The phone is a multi-mode device that switches state and scene frequently: one moment it turns into a shopping mall for you to shop in, the next into a café for you to chat and read celebrity gossip. One of the core technologies of computing devices like the phone is multiprocessing and multithreading, but the human habit of allocating time is not time-sliced polling.

Imagine a scene from before the internet: I want to read a book, so I have to set aside a particular stretch of time, find a quiet place, take a paper book and read it — maybe even make a special trip to the library to find the book. If while reading I hit a passage I find especially interesting and want to share it with a friend, I can only memorize it or note it down, and then when I have arranged time to meet the friend, tell them about it. Here, reading and socializing are separate and independent.

Now it is very convenient. I read with the phone beside me, see something interesting, snap a photo of it at once, and send it to a friend. Very convenient, right? Here you have actively mixed reading and socializing together. Worse, you may switch without noticing to short videos, or to Taobao and Pinduoduo. The moment you pick up the phone you may be drawn in by shorter, flatter, faster information, and it takes a lot of willpower to finish the reading. The human attention mechanism means we are in fact poorly suited to switching states this often.

The earliest smartphones were called PDAs — personal digital assistants. After decades of development, smartphones have completely failed to deliver the originally expected effect. Instead of becoming our personal assistants, phones became the field where the big vendors run their horses over our attention; we must respond to others at any moment, or hunt for dopamine on the phone at any moment.

AI is advancing by leaps and bounds now, and I think this time we may have a shot at really turning the phone into a personal assistant — one that, according to our needs, presents a different state in different scenarios to help us rather than distract us. But who knows whether things will just get worse. At least before that ideal state arrives, we still have a lot to do ourselves to manage our own attention.

Going back to single-mode devices

Going back to pen and paper — current AI is a big help here. When reading and taking notes, for instance, you can stay entirely away from the phone and the laptop and go back to pen and paper. After you finish the book, take out the phone and photograph your notes; today’s AI can recognize what you wrote quite accurately. That way we avoid reaching for the phone at every turn without giving up the convenience of digitization altogether.

Or: the bedroom is for sleeping, so you probably need a more traditional alarm clock, instead of using the phone to check the time and set alarms. Then the phone can get out of the bedroom.

Make good use of screen-time management apps

Phones and computers today have plenty of time-management software — iOS’s Cape, for example. Such apps can generally group your apps and disable groups during different time windows. That gives us a rough version of scene switching: put entertainment apps in a group and enable it only during specific periods. To be honest, though, with today’s phone operating systems this area is still very crude, far from actually reaching the level of scene switching. Third-party screen-time apps can only paper over the cracks. But at least it is an initial attempt. I think this should be a direction for how phones and phone software evolve in the coming intelligent era: software and features should appear only when I need them.

Turning off feature modules inside apps

Some apps do a bit better here. WeChat, for example: in its Discover page, Channels and Moments can both be turned off according to your needs; Alipay likewise lets you turn off its short-video module. But WeChat and Alipay still have a vast number of features you cannot turn off. Most apps, especially domestic ones, are actually terrible at letting users customize on demand, and that is probably determined by their business models. Every app can scroll short videos; every app can push high-interest loans. We ought, while our heads are clear, to be able to decide which features of an app we need, but reality is grim; we can only dance in shackles.

Exploring rebuilding the information feed with “Lobster”

All the so-called personalized recommendation systems today do not actually recommend according to your wishes. They amplify the fragile, lazy part of your nature and optimize for metrics like click-through rate. In this era of fast-developing agents like “Lobster,” there is in fact an opportunity to rebuild the information feed. While our heads are clear, we can tell our own agent what kind of thing we want to see, and let it do an initial round of filtering, playing the role of a real assistant. In theory that should greatly relieve information overload. Some companies are already exploring this direction — Bluesky’s Attie, for example.

Moutai was once the “political correctness” of state banquets; today it slips quietly into family dinners, weddings and gatherings of friends. One bottle of liquor, three chapters of a life: from government liquor to business liquor to family liquor. Moutai’s thirty-year transformation was not a simple change of channels but a self-evolution forced by the macroeconomic cycle — from “a scarce resource for institutions” to “genuine consumption by individuals,” from “selling liquor” to “selling a way of life.”

According to Kweichow Moutai’s own historical records and a review of its financial reports, government procurement once accounted for nearly 50% of sales. After the Eight-Point Regulation at the end of 2012, the company put forward “three transformations and five shifts,” and business settings rose fast. From 2021, with the property cycle at a trough and consumption weak, Moutai accelerated its push toward the consumer. iMoutai launched in 2022, and the direct-sales share climbed quickly from its low in 2018; the company explicitly committed to being “consumer-centered and market-oriented,” focusing on three transformations — customer base, settings, service — and rolling out a coordinated model of “direct sale + distribution + agency sale + consignment,” pushing channels further down into household and private consumption.

I. The era of government liquor: the national drink’s “political endorsement” (1950s–2012)

In 1951 local government merged three private distilleries into the Moutai distillery, and Zhou Enlai made it the liquor of state banquets. When Nixon visited in 1972, Moutai became a diplomatic gift; for National Day and military parades it was flown in for the occasion. Early sales ran mainly through state-run sugar-and-liquor companies and group purchasing; government and state-owned-enterprise purchases accounted for nearly 50%, and the wholesale price climbed with high economic growth, exceeding 2,000 yuan before 2012.

In those years what Moutai poured was not only liquor but status. The brand premium came from a “political halo”: scarcity plus endorsement, and terminal demand that could not be met. Company filings and historical reporting both show sales growing at 44% in this period, entirely dependent on institutional, government settings.

II. The anti-corruption shock and the forced rise of business liquor (2013–2021)

The Eight-Point Regulation at the end of 2012 landed like a storm. Spending on the “three public expenditures” collapsed, and Moutai took a direct hit. In 2013–2014 the share price halved from its high (a drop of about 50%), the wholesale price fell from above 1,200 yuan to a low of roughly 830, and sales growth collapsed from 44% to 2%. For the first time the company’s annual report admitted to “unprecedented pressure.”

Crisis forced the transformation. From 2013 Moutai advanced “three transformations and five shifts”: from government settings to high-end business ones — corporate banquets and gifts. Distributors expanded rapidly and restaurant accounts became the main channel; after 2015 the company stopped adding distributors at scale and turned to e-commerce and to pushing its series liquors down-market. Business settings replaced government ones as the main source of growth.

After 2018 the share price reclaimed its throne as “king of stocks.” The record shows that business liquor taught Moutai to survive by the market: from an iron rice bowl to a market rice bowl. Channel profits were squeezed, but a far broader institutional space opened up.

III. The metamorphosis into family liquor: breaking through to consumers in the property slump (2022 to today)

From 2021 Chinese property entered a downcycle, high-end consumption weakened, and Moutai’s wholesale price came under pressure. In the first three quarters of 2025 revenue grew only 6.32%, and the full-year target was cut to single digits for the first time. In March 2026 the wholesale price of Feitian briefly fell below 1,600 yuan, a multi-year low.

The company moved decisively to go “one step to the consumer.” iMoutai launched in March 2022, and direct-sales revenue jumped 105.4% that year. In 2024 direct-channel revenue reached RMB 74.843 billion, or 43.87% of the total, at a gross margin of 95.33% (far above the 89.42% of wholesale agency); iMoutai alone took in RMB 20.024 billion for the year. In the first half of 2025 the direct share held steady at around 44.8%.

Moutai then upgraded its strategy across the board, issuing a market-based operating plan and applying an “agency sale” model to personalized products (aged Moutai, premium Moutai, zodiac bottles, Guyue Feitian, small-format Feitian and others), formally establishing the new coordinated structure of “direct sale + distribution + agency sale + consignment.” Under agency sale, title does not transfer and the distributor need not prepay in full — only post a deposit — and reaches consumers through online retail, physical stores, restaurants and private channels, earning a service fee of about 5% (a commission). A 500ml bottle of 53-degree premium Moutai, for example, retails at 2,299 yuan, of which the distributor earns about 115 yuan. Every agency-sale product must be sold through the iMoutai platform; former group-buying clients can scan a store-specific QR code, ordering online and collecting in store, which mainly serves retail customers.

A Moutai executive put it plainly: “We must change of our own accord, serve our channel partners well, and help them shift roles — from the traditional job of moving goods to carrying Moutai’s offline display and service functions, so that company and distributors together serve the consumer.” The move takes back pricing power, steadies prices on non-standard products, and curbs channel speculation, while turning distributors from merchants who sit and wait into merchants who go out — asset-light, service-heavy, and genuinely working the end market.

The shift shows most clearly in the settings: from business banquets to family dinners, weddings, birthday banquets and gatherings of friends. Stores have set up “family banquet corners,” the iMoutai share of users keeps rising, and consumers report using Moutai more often at family gatherings. The post-90s and post-85s cohorts are becoming the main force, with buy-and-drink-now and small formats for personal use the emerging pattern. Moutai moves from “the glory of government liquor, the survival of business liquor” to “the homecoming of family liquor” — a glass that keeps real life company.

IV. The logic beneath it, and what it teaches

Moutai’s three transformations are, at bottom, the macro cycle forcing a micro one: anti-corruption weaned it off government dependence, and the property slump catalyzed its awakening to consumers. The company moved from policy dividend to consumer sovereignty, completing a return to commercial logic. The agency-sale model is the latest footnote to that shift — it turns channel profit from speculative to stable and forces distributors to work the consumer end rather than hoard and flip. Analysts note that distributors with real consumer-facing capability and community networks will win more quota, while the old pure goods-flipping model is cleared out as a matter of course.

For the industry this is a template. Baijiu makers must embrace digitalization and scene-building early and compete for younger customers and private consumption. Whether Moutai’s “family liquor” strategy succeeds will turn on the bottle-opening rate and its penetration among younger drinkers. So far execution has been firm: iMoutai is the core consumer lever, and both the wholesale price and the share price are steadying through the transition.

V. Closing

Government liquor was glory, business liquor was survival; family liquor is where it belongs. Moutai’s transformation is not only a corporate story but a mirror of the times — from a symbol of power to a companion of daily life, reflecting the micro-level upgrade of Chinese consumption from institutions to individuals. One hopes for more innovation still, so that the national liquor’s fragrance truly reaches ordinary households.

On your own dinner table, which kind of liquor is Moutai? Comments welcome.

In this year’s annual report I called Coca-Cola and Gillette invincible companies, and described how great they are — great businesses run by outstanding managers.

But I think you are best served by not treating my view as an unconditional recommendation to buy these companies.

Otherwise you may pay too high a price for them — at least in the short run.

There is always a risk: it can take a business years to grow into the price you paid. Even the greatest business can see its share price run ahead of its business performance.

So Buffett holds these companies and yet does not recommend buying them.

How, then, should we read this? I think the argument never ends because two key elements get overlooked: transaction friction costs and position sizing.

That holding equals buying is a fact rests on one basic assumption: that buying and selling shares carries no friction cost. Under that assumption we can run a simple thought experiment and come away with an extremely important perspective on jettisoning sunk costs. Suppose buying and selling cost nothing, and every evening after the close the system automatically sells your shares at the closing price, and before the open the next morning you must decide what to buy — what do you buy? If your decision was to hold the portfolio unchanged, then you should buy back exactly the shares that were sold, in exactly the same quantities. That is the whole meaning of holding equals buying.

On the effect of transaction friction, Munger gave a fine illustration at the 2023 Daily Journal annual meeting: a shareholder asked why Berkshire was bearish on bank stocks while the Daily Journal kept holding its bank stock. Munger said the Daily Journal’s bank shares were bought at a very low cost and carried a large unrealized gain; sell them, and 40% of the proceeds would go to the government as capital gains tax. So he would rather not sell.

Even without a capital gains tax, commissions alone bias you toward holding for the long term. With no friction cost at all, any small new development at any company we hold or watch — a change in fundamentals or in price — shifts our win rate and our payoff odds, and in theory justifies a small rebalancing. Once friction costs enter the picture, we rebalance only when fundamentals or prices move materially.

The “buy” in “holding equals buying” really refers to rebuilding the position under ideal conditions, not to adding to a position already held. Adding to a holding, or trading other positions in a way that shifts this holding’s relative weight, both amount to a new view of the existing portfolio.

So in the end, to settle whether holding equals buying in substance, we have to be clear about what holding is and what buying is — and those words carry different meanings in different contexts and in different mouths.

Yet more important than parsing the sentence: a value investor should learn to discard the pull of cost basis on his thinking and audit his holdings constantly. He may not trade often, but at the level of thought the comparison of assets never stops, because money never sleeps — and investing, at bottom, is nothing but the comparison of opportunity costs.

The same stock is a different object in different eyes. To some it is a claim on a company’s dividends; to others it is a chip at the gaming table.

The dividend layer

The bottom layer of a stock is sharing in the profits. Historically, companies have made money and been happy to share it with shareholders through dividends or buybacks. Kweichow Moutai and Tencent Holdings are companies of this kind. The risks here are:

1. The company is not really making money; it only looks like it on the statements
   1. Profit with no cash flow;
   2. Risk deferred — banks and insurers, for instance, may look profitable today while concealing enormous bad-loan or claims exposure;
   3. Accounting fraud
2. The company makes money but does not share it with minority shareholders
   1. No dividend, buybacks only at inflated prices, and the cash flung at random investments
   2. Management siphons value out of the company through related-party dealings of every kind
3. The company used to make money and is now in decline, with no more to be made
   1. The company decays and loses its competitiveness
   2. It is disrupted from above and its industry turns into a sunset industry

At this layer we must put a price on the earnings and the dividend, and judge whether they can last.

The story layer

Building on the eventual sharing of dividends, the market reaches a consensus that a company will make a great deal of money in the future. That usually gets told as a coherent story: EVs will replace combustion cars, and Xiaomi will become an EV giant; AI is rising, and Baidu will be great again; Pinduoduo will use China’s vast manufacturing capacity to find markets for goods and own the “save money” mind. For such companies every first-layer risk is present, and there are more besides:

1. The story is right, but the company is not the protagonist — AI really is rising, and it may have nothing to do with Baidu;
2. The story is simply wrong — EVs replacing combustion cars may not be a real trend at all.

At this layer we must price the story, and the difficulty rises by perhaps an order of magnitude. There is some venture-capital logic in it.

The gambling layer

One level above the story, we sometimes need not believe the story at all; we need only believe it is the hot speculation of the moment. That is the gambling layer — a casino needs something to bet on. Under gambling logic the risks of the first two layers cease to be risks. There is only one:

1. Hot themes rotate at any moment, and what the market will be betting on next is impossible to predict. At this layer there is no way to price anything at all; perhaps only high-frequency quant funds find an edge.

As an investor I care only about the first two layers, and I want most of my position in the first and a small part in the second. Stay away from the third.

Because the book describes the American situation, much of its content is only useful at the level of principle; the specifics need to be adapted to China’s actual conditions. So I only flipped through it roughly. But the 15 principles of personal finance it distills run through the whole book and are strongly universal. Let me excerpt and list them here, with a short note on my own understanding.

1. The trade-off between risk and return: risk and return are generally positively correlated, so a financial advisor who pitches high returns with low risk can basically be defined as a fraud. But it isn’t that absolute — through superior knowledge and information asymmetry, low risk with relatively high return is possible; and conversely, a low-return product can also be high-risk, as with some online lending platforms.
2. The time value of money: you can’t understand investing without understanding compounding. In the Chinese way of lending money to friends, you borrow 10,000 yuan ten years ago and repay 10,000 ten years later — at a 7% annualized return that’s basically doubled.
3. Diversification helps reduce risk: as the saying goes, when the east is dark the west is bright. But pay attention to the correlation between asset classes.
4. Not all risk is the same: diversification cannot eliminate systemic risk. If the whole economy collapses, whatever stocks and bonds you bought are basically finished.
5. The source of competitive investment markets: through the bidding mechanism, markets are generally efficient, and beating the market is extremely hard. The line “the pig thinks so too” is exactly this principle. I often hear people say they hope to earn 50% to 100% annualized in the stock market; I can only wish them a happy gamble.
6. The effect of taxes on personal financial decisions: the casino operator is the one who makes money.
7. The importance of liquidity: high liquidity commands a premium, low liquidity a discount. Generally keep 3–6 months of living expenses; for T+0 or T+1 you can put money into money market funds, reverse repos, and the like.
8. Without a plan nothing gets done: I started doing an annual financial review and plan at the end of 2015, and the effect is obvious.
9. Knowledge is the best protection: don’t be the chives.
10. Protect yourself from major disasters and calamities: pure protection cover is the essence of insurance.
11. The time scale of investing: stocks should be held at least 5 years; Bitcoin I’ll hold at least 20.
12. The principal-agent problem — watch the sales pitch: I find reading the contract rather than the ad more reliable.
13. Pay yourself first: when money comes in, think first of saving, of investing, or of improving yourself.
14. Money isn’t everything: a correct platitude.
15. Get started! The earlier the better with investing; if you haven’t started before, start today.