courts are not primarily driven by demographic bias. Legal factors dominate sentencing, and court-to-court variation is minimal (courts explain less than 1% of variance after case-level controls). That’s genuinely good news, and it’s worth saying clearly.

But the employment gap is a real problem. A 3.7-month penalty for being unemployed, after controlling for everything else, means that economic status is influencing outcomes in ways the law doesn’t intend. Whether this reflects conscious bias, unconscious assumptions about defendants’ “stability” or risk of reoffending, or some unmeasured confound, the data can’t say. But the pattern is consistent and substantial.

The fact that we can even have this conversation is thanks to shuukh.mn publishing every court decision. That level of judicial transparency is more than most countries offer. Mongolia doesn’t always get credit for the things it does well in governance, but open court records is one of them.

If you’re interested in the details, the full analysis code and notebooks are available on GitHub.

Mongolian Meat Prices, Seven Years Later

Thu, 12 Mar 2026 08:00:00 +0800

In 2019 I predicted meat prices would drop. They did. Then things got complicated.

Back in May 2019 I wrote about Mongolia’s meat prices, dug into the data on exports, animal losses, and inflation, and even made some forecasts about my predictions on meat prices going forward. At the time beef was 10,904 MNT per kilogram in Ulaanbaatar and mutton was 9,777 MNT. I argued that exports weren’t the main price driver, that animal losses were, and that meat was actually getting more affordable relative to wages.

Seven years is a long time. Since then we’ve had a global pandemic, border closures with China, and the worst dzud since 2010. I wanted to revisit the original analysis with fresh data from the National Statistics Office and see what held up and what didn’t.

The roller coaster didn’t stop

As of February 2026, beef in Ulaanbaatar costs 24,581 MNT per kilogram. Mutton is 19,181 MNT. That’s an increase of 125% for beef and 96% for mutton since May 2019.

But the path there wasn’t a straight line. After I published the original post, prices did come down through the summer of 2019 (my forecast actually got that right). Then COVID hit in early 2020. Mongolia closed its borders, including with China, and exports dropped. Prices stayed relatively flat through 2020 and into 2021.

The real acceleration started in 2022. A combination of post-COVID inflation, rising animal losses, and a weakening tugrik pushed prices up fast. By 2023 both beef and mutton were roughly double their 2019 levels. And then the dzud hit.

Exports: still not the main driver

In the original post I argued that meat exports weren’t the primary driver of prices. The data since 2019 only makes that case stronger.

COVID cratered exports. In 2020 meat exports dropped to 18,769 tons from 33,193 in 2019. In 2021 and 2022 they stayed below 10,000 tons. Yet prices kept climbing.

Then exports bounced back in 2023 to 34,194 tons (nearly matching the 2018 peak of 34,887 tons) and fell again to 24,736 tons in 2024. The correlation between exports and prices remains weak. Prices went up when exports dropped. Prices went up when exports rose. The relationship just isn’t there.

The dzud came back

This is the big story. In 2024, Mongolia lost 9.4 million adult animals. That’s the worst year since 2010’s devastating dzud when 10.3 million died.

But 2024 didn’t come out of nowhere. Animal losses had been creeping up for years: 2.6 million in 2018, 3.0 million in 2021, and then 4.9 million in 2023 before the 9.4 million in 2024. Two consecutive years of heavy losses (14.3 million animals across 2023 and 2024 combined) is something Mongolia hasn’t experienced in recent memory.

The total national herd dropped from about 71 million animals in 2019 to 57.6 million by the end of 2024. That’s a loss of roughly 13 million head in five years, or about 19% of the herd.

Looking back at the original post, I wrote that the 2.6 million animals lost in 2018 was “a very likely explanation for rising prices.” The 2023/24 losses were more than five times that number across two years. The connection between animal losses and prices remains the strongest signal in this data.

Meat prices finally outpaced inflation

This is where things changed from the original analysis.

In 2019 I showed that meat prices had been remarkably stable relative to inflation since 2011. For years, actual prices tracked below what you’d expect if they simply followed CPI. I interpreted this as meat getting cheaper in real terms.

That story flipped around 2022. Both beef and mutton prices broke above the CPI trendline and haven’t come back. Beef at 24,581 MNT is well above the 14,000 or so MNT you’d expect from CPI alone. Mutton tells a similar story.

The combination of back-to-back dzuds, post-COVID inflation (13.8% in 2021, 13.2% in 2022), and a weakening currency appears to have broken the pattern. Meat prices are now outpacing inflation.

Are Mongolians still spending less on meat?

Despite the price increases, the answer is still mostly yes.

I used the same methodology from the original post: a family of three in Ulaanbaatar, each person eating 250 grams of meat per day (roughly in line with average US consumption), earning the average UB salary.

In 2011 that family would have spent about 22.5% of their income on beef (or 19.7% on mutton). By 2019 it was 19.1% for beef and 16.0% for mutton. In 2024: 16.9% for beef and 13.7% for mutton.

The trend held. Wages in Ulaanbaatar have grown faster than meat prices. The average monthly wage went from 470,300 MNT in 2011 to 2,390,700 MNT in 2024, an increase of over 400%. Beef prices increased about 300% over the same period.

There was a noticeable bump in 2022 when wages hadn’t caught up with the post-COVID price spike, but by 2024 the long-term downward trend had reasserted itself. For the average UB household, meat consumes a smaller share of income than it did a decade ago.

The Tsuivan Index, revisited

This was a reader favorite from the original post. I created a price index using December 2010 as the base (100) for several common goods and services tracked by the NSO.

The results are pretty wild. Women’s haircuts have increased 885% since December 2010 (index of 985). Men’s haircuts are up 727% (index of 827). Both have far outpaced beef (505%, index of 605) and mutton (417%, index of 517). Canteen food (the NSO’s current category closest to tsuivan) sits at 494.

In the original post I noted that haircuts were the most surprising price increase. Seven more years of data just made that gap wider. Haircuts continue to outpace everything else in this basket by a large margin. This makes some intuitive sense: haircuts are pure labor, and labor costs in Ulaanbaatar have been growing fast. You can’t import a cheaper haircut from China.

What I got right and what I got wrong

Right: My forecast predicted prices would drop after the spring 2019 peak. They did. Animal losses remain the strongest predictor of price movements. Exports still don’t correlate well with prices. Meat has continued to get more affordable as a share of income.

Wrong (or at least different now): The inflation comparison reversed. In 2019, meat prices tracked below CPI. By 2026 they’re well above it. The 2023/24 dzud was a supply shock large enough to break the pattern that had held for a decade.

The question I raised in the original post about government reserve meat remains as relevant as ever. With 14.3 million animals lost across 2023 and 2024, the pressure on herders and on consumer prices is real. Whether the government’s reserve meat program can meaningfully buffer against shocks of this scale is still not clear from the available data.

All data in this post comes from Mongolia’s National Statistics Office (1212.mn).