US stocks: After the shock, the music continues?

As the negative news of the earnings season is gradually released, Chinese stocks may have a chance to recover in the short term with the help of the external US dollar interest rate cut.

After a series of accidents such as Nvidia Blackwell delay, giant shock, employment decline, and yen interest rate hike, the market looks back carefully and finds that the US economy has its own operating inertia, not "clear sky in the last second, stormy in the next second".

In the last strategic weekly report "US stocks burst "ghost stories", there is no bottom line for the drop?", Dolphin Jun also said that the employment and unemployment data in July do not need to be too serious because of weather disturbances. The adjustment of US stocks is more like a combination of various "ghost stories" against the background of continued rise, taking the opportunity to kill valuations. After the killing, it is highly likely that as long as the economic situation does not take a sharp turn for the worse, but takes the path of a soft landing, then the start of the interest rate cut will reduce the risk of US stocks.

And the subsequent economic data in July, whether it is prices or retail sales, are pointing to a soft landing of the economy, not a recession.

1. Consumption locomotive: slowdown, not collapse

When the June retail sales of the United States came out, the seasonally adjusted month-on-month total was negative growth. Some people did not look at the subdivided structure and directly called the U.S. consumption locomotive out of business. As a result, they were slapped in the face in July, because the month-on-month growth of automobile and parts retail, which accounted for the largest category in retail sales and had a roller coaster monthly fluctuation, was greatly pulled back.

Not only that, other optional consumption such as 3C, electrical appliances, building materials, gardening, and even catering, which has been sluggish in recent months, have begun to return. The growth rate of essential consumption such as food and beverages, medical health, and daily necessities is also accelerating this month.

If the monthly fluctuations are smoothed out and the changes since the beginning of the year are looked at, it is still very obvious that rigid demand is resilient, while optional consumption such as automobiles, furniture and home furnishings, sports hobbies are generally poor, and catering is gradually weakening. The trend of consumption growth slows down, rather than a flash crash or collapse.

2. Another month of "good news" on inflation

The data in June made people worry whether inflation was over-hit, but in July, as housing costs and prices returned to 0.4% month-on-month, the core CPI rose from 0.06% to a safe range of 0.17% - three consecutive months of 0.2% core price month-on-month growth should give the Fed enough confidence to cut interest rates.

In particular, this wave of CPI decline seems to be more sustainable: in addition to the two high-volatility categories of food and energy, which have also leaked inflation, and the continued negative growth of commodity prices looks more like deflation, core services (excluding housing costs) are also falling significantly. For example, the more critical medical prices and transportation prices seem to be on a month-on-month decline in the past three months.

And what is relatively uncertain here, in Dolphin Jun's opinion, is still some price sectors with a heavy "human cost" content, such as education, other personal services, garbage removal, etc.

However, with the current rapid increase in labor supply and the gradual rise in unemployment, the wage growth rate is also returning to a gradually controllable range. The nominal wage growth in July was only 0.2%, which is also within the safe range of monthly growth.

The inflation data for the whole of July put together just right to point to the path of a soft landing for the US economy in the future, rather than a direct recession.

At the same time, combined with the various trading platforms covered by Dolphin Jun this earnings season, whether it is Amazon's revenue guidance, Airbnb's quarterly performance and guidance, or even Uber's implied weakening of North American orders, they are actually confirming that consumption and economic growth are gradually weakening, but these weakenings are supported by healthy resident balance sheets, more like a slow and natural process, rather than a sudden storm.

From the micro-entity point of view, the US economy is also a "soft landing", rather than no landing, or a hard landing.

3. After the major adjustment, what are the opportunities for Chinese and US stocks?

Under the soft landing path, the growth rate of EPS on the numerator side is slowing down, but the risk-free interest rate expectations of the denominator are also declining under the expectation of inflation and interest rate cuts. After the recent social retail and price data, the market has slightly revised up the expectation of interest rate cuts. The current expectation of interest rate cuts is two to three times this year, which is also within the basic reasonable expectations. To a certain extent, at least the interest rate cuts this year have been fully traded, and the high valuations under the previous optimistic expectations have also converged.

In this case, Dolphin Jun is more willing to look for rigid assets with relatively high sensitivity to discount coefficients and relatively slow slowdown in the numerator, or with overseas growth, or overseas markets can make up for the slowdown in North American business and hedge the negative impact of the downward trend of the US dollar on its own business.

In addition to the seven technology giants we are familiar with, all of which have overseas businesses accounting for nearly 50% to hedge the exchange rate losses caused by the weakening of the US dollar, some small vertical giants with better quality, such as Uber, which performed well in this earnings season, have overseas businesses, and the growth of overseas businesses is not bad. In addition to hedging the weakening of North American business, it also pulled up the growth of the overall business, and the current investment does not seem to be as large as that of the US stock giants.

While Chinese companies have not benefited significantly from the expected interest rate cut in the US stock market, Dolphin Jun believes that after the US dollar interest rate cut is launched, the RMB interest rate cut space can be slightly opened, which will help alleviate the current actual borrowing costs are too high, and thus help alleviate the expectation of a weak economy.

In addition, in the second half of the year, the government's fiscal spending is expected to catch up. If the economy does not decline further and the negative impact of the short-term earnings season is gradually released, Chinese companies may have the opportunity to recover in the short term with the help of the external US dollar interest rate cut.

In this case, Dolphin Jun began to add back the positions that he had cut before.

4. Portfolio Adjustment and Returns

With the end of the US stock earnings season, the US stock valuation cut triggered by the yen rate hike has come to an end. Dolphin Jun has slightly increased the positions that were cut in July based on the certainty and upward elasticity of individual stock performance.

Based on the performance of the earnings season, the certainty of the performance path, and the valuation cost performance, Dolphin Jun still gave priority to high-end wafer manufacturing and flash memory tracks in AI. As for Nvidia, Dolphin Jun is afraid of heights and the volatility is too large, so Dolphin Jun still focuses on observation.

Among the technology giants, only Apple was carefully selected. Dolphin Jun believes that it has a relatively strong certainty in hardware repair, and the investment cost in the AI ​​era is relatively low, but the certainty of the results is relatively high.

The first batch of selected stocks and the reasons for adding positions are as follows:

At the end of last week, the portfolio return rose by 2.6%, outperforming the Chinese asset index - MSCI China (+1.7%), Hang Seng Technology Index (+0.6%) and CSI 300 (+0.4%), but underperforming the S&P 500 (+3.9%).

From the beginning of the portfolio test to last weekend, the absolute return of the portfolio was 42%, and the excess return compared with MSCI China was 64%. From the perspective of net asset value, Dolphin Jun's initial virtual assets were 100 million US dollars, and now it has risen to 144 million US dollars.

V. Contribution of profit and loss of individual stocks

The risk of Japanese yen interest rate hikes has subsided, and the thunder of the earnings season has ended. After the recent CPI and social retail data, the US economy still seems to be on the road to a soft landing. US stock trading has returned to normal, and the expectation of a soft landing rate cut + a weaker US dollar is still beneficial to companies with a high proportion of overseas business. Trading has returned to technology giants again.

Last week, the stocks that Dolphin Jun focused on had large fluctuations in the stock pool. Dolphin Jun explained as follows:

VI. Portfolio asset distribution

Alpha Dolphin virtual portfolio holds a total of 13 stocks and equity ETFs, of which 5 are standard and 8 equity assets are low-allocation. The rest are distributed in gold, US bonds and US dollar cash. As of last weekend, Alpha Dolphin's asset allocation and equity asset holding weights are as follows:

How can A-shares pass 3,000 points?

This week (August 12 to August 16), the A-share market continued to consolidate at a low level. The Shanghai Composite Index rose 0.74% in the five trading days of the week and closed at 2,879 points on August 16; the Wind All A Index rose 0.32% this week.

In terms of sectors, 12 of the 31 primary industries of Shenwan rose this week, among which communications, banking, media, computers, electronics and other sectors rose the most, while agriculture, forestry, animal husbandry and fishery, building materials, basic chemicals, food and beverages, beauty care, light industry and other sectors rose the least this week.

This week's style is relatively chaotic, and the small and medium-sized micro-cap and large-cap indexes have all risen well, and from the perspective of industry distribution, it is also a bonus at a glance. Reflecting on the performance of style indexes and broad-based indexes, Wind Micro-cap Stocks, CSI 2000, SSE 50, FTSE China A50, Dividend Index and other indexes rose the most. The rise and fall of the Beijing Stock Exchange 50, Science and Technology Innovation 100, Value Creation, CSI 500, CSI 380 and other indexes fell the least.

In terms of Hong Kong stocks, the Hang Seng Index rose 1.86% this week; the Hang Seng Technology Index rose 0.85% this week. In terms of sectors, 10 of the 12 Hang Seng industry indices rose this week; among them, telecommunications, energy, industry, finance and other sectors were the top gainers this week, while real estate construction and comprehensive industries fell this week.

Most of the major overseas asset classes closed higher, and the three major US stock indexes saw a surge of 3-5% this week; gold rose, crude oil prices fell slightly, and non-ferrous metals rose; the US dollar index fell.

Figure: Performance of major global asset classes this week; Source: Everbright Securities, 36Kr

01 One of the principles of buying A-shares is to find highlights from macro data

This week, there are many events, including July economic and financial data in China, and July inflation data in the United States overseas, and it happens to be the earnings season. The large amount of incremental information has a significant disturbance to the market, which has also led to a seemingly split but conservative style tendency in A-shares this week.

On the surface, the economic data in July continued the overall weak and structurally differentiated trend in June.

In terms of retail sales, essential consumption is relatively stable due to its rigid attributes; optional consumption has slowed down relatively significantly, and the weakening of clothing, shoes and hats, gold and silver jewelry is more striking; durable goods and real estate chains have a certain marginal improvement compared with June. The structural differentiation of retail sales can be seen in two characteristics: first, the marginal improvement of durable goods and real estate chains is mainly driven by the old-for-new, which is one of the key directions of the current policy of restoring and expanding consumption; second, the consumption in first-tier cities is more obviously weakened, which may be related to the change in income expectations caused by industry adjustments, and the indirect impact of real estate drag on consumption.

In terms of supply data, the year-on-year growth rate of industrial added value above designated size continued to slow down in July; from a structural point of view, the upstream mining industry was mainly affected by the booming prices of raw materials and the acceleration of production, while the manufacturing industry slowed down slightly overall, but there was still differentiation in structure. Industries with marginal improvement include high-tech industries, electronic computers and communications, etc.; industries with marginal decline mainly include utilities represented by electricity and gas, automobiles, equipment manufacturing, electrical machinery, chemicals, and pharmaceuticals. Among them, the overall prosperity of industries such as automobiles, electronic computers and communications, and chemicals is relatively high, and the marginal improvement or deterioration is not large.

Although fixed asset investment has weakened slightly overall, the absolute level of infrastructure and manufacturing as a whole is still high, and infrastructure investment is still accelerating. Manufacturing investment has only slightly declined on the margin, and the decline in real estate investment has slightly expanded. The above results are consistent with my country's industrial transformation in recent years. It can be said that they are all within expectations, and the monthly marginal changes have little impact on the overall trend.

The financial data in July still reflects the old problem of insufficient total demand. The decline in new credit is large. For the corporate sector, it still reflects the optimization of the credit structure under industrial transformation. For the resident sector, it is mainly dragged down by real estate, which is also related to the changes in income expectations caused by industry adjustments, which is cross-confirmed with consumption data.

From the economic and financial data for July released this week and the inflation data released last week, although the overall situation reflects the fact that total demand is weak, the structural highlights still provide direction for investment. Since the full liberalization, the direction of economic recovery has been determined, but the economy is like water, and it takes time to warm up. Under the circumstances of internal economic transformation and external interference, the kinetic energy switching determines the rhythm and pattern of the overall weak recovery.

Therefore, in the context of investors' downward risk appetite, mature and stable large-cap value stocks are naturally favored by investors based on the structural characteristics of the current economic fundamentals to reflect A-shares. Among them, most of my country's dividend stocks are value-oriented, and the investment value of stable dividend income is magnified in the interest rate cut cycle and asset shortage. In addition, there are certain liquidity problems in A-shares this year, which also drives funds to the dividend sector with value attributes.

For this week's economic and financial data, it is just a reconfirmation of the current economic form. Therefore, reflected in A-shares, the industry sectors with strong dividend attributes such as big finance, energy, and utilities have the highest increase, which also shows that the market has continued the main line style of the past two years. Judging from the supply and investment data, the overall prosperity of electronics, computers, communications, chemicals, and automobiles is good. In the consumption data, durable goods and real estate chains (mainly home appliances) have also shown marginal improvements. These industries are also relatively strong in A-shares, which essentially reflects the investment logic of buying strong fundamentals and buying policy-supported A-shares this year.

Looking ahead to the future market, macro data usually have a certain inertia, and the changes are mainly marginal. It should be noted that the current easing is to help the future rebound, so the overall weak recovery rhythm will continue. Then, in the context of shrinking A-share trading, the current logic of buying strong fundamentals and buying policy-supported will not change. In terms of the reflected market, the main line status of dividends will not be shaken due to the weakening of valuation advantages, but the frequency and amplitude of adjustments will increase accordingly, which is normal under the low-risk preference of investors; the relatively strong industries corresponding to the macro data are naturally the main direction of investment; another result of the shrinking trading volume is to suppress speculation in small-cap stocks. The short-term rebound of small and medium-cap stocks under the valuation advantage after full adjustment does not represent a change in the main line logic of A-shares at present.

02 How does overseas interest rate cut trading affect A-shares?

Looking at overseas, although the inflation data in the United States in July continued the downward trend, it was slightly lower than expected, which caused some interference on whether the interest rate cut could be carried out as scheduled in September this year, and some differences among investors. In terms of the focus of the interest rate cut decision, future employment data is crucial.

At present, the overseas market is a mixture of interest rate cut transactions and recession transactions, so we can see that major asset classes sometimes adjust and sometimes surge, and the style is open and close, and the impact on A-shares is still on the realization of interest rate cuts. Based on this, the Fed's interest rate cut coincides with our current round of interest rate cut cycle, and it also helps to alleviate the pressure of exchange rate depreciation.

From the perspective of the monetary policy trends of China and the United States in recent years, the specific historical period has caused the temporary failure of the adjustment mechanism corresponding to the interest rate gap between China and the United States. Influenced by the international trade environment, the monetary policies of the two countries tend to fight each other, and the correlation is obviously weakened. The overlap of this round of interest rate cut cycles is more due to historical background. The Fed's interest rate cut has an impact on my country's interest rate level, but the impact on A-shares is relatively indirect. The pricing of A-shares still depends on its own fundamentals.

Reflecting on the trading level, as the Fed's interest rate cut is realized, it will be conducive to the release of overseas technology companies, which will give rise to the demand for related small and medium-sized start-ups in my country, which is determined by the global division of labor in science and technology. Therefore, the A-share style at that time may be that dividends are still the core theme, and at the same time, some funds will flow to small and medium-sized start-ups with greater flexibility. This is because the fundamentals of those small and medium-sized start-ups have been strengthened with the overseas interest rate cuts, and at the same time, they have valuation advantages. Investors will also be driven by the money-making effect under the asset shortage.

It is worth noting that in previous strategy articles, it was repeatedly mentioned that the current market has expected that small and medium-sized start-ups will have a wave of high prosperity under the linkage of the first and second levels in 25 years, so there may be a rush to run at the end of the third quarter or the fourth quarter of 24. Since the beginning of this year, small and medium-sized start-ups have been weak for a long time, but the band market under short-term speculation has continued. In the future, overseas interest rate cuts and the above market expectations may be catalysts for short-term market conditions. In the current market environment where risk appetite is low, speculative trading is not impossible, but risk exposure must be controlled.

*Disclaimer:

The content of this article only represents the author's views.

The market is risky and investment should be cautious. In any case, the information or opinions expressed in this article do not constitute investment advice to anyone. Before deciding to invest, if necessary, investors must consult professionals and make prudent decisions. We do not intend to provide underwriting services or any services that require specific qualifications or licenses to the parties to the transaction.

The low-key Visa: The secret of the "world's largest payment giant"

As the founder of the world's largest payment empire, how did Hawke use a chaotic and orderly organization to smash the traditional authority he hated and build a real invisible business empire?

In the vast sea of ​​commerce, Visa, with its secretive and huge body, has quietly become one of the largest business organizations in the world.

Its transaction volume is ten times that of the retail giant Walmart, and its market value is more than twice that of General Electric.

Visa's success lies not only in its scale, but also in its almost invisible construction of a real invisible business empire.

The creator of all this is Dee Hawke, who is still a "bullied little sheep" at the age of 36 and is also named one of the "eight people who can change people's lifestyles in the past 25 years" by the influential American Fortune magazine.

01 Founder of Visa

Hawke was born in the early days of the Great Depression in 1929 and grew up in a poor family in Utah. He did a variety of hard physical labor jobs, and "Root, hog, or die" was the cruel truth of that era.

Dee Hawke is an endless thirst for knowledge. His childhood was full of curiosity and exploration of the world. His rich survival experience made him full of love and awe for nature.

Although he obtained a college degree, Hawke was self-taught most of the time. At the age of 14, he forged his age and got into a canning factory to work as a sewage dumper. He also worked as a dairy farm clerk, a porter, a slaughterhouse worker, and a farm pesticide sprayer...

Relatives around him said that he was too rebellious and it was difficult for him to succeed.

This self-taught financial innovator started his career at a consumer finance company in Los Angeles in 1951. Here, he lived a life driven by others and was even assigned to search for lost deposit certificates in the garbage dump.

Doing a meaningless job that was fooled by others, Hawke was disgusted with the "industrial age thinking" of centralized power and bureaucratic hierarchy here, where people were treated as gears and consumed.

We are in the midst of a "global epidemic of institutional failure," and Hawke hates this kind of bureaucracy with "mediocrities" involved.

Hawke, full of confidence, began to try his new system. Sure enough, his team's performance improved, but he was fired. Bureaucracy is not like nature, where the "fittest survive." When certain people are in power, they naturally create prejudices.

The experience of being fired made Hawke hate authority even more, but the situation soon got worse.

Before being fired, Hawke began to use credit cards frequently, and his uncontrolled consumption put him in debt that he could not extricate himself from. He and his wife experienced unprecedented financial pressure, and they had no savings to repay their debts. Hawke had to work multiple jobs.

He cut up his credit cards and vowed never to do it again.

The importance of a stable credit system with low financial risk began to sprout in his mind.

In 1965, Hawke moved to Seattle. As a father of three children, he lived in poverty and joined the National Commercial Bank as a handyman. This job made him depressed and he even wanted to escape all the time. Fortunately, he stayed...

In 1966, he was selected by his boss to solve a big problem in the credit card system. Hawke had the opportunity to change his fate. He began to transform from a bruised little sheep to a giant of the super credit empire.

"I have never met such a powerful person." "We are all limited in our way of thinking, but he is not." This is the evaluation of Hawke by his colleagues.

"One of the eight people who changed people's lifestyles in the past 25 years" is the evaluation of him by an influential American magazine.

How did Hawke complete the transformation?

02 Symphony of the Payment Revolution

In 1958, Bank of America opened the prelude to the credit card era.

In early 1966, Bank of America announced the authorization plan for credit cards, and the credit card business began to grow explosively.

But there were no magnetic strips on credit cards at that time, and merchants did not have electronic card readers. All settlement processes needed to rely on primitive actions such as paper checks, telephones, and punching. The massive number of users made the credit card center almost unable to bear the burden, and banks, merchants, and users were all miserable.

What's more troublesome is that when merchants have discounts and exchange rates change, a small amount of credit consumption will bring huge troubles such as amount issuance, settlement, and reconciliation. Unfair competition between banks, information delays, and loopholes in some rules will also bring systemic risks, and fraud will begin to spread...

In 1968, the loss of credit cards exceeded tens of millions of dollars.

Hawke became one of the members to solve this problem. Hawke, who had suffered from the system for more than ten years, had long hated the administrative management system. As he had imagined, these problems can only be solved by breaking the authority of banks between users and merchants and breaking the "control-centered" order.

Hawke proposed to establish a non-stock company jointly owned by member banks, uniting these competing banks to jointly weave an unprecedented payment network "disruptive dream".

It was not easy to convince 200 licensed banks and thousands of staff to join this organization, not to mention that his proposal was fiercely opposed by the audience. But as a debater, Hawke is unyielding.

"Credit card is just banking jargon. It happens to appear in the form of a card, but it has nothing to do with money. The monopoly of traditional banks and the state's control over money will be broken." Hawke's cognition is very visionary, "What we are really engaged in is the business of exchanging monetary value."

He demonstrated how the current system is collapsing and put forward his new idea - to create a natural system of "chaotic management order" (Chaordic Organization) that is as open as nature.

Hawke's organization is very transparent. Multiple internal institutions similar to the board of directors have unchangeable authority and autonomy in their respective functional areas, but no organization is above other power institutions. Like a carefully woven symphony orchestra, it ensures that every voice is heard, but no single voice can dominate the entire movement.

In the new system, organizations are decentralized and autonomous, making them more flexible and adaptable. In this network, as long as they meet the regulations, everyone can join in and further expand the network.

All members, partners and end customers provide services and are served under common rules, and eventually become a unified global brand and payment system. It is like an ocean or an ecosystem, with strong self-regulation capabilities.

Today, we who have Internet thinking may not find this design novel, but more than 50 years ago, when there was no Internet, Hawke was definitely an explorer with advanced ideas.

In 1970, after Hawke's electronic authorization system ran for 90 days, everyone became willing to join. Hawke, who was elected as president, set off a major project that affected the world.

03 Technology leadership

Hawk saved credit cards and turned them from a department with serious losses to a profitable department.

"The most important value exchange system in the world." This is Hawke's ambition. How did he do all this?

It is Visa's technical strength that supports all this.

Hawke firmly believes that technology can change the payment system. He has a clear vision of the potential of the electronic payment system, which has promoted Visa's transformation to an electronic value exchange system.

Hawke established a set of operating procedures, which is the rules of the game for the entire ecosystem. He set the card design standards and the rules for the use of logos in advertisements, as well as how to complete the work between various members and the penalties for members violating the rules. Basic provisions that must be implemented in contracts signed with merchants.

Then he began to build his own technical infrastructure. In 1973, National BankAmericard Inc. (NBI) launched the first electronic authorization system, which marked the birth of Visa's electronic authorization system. Subsequently, the system has undergone years of development and improvement and has become a key technology supporting Visa's global transactions.

Data has become Visa's blood, and the technical nervous system ensures the fast and accurate flow of information. They have established some rules for the system, just like the laws of nature to ensure its operation.

The electronic authorization system ensures real-time processing of electronic authorizations and ensures smooth payment. Visa builds a global network, uses advanced encryption technology and security measures to protect data security, analyzes transactions in real time to reduce fraud risks, integrates into POS terminals to simplify the payment process, and allows sellers to recognize buyers and buyers to recognize sellers, a global payment tool that transcends language, law, currency, customs or culture.

Visa not only expands its influence by sponsoring Olympic events, but also gains more users through TV commercials. It quickly surpassed American Express and MasterCard from its initial 20% market share to become the world's number one credit card brand.

In 1976, Bank Americard was officially renamed Visa, starting its international journey.

In 1983, Visa launched a global ATM network, allowing cardholders to withdraw cash around the clock around the world. Once the service was launched, Visa's international payment status was instantly consolidated.

In 1984, when the Visa system was in its heyday, Hawke put his business suit into the closet, rushed to the remote wasteland, drove a crawler tractor and started a 10-year farmer's life, returning to his true nature.

In 1994, he returned to the arena and began a series of academic discussions on his mixed-order organization. On July 16, 2022, Hawke, 93, died at his home in Washington State at the age of 93.

04 Post-Hawk Era

Following Hawke's clear system thinking and technical ideas, Visa has taken every step firmly after Hawke left.

In 1986, Visa developed a clearing and settlement system that supports 21 currencies, and its technological advantages allowed it to break through again.

In 2008, with the popularity of smartphones, Visa launched a mobile payment platform, fully opening up its strategic layout in the field of mobile payments. In 2014, Visa Checkout once again simplified the online payment process and improved the consumer experience.

Although it has more than 3 billion credit cards, tens of thousands of banks, tens of millions of commercial institutions, and more than a billion users worldwide, it has always remained at the forefront of technology. Visa's advanced authorization system processes more than 65,000 transactions per second and prevents more than $25 billion in fraud each year, demonstrating its data-driven and technological strength.

All of this is inseparable from Hawke's chaotic and orderly ecosystem and the cutting-edge positioning of technology companies, which allows it to continue to grow and evolve like nature.

Of course, Visa is not sitting back and relaxing. The challenges of emerging payment methods such as Apple Pay, Google Pay, Alipay and WeChat Pay, and the development of blockchain technology and digital currencies may reshape the landscape of the payment industry.

With the development of blockchain technology, Visa has begun to explore cryptocurrency payment solutions to meet the growing market demand for digital currencies. In the future, Visa will still need to continue to innovate and adapt to changes to maintain its leading position in the global payment field.

Turing giants split again, Hinton-supported California AI restriction bill preliminarily passed, LeCun, Fei-Fei Li, and Andrew Ng called for a fight

Which models and companies will be restricted?

California's "AI Limitation Bill" was successfully passed in the face of strong opposition from a number of AI giants, technology giants, startups and venture capitalists.

As we all know, apart from the interpretation of various science fiction movies, AI has not killed anyone in the real world or launched a large-scale cyber attack.

However, some US lawmakers still hope that sufficient security can be implemented before this dystopian future becomes a reality.

Just this week, California's "Frontier Artificial Intelligence Model Safety Innovation Act" - SB 1047, once again took an important step towards becoming law.

In short, SB 1047 will prevent AI systems from causing large-scale casualties or causing cybersecurity incidents with losses of more than $500 million by holding developers accountable.

However, due to the strong opposition from academia and industry, California lawmakers made some compromises - adding several amendments suggested by AI startup Anthropic and other opponents.

Compared with the original proposal, the current version reduces the power of the California government to hold AI labs accountable.

Bill address: https://leginfo.legislature.ca.gov/faces/billTextClient.xhtml?bill_id=202320240SB1047

But even so, (almost) no one likes SB 1047.

Yann LeCun, Fei-Fei Li, Andrew Ng and other AI experts have repeatedly expressed their dissatisfaction with this bill that "stifles open source AI and forces AI innovation to be suspended or even stopped."

Various joint letters are also emerging in an endless stream:

Not only more than 40 researchers from the University of California, the University of Southern California, Stanford University and the California Institute of Technology strongly called for the bill not to be passed.

And 8 congressmen representing various districts in California also urged the governor to veto the bill.

Even LeCun copied the original stalk when he asked for a suspension of AI research before-please suspend AI legislation for six months!

So the question is, why did the previous article use "almost everyone"?

Because, in addition to LeCun, the other two Turing giants, Yoshua Bengio and Geoffrey Hinton, strongly support the passage of this bill.

Even feel that the current terms are a bit too loose.

As senior AI technology and policy researchers, we write to express our strong support for California Senate Bill 1047.

SB 1047 outlines the basic requirements for effective regulation of this technology. It does not implement a licensing system, does not require companies to obtain permission from government agencies before training or deploying models, relies on companies to assess risks on their own, and does not even strictly hold companies accountable in the event of a disaster.

This is a relatively loose legislation relative to the scale of the risks we face.

It would be a historic mistake to cancel the basic measures of the bill - a mistake that will become more obvious in a year, as the next generation of more powerful AI systems will be released.

Today, SB 1047, despite strong opposition from some US congressmen, well-known AI researchers, large technology companies and venture capitalists, has passed the California legislature relatively easily.

Next, SB 1047 will enter the California State Assembly for a final vote. Due to the addition of the latest amendment, the bill will need to be sent back to the California State Senate for a vote after it is passed.

If both votes are passed, SB 1047 will be sent to the governor, awaiting a final veto or signing into law.

01 Which models and companies will be restricted?

According to the requirements of SB 1047, developers or companies that develop models are responsible for preventing their AI models from being used to cause "significant harm."

For example, manufacturing weapons of mass destruction or launching cyber attacks with losses exceeding $500 million. By the way, CrowdStrike's "Global Windows Blue Screen Incident" caused losses of more than $5 billion.

However, the rules of SB 1047 only apply to AI models that are very large in scale - that is, the training cost is at least $100 million and the floating-point operations exceed 10^26 times. (Basically it is based on the training cost of GPT-4)

It is said that the amount of computing required for Meta's next generation Llama 4 will double by 10 times, so it will also be regulated by SB 1047.

For open source models and their fine-tuned versions, the original developers are responsible. Unless the cost is three times that of the original model.

In this case, it is no wonder that LeCun's reaction is so fierce.

In addition, developers must create testing procedures that can address the risks of AI models and must hire third-party auditors every year to evaluate their AI safety practices.

For AI products built on models, corresponding safety protocols need to be formulated to prevent abuse, including an "emergency stop" button that shuts down the entire AI model.

Wait...

02 What is the function of SB 1047 now?

Today, SB 1047 no longer allows the California Attorney General to sue AI companies for negligent safety measures before a catastrophic event occurs. (Anthropic suggestion)

Instead, the California Attorney General can seek an injunction requiring a company to stop an operation it deems dangerous, and can still sue AI developers if their models do lead to catastrophic events.

SB 1047 no longer creates the new government agency "Frontier Model Division (FMD)" that was originally included in the bill.

But it still establishes the core of FMD, the Frontier Model Committee, and places it within the existing Government Operations Agency, and expands its size from 5 to 9 people. The committee will still set computational thresholds for covered models, issue safety guidance, and issue regulations for auditors.

SB 1047 also has looser language on ensuring the safety of AI models.

Now, developers only need to provide "reasonable care" to ensure that AI models do not pose a significant risk of disaster, rather than the "reasonable assurance" required previously.

In addition, developers only need to submit a public "statement" outlining their safety measures, and no longer need to submit certification of safety test results under penalty of perjury.

There is also a separate protection for open source fine-tuning models. If someone spends less than $10 million on fine-tuning the model, they will not be considered a developer, and the responsibility will still be borne by the original large developer of the model.

03 Fei-Fei Li once wrote an article to criticize

The impact of SB 1047 on the AI ​​community can be seen from the column published by "AI Godmother" Fei-Fei Li in Fortune magazine:

"If it becomes law, SB 1047 will harm the nascent AI ecosystem in the United States, especially those parts that are already at a disadvantage: the public sector, academia, and small technology companies. SB 1047 will unnecessarily punish developers, stifle the open source community, and restrict academic research, while failing to solve real problems."

First, SB 1047 will over-punish developers and stifle innovation.

In the case of AI models being abused, SB 1047 puts the blame on the responsible party and the original developer of the model. It is impossible for every AI developer, especially programmers and entrepreneurs who are just starting out, to predict every possible use of their model. SB 1047 will force developers to take defensive measures—something that should be avoided at all costs.

Second, SB 1047 will constrain open source development.

SB 1047 requires that all models above a certain threshold include an “emergency stop button,” a mechanism that can shut down the program at any time. If developers worry that the programs they download and build will be deleted, they will be more hesitant to write code or collaborate. This emergency stop button will seriously affect the open source community—not only in AI, but also in countless sources of innovation in various fields including GPS, MRI, and the Internet itself.

Third, SB 1047 will weaken AI research in the public sector and academia.

Open source development is important in the private sector, but it is more important to academia, which cannot make progress without collaboration and access to model data. If researchers cannot access appropriate models and data, how can they train the next generation of AI leaders? The emergency stop button will further weaken academia, which is already at a disadvantage in data and computing. SB 1047 will deal a fatal blow to academic AI when we should be investing more in public sector AI.

Most worryingly, SB 1047 does not address the potential risks of AI progress, including bias and deepfakes. Instead, it sets a very arbitrary threshold - reaching a certain computing power or a model that costs $100 million to train. Rather than providing protection, this measure will restrict innovation in all fields, including academia.

Fei-Fei Li said that she is not against AI governance. Legislation is essential for the safe and effective development of AI. However, AI policy must support open source development, develop unified and reasonable rules, and build confidence for consumers.

Clearly, SB 1047 does not meet these standards.

The AI ​​unemployment wave is coming. Tens of thousands of people have been laid off in the gaming industry in 23 years. Blizzard employees are deeply saddened by the fact that their jobs have been taken away by AI.

You didn’tThe entire industry is being swallowed up by AI

Anxious game practitioners are about to suffer from PTSD about AI. Every time they see the CEO's email about AI, their hearts will tremble. In 2023 alone, 10,500 people in the game industry will be laid off.

Workers in the video game industry are suffering from the critical blow of AI!

When Noah saw the email from the company's CEO, a wave of anxiety came over him.

It was the spring of 23, and the Activision artist saw the CEO's message, emphasizing that AI has become the top consideration for the video game publisher.

The CEO said that the system is still being tested, but the performance of AI is promising.

Employees of the "Call of Duty" studio also received similar emails, approving them to use Midjourney and Stable Diffusion in the game to create concept art.

That spring, there were rumors that AI was about to replace certain jobs, and people were whispering and worried about their future.

Executives saw reason to be excited about AI, but many game artists, writers and designers saw a huge threat to their livelihoods.

Employees like Noah are worried and grief-stricken.

"I feel like we are abandoning humanity."

Then, jobs started to disappear.

In 2023, the gaming industry staged a battle royale of 10,000 people

Clearly, AI has put the gaming industry in trouble.

In 2023 alone, 10,500 people were laid off from the industry.

This year, layoffs in this nearly $200 billion industry will only get worse.

And the number of layoffs at the neighboring film company has reached 11,000, and it is still increasing.

Microsoft, the parent company of many studios such as Xbox and Activision Blizzard, closed Tango Gameworks and Alpha Dog Games in May this year.

Now, we see that generative AI built by OpenAI and others has penetrated almost all industries and destroyed many people's careers along the way.

Of all the industries, which one is the most affected by AI?

One answer is the gaming industry.

Although the industry's economic influence eclipses that of Hollywood, they have one obvious disadvantage: there are no unions.

And AI has already invaded.

A recent survey conducted by the organizers of the Game Developers Conference found that among more than 3,000 respondents, 49% said that their companies were already using AI, and 4/5 said they were concerned about it.

"AI is here, and it's here now!" said Violet, a game developer, technical artist, and veteran of AAA games for more than a decade.

"Everyone sees that it is being used. The question is: how to use it? To what extent should it be used? The genie is out of the bottle, and Pandora's box has been opened..."

According to emails obtained by Wired and interviews with artists, developers, designers and other practitioners in the game industry, the industry landscape was already precarious, and now the rise of AI will accelerate it further.

Historically, the automation of work rarely occurs evenly, and it is difficult to complete it "cleanly" immediately.

Most of the impact of AI is felt through "deskilling": more roles are handed over to machines or programs, and employees are fired, dismissed, and will not be rehired.

Based on past experience, this time the aggressive AI will not be an exception.

Although the experience of video game companies may not necessarily cut off the entire part because of AI, many companies have already used it to take shortcuts, improve productivity, and make up for the lack of manpower after layoffs.

This process is not as simple as a one-size-fits-all approach, but is based on various opaque execution decisions, which is very complicated and the ending is often ambiguous.

Rather than saying it is Skynet, it is better to say that this is a group effect.

01 Game practitioners strongly oppose AI

Molly Warner is a game environment artist who is responsible for the game "Overwatch" at Blizzard.

When she saw this email from the CTO, her anxiety reached its peak.

"Almost everyone I know is strongly opposed to using AI-generated images."

In May 2023, Bobby Kotick, then CEO of Activision Blizzard, once again raised how generative AI will affect the gaming industry at the company's collective meeting.

In his speech, he cue Sam Altman and OpenAI.

I have known Altman for a long time.

I don't know how many of you realize that many modern AIs, including ChatGPT, started with the idea of ​​beating games, whether it is Warcraft, Dota, StarCraft, Go or chess.

Last year, I saw AI, and I felt the same way when I saw the first Macintosh: AI will have a huge impact on our society, both positive and negative.

In July 23, Vance announced that Activision Blizzard had obtained access to GPT-3.5.

Employees were allowed to use AI tools to create concept art, write marketing materials, and even use AI to write user surveys.

Many game workers and artists are uneasy about this proliferation of AI.

Everyone began to worry about their livelihoods, but few people spoke up. The reason for keeping silent was the fear of losing their jobs.

Activision repeatedly assured artists that AI would only be used for internal concepts, not final game assets; AI would not be used to replace employees.

However, what was coming still came.

By the end of the year, Activision launched an AI-generated skin in Call of Duty: Modern Warfare 3.

In late January 24, 1,900 Activision Blizzard and Xbox employees were laid off.

The team most affected was the 2D artists.

Lucas Annunziata, a former environmental artist at Blizzard, posted on X, "What a fucking day."

At Activision, the situation was not much better.

Many 2D artists were laid off, and the remaining concept artists were forced to use AI to assist their work.

Now, company employees have been required to participate in AI training, and AI is being promoted throughout the company.

02 The entire industry is being swallowed up by AI

A company employee said, "When an art director can get something fast and good by giving AI some wrong guidance, why spend a lot of money to hire a group of expensive concept artists and designers?"

At present, everyone has basically reached a consensus: concept artists, graphic designers, asset artists and illustrators are the groups most affected by AI so far.

After working for 6 years and producing more than 6,000 graphic design works, the employee learned that he was replaced by AI.

And the 3D artist said that since the release of Midjouney V5, he is no longer an artist.

Here, creativity is no longer important, and these Internet contents copied from artists have saved the boss a lot of time and money.

He was sad and angry.

After all, in the eyes of many customers, 2D images generated by AI are good enough. They are more concerned about cost rather than quality.

Fortunately, tasks such as 3D animation and programming are much more difficult to automate at present.

A recent report from consulting firm CVL Economics shows that the gaming industry is adopting AI more than its peers in television, film or music.

According to a survey of 300 CEOs, executives and managers, nearly 90% of video game companies have begun using generative AI.

The CVL report found that games "rely on GenAI more than other entertainment industries to perform tasks such as generating storyboards, character design, rendering and animation. In fact, it is expected that GenAI may be involved in more than half of the game development process in the next 5 to 10 years."

03 Infringement suspicion

For the overall situation of large game companies like Activision Blizzard, many people don't know.

In fact, the company consists of a winding supply chain consisting of studios, developers, third-party developers and testers.

The situation of 3A game companies is quite fragmented, so we can't see who is doing what, and we can never see which link uses AI.

And this ambiguity also eliminates most concerns about infringement.

US law insists that any work that requires copyright must have a human author.

But is it an infringement to use unlicensed intellectual property to train AI? This is still an open question.

Today, copyright uncertainty, concerns about the security of LLM systems, and employees' concerns about job losses have exacerbated the division of the entire industry.

Now, it can basically be divided into two camps.

It is reported that Blizzard, unlike its brother studio Activision, does not allow developers to use publicly available AI generators.

04 Lay off your own employees and then look for outsourcing

And the pace of layoffs in the entire industry has not stopped.

Last May, EA CEO Andrew Willson said in a quarterly earnings call, "In every agricultural and industrial revolution, labor will be replaced in the short term, and then in the long run, opportunities will increase. We hope that AI can bring the same opportunities."

In February 2024, EA laid off 5%.

Employees of Riot Games, the company behind League of Legends, revealed that the leadership had said that they did not intend to replace anyone with AI because they knew the value of the artists the company had and the extent to which Riot's art reflected its brand integrity.

However, in January this year, Riot Games laid off 530 employees.

Cross was also one of the people laid off. He doesn't think his job will be completely replaced by AI, but it's interesting to consider the number of people who were laid off.

During the pandemic, many studios hired and expanded frantically, and then began to lay off employees.

Many companies have followed the practices of large technology companies and relied more on outsourcing and contract workers.

Interestingly, a few hours after being laid off, Cross was approached by a company that outsources art for game studios.

The company asked if he could make skins for a version of League of Legends, which was Cross's old job at Riot, and then he could only pay by piece after outsourcing.

In Cross's view, the biggest problem brought by AI is that art is undervalued in games.

Like other jobs, this is a race to the bottom, and companies will use any means to lower wages, usually outsourcing.

Noah, an employee of Activision Blizzard, also said that the company is now outsourcing a large number of 3D art assets because the internal art department can't keep up with the work after layoffs.

The same is true in other countries. come this far to stop

OpenAI’s “Game of Thrones” is not over yet. Four new leaders have joined the team, and half of them are Chinese.

OpenAI has seen a series of core senior personnel changes. In troubled times, heroes emerge, and a group of rising stars take the lead.

Since Altman was dismissed and reinstated in November 2023, OpenAI seems to have been in constant power struggle.

Recently, many executives have left one after another. According to the Financial Times, of the original 11 co-founders of OpenAI, only two are still working at OpenAI.

In addition to Altman, only Polish computer scientist Wojciech Zaremba remains at OpenAI as a researcher.

Greg Brockman, president and co-founder of OpenAI, who has been fighting side by side with Altman, will extend his vacation until the end of this year. It is generally believed that this is a precursor to his upcoming departure.

Another co-founder, John Schulman, left permanently and joined competitor Anthropic. Peter Deng, vice president of consumer products, also left OpenAI.

According to statistics, nearly 75 core employees of OpenAI have left and founded about 30 AI startups.

OpenAI has become the "Huangpu Military Academy" of the AI ​​industry, and its former employees have set up their own companies with the experience and connections they accumulated in the company.

Although this talent output is beneficial to the development of the entire industry, it is undoubtedly a huge loss for OpenAI itself.

01 Newcomers take over

Although there is a loss of talent, OpenAI is also continuing to spend a lot of money to recruit and promote talents, and others have stepped up to take over the vacancies.

This scene makes people sigh: "Game of Thrones" happens in reality, "Chaos is the ladder to rise."

Based on conversations with former and current employees and other insiders, the following is the situation of OpenAI's new batch of leaders.

02 Jakub Pachocki-"Genius Scientist" who succeeded Ilya

Jakub Pachocki is currently the chief scientist of OpenAI and an old employee who has worked at OpenAI since 2017.

Judging from Pachocki's past experience and his rocket-like promotion speed within OpenAI, he is worthy of the title of "Genius Scientist".

Before joining OpenAI, Pachocki obtained his Ph.D. from CMU in just three years and conducted a seven-month postdoctoral research at Harvard University.

In the first five and a half years after joining, Pachocki led the Dota, reasoning and deep learning science teams, all of which belong to OpenAI's transformative research programs.

He played a key role in OpenAI's development of the Dota 2 game robot, which eventually reached a professional level through continuous self-play.

Starting in 2021, Pachocki was promoted to research leader, research director and other positions. He is the leader in the development of GPT-4 and OpenAI Five, and is praised for his research contributions in large-scale reinforcement learning and deep learning optimization.

Altman praised him generously, describing Pachocki as "undoubtedly one of the greatest thinkers of our generation" and "responsible for managing many of our most important projects."

Altman once commented on Pachocki's contribution to the GPT-4 project: "The outstanding leadership and technical foresight shown by Jakub Pachocki are remarkable. Without his contribution, we would not have achieved what we have today."

In a previous MIT interview, Pachocki compared the development of a language model like GPT-4 to "building a spaceship" - every component must be perfect.

He also mentioned that the basic construction method of the GPT model has not changed much since the first version was released in 2018.

After Ilya Sutskever left in May this year, Pachocki officially took over the position of chief scientist.

According to current and former employees, Pachocki's influence was growing before his promotion.

Altman values ​​Pachocki very much and has previously assigned him tasks similar to Ilya, although Pachocki should report to Ilya, which has led to tensions between the two.

During the "palace fight" last November, Ilya and board member Helen Toner jointly launched the "expel Altman" action, but Pachocki did not follow Ilya and stood firmly in the camp supporting Altman.

However, when Ilya officially announced his resignation, Pachocki's tweet still showed respect and admiration for the predecessor.

Ilya brought me into the world of deep learning research and has been my mentor and great collaborator for many years. His incredible vision for deep learning is the foundation of OpenAI and the field of artificial intelligence today. I am very grateful to him for countless conversations with us, from high-level discussions about the future of artificial intelligence progress to in-depth technical whiteboard meetings. Ilya-I will miss working with you.

Employees said that after Ilya left, Pachocki was supported by Altman and has become an important decision maker within the company.

Pachocki was born in Poland and graduated from the Department of Computer Science at the University of Warsaw. Not only did he get his bachelor's degree in just three years and had two internships at Facebook, but he also participated in programming competitions many times during high school and college.

Jakub Pachocki (third from left) at the ICPC 2012 competition

According to statistics from the Programming Hall of Fame, Pachocki represented the University of Warsaw in the ACM-ICPC World Finals twice in 2011 and 2012, and won the gold medal in the second year.

In terms of individual competitions, Pachocki's performance was equally amazing: he won a silver medal in IOI in high school, and won medals in programming competitions hosted by Topcoder, Facebook, and Google many times.

Among them, the best result was winning the gold medal in the 2012 Google Code Jam competition, with a prize of $10,000.

He once mentioned in an interview that programming competitions such as Google Code Jam are more like math homework or solving logic problems, and winning requires extremely high cognitive abilities.

02 Barret Zoph

Before co-founder John Schulman left, he and his only direct subordinate Barret Zoph jointly led the post-training team.

According to an employee, Schulman was responsible for setting the high-level agenda, while Zoph was responsible for the daily management and research of the team, ensuring that the project was completed on time and that the basic model could be smoothly deployed to products such as ChatGPT and the API for developers.

Zoph is well respected within OpenAI, and now that his boss has left, he has naturally become the sole leader of the team.

Zoph graduated from the University of Southern California. During college, he joined the natural language group of the school's Information Sciences Institute (USC Information Sciences Institute) and worked on statistical machine translation with two professors, Kevin Knight and Daniel Marcu.

After graduating from college, Zoph joined Google Brain and worked as a research scientist for 6 years and 8 months, dedicated to training large language models and applying them to various applications.

During his time at Google Brain, Zoph has made fruitful academic achievements. Google Scholar shows that his papers have been cited 62,284 times in total.

In August 2022, Zoph left Google Brain and joined OpenAI to build ChatGPT. His main research direction is training large sparse language models and AuoML, such as neural structure search (NAS).

At OpenAI's May press conference, Zoph demonstrated the real-time visual function of GPT-4o, allowing GPT-4o to solve mathematical problems in real time through the mobile phone camera.

Mark Chen

Mark Chen has been the director of frontier research since he joined OpenAI in 2018, and is responsible for a working group under the leadership of Bob McGrew, vice president of research.

Chen also showed extraordinary talent very early, and won full marks in mathematics competitions such as AMC10, AMC12, and AIME, which paved the way for him to enter MIT.

When he graduated from MIT, Chen received a double bachelor's degree in mathematics and computer science. During college, he interned at Microsoft and Trading, and was a visiting scholar at Harvard University.

After graduating from college, Chen entered the financial field to engage in quantitative research, building machine learning algorithms for stock and futures trading.

In his last company, Integral Technology, Chen has reached the partner level.

Currently, he also serves as a coach for the US IOI training team.

Sure enough, people who are good at math are good at everything.

Rest in peace, Jim Simons.

You have proved with your own example that being really good at math has a positive impact on almost everything else.

Chen focuses on multimodal modeling and reasoning research at OpenAI. He led the team that created DALL·E and the team that incorporated visual perception into GPT-4.

Chen also led the development of Codex and contributed to the advancement of the GPT model, including the development of image GPT, and was a member of the 17-person gold medal team of GPT-4o.

At the launch conference in May this year, Chen also took the stage to demonstrate the voice function of GPT-4o.

Chen's influence gradually emerged during the "palace fighting storm" in November last year. According to a former employee, Chen, Barret Zoph, and post-training researcher Liam Fedus were the main liaisons between the leadership and employees at the time.

For example, the three conveyed a joint letter from employees supporting Altman, which became a key link in pushing Altman to reinstate. Most employees in the letter said that if Altman did not reinstate, they would join Microsoft.

03 Lilian Weng

Lilian Weng is currently the head of OpenAI's security system, mainly engaged in research on machine learning and deep learning.

Weng graduated from Peking University with a bachelor's degree in information systems and computer science. She went to the University of Hong Kong for a short exchange and then received a doctorate from Indiana University Bloomington.

During her doctoral studies, Weng's research areas were complex systems and networks, focusing on social media, social games, human-computer interaction, and complex information network modeling.

Open her Google Scholar profile, and you can also see Weng's papers on memes and social networks.

She has interned in user analysis at companies such as eBay and Mozilla, and then successfully "changed careers" and joined Facebook and Dropbox to work in software engineering and data science.

Since 2018, Weng has joined OpenAI as a research scientist. As the technical director of the robotics team, she focuses on algorithms for training robotic tasks.

Later, Weng also led the research team of applied artificial intelligence and is currently the leader of the security team.

In July this year, OpenAI transferred the former head of the security team, Aleksander Madry, to a team focused on the basic work of reasoning. According to people familiar with the matter, the security team originally led by Madry was transferred to Lilian Weng.

Now, Weng will manage teams focused on both long-term and short-term AI safety. This organizational decision worries some researchers because the incentives for long-term and short-term safety may conflict with each other.

It is worth mentioning that the blog posts written by Weng on his personal website are very popular. They are basically long articles of 10,000 words, with both technical dry goods and opinion output, which are references for many people in the industry.

They made millions guessing what Trump would say next

The idea is very interesting, but it is indeed dancing on a tightrope.

"Do you dare to make a bet?"

A simple verbal bet between friends has now been made into a large social platform with over a million people participating.

In recent months, a prediction market platform called Polymarket has rapidly emerged and attracted a lot of attention.

Polymarket allows users to bet on real events with cryptocurrency. Some popular bets include the results of the US election, what words Trump and Musk will mention in their conversation, how many tweets they will post, when GPT-5 will occur, etc.

The platform believes that it has gathered "collective wisdom" and its predictions for the future are more accurate than traditional expert models and opinion polls.

Why is Polymarket so popular? It has even been listed as a data source for public opinion predictions by various media? What kind of website is it?

01 Trump and "Tampons"

"Bet on what you believe in." On Polymarket, users can bet on a wide variety of events, including political elections, sports events, economic indicators, pop culture events, business, scientific events, etc., with a focus on both serious and absurd ones.

Not long ago, Polymarket users bet about $5 million on Trump's conversation with Elon Musk on August 12, predicting which words Trump would say.

In this conversation, users who bet at the last minute that Trump would say "Tampon" received the highest return.

The most money was bet on predicting whether Trump would say "cryptocurrency", followed by "MAGA" and "illegal immigrants". Trump finally said three words, namely "MAGA", "illegal immigrants" and "tampons".

When Trump said the slogan "MAGA", the odds were 59%, meaning that bettors would get less than double the return. As for "tampons," the odds were 7% when he mentioned the word, meaning bettors received a whopping 14 times the return.

"Congratulations to my tampon brothers," a Polymarket user commented when Trump said the word.

A user who didn't make the prediction commented, "Tampons are the first bet I've lost in 51 bets, and I'm sad."

Users bet on what Trump will say in a conversation with Musk | Image source: Polymarket

In contrast, users who bet on "cryptocurrency" suffered a heavy blow because this market had the most money, but it didn't happen in the end. The odds for Trump mentioning "cryptocurrency" were initially 65%, but fell to zero at the end of his speech, causing those who bet on this item to suffer heavy losses.

The user known as "bama124" has joined Polymarket since July 2024 and has traded in 77 different markets in total, with a trading volume of more than $10 million. The user lost $67,000 on bets that Trump would avoid using the word "tampon", but ended up with a profit of $964,402 by correctly betting that Trump would not say the words "cryptocurrency", "bitcoin", "Tesla", etc.

Polymarket's betting market activity has surged with Trump's return to social media platform X (formerly Twitter). These markets allow participants to bet on various outcomes of Trump's online behavior and public statements, involving huge amounts of money.

One of the more active betting pools is predicting how many tweets Trump will send in the next week. Participants can bet on different ranges, from 11-15 tweets to more than 50 tweets.

Another active betting pool focuses on whether Trump will tweet again before the election. Any original post, reply or quote tweet from Trump will be counted as a "yes", and retweets are not counted.

In addition to this, markets on Polymarket also focus on other events, such as predicting the country with the most Olympic medals, or predicting movie box office performance, predicting Premier League results, predicting the number of interest rate cuts by the Federal Reserve, etc.

Users bet on the first weekend box office of "Alien: Takeaway" | Image source: Polymarket

There are also "Will August 2024 be the hottest on record?" "Will the United States confirm the existence of aliens in 2024?" "When will GPT-5 be released?" "When will Sony release the Playstation 5 Pro?" "Will Taylor Swift get pregnant in 2024?" "Will Taylor Swift support Kamala Harris?" "Bieber Baby: Boy or Girl?" and so on.

User bets on "Bieber Baby: Boy or Girl?" | Image source: Polymarket

User bets on "Will the United States confirm the existence of aliens in 2024?" | Image source: Polymarket

There are currently more than $100,000 in bets on whether the US government will confirm the existence of aliens by the end of the year, but Polymarket shows only a 3% chance, which is about the same as the probability of predicting that Trump will go to jail before the election.

But overall, the platform is most popular for betting on the results of the US election. So far, Polymarket users have bet more than $500 million on the election results.

US election results prediction | Image source: Polymarket

Prediction markets have even begun to become an information tool, believed to reflect collective emotions and event probabilities.

It is said that the US political circle is now gradually becoming popular to go beyond polls and focus on betting markets to understand voters' ideas. Predictions on Polymarket are also frequently cited as a kind of data by media such as The Wall Street Journal, The New York Times, The Washington Post, Fortune, and Bloomberg.

Jason Furman, an economist at Harvard University and former chairman of the Council of Economic Advisers in the Obama administration, also said that the White House "often refers to prediction markets about election results and specific events."

In the past few months, Polymarket has reportedly successfully predicted that US President Joe Biden would withdraw from the re-election campaign and that Donald Trump would choose JD Vance as his deputy.

Polymarket executives directly stated that people have become increasingly distrustful of traditional polls and the media's use of polls as a source of predictions. The platform has found "a crazy product-market fit because we are currently experiencing what may be the most unpredictable and turbulent election in American history."

Many people lie to pollsters, but tell the truth when real money is used for betting.

The Polymarket team said, "It is very easy for people to only believe the media that confirms their biases and expected results, and algorithms will ultimately only provide people with information they agree with. Polymarket tells people the true probability, regardless of what anyone wants to happen."

However, like traditional polls, Polymarket is not infallible.

In June 2024, users of the platform predicted that Trump would mention Bitcoin in the debate with Biden with a probability of 61%, but this did not happen in the end. Similarly, they failed to successfully predict the results of the 2024 French election, and of course, some traditional polls did not predict it.

02 The birth of "the world's largest prediction platform" Polymarket is not only a novel existence in the American election scene. It is also a hot startup, with $70 million raised to date, including a $45 million round in May.

Investors include Ethereum founder and "Zero to One" author Peter Thiel, who is best known for buying 10.2% of Facebook in 2004 for $500,000, becoming Facebook's first outside investor.

Polymarket receives investment from Peter Thiel and others | Image source: X

The founder of Polymarket, who is almost unknown outside the cryptocurrency circle, is a 26-year-old New York native named Shayne Coplan, who studied computer science at New York University, is obsessed with cryptocurrency and prediction markets, and has appeared in Fortune magazine.

Shayne Coplan began learning programming as a teenager, and rumor has it that he participated in the initial sale of Ethereum in 2014 at a price of 30 cents, making him the youngest participant. ETH is now trading at $2,600.

Shayne Coplan said on X that Polymarket was the first place to predict that J.D. Vance would be former President Donald Trump's vice presidential candidate. He also wrote that Polymarket "priced" Biden's probability of withdrawing from the presidential race earlier and with higher probability than the media predicted.

Polymarket founder Shayne Coplan｜Image source: Polymarket

In July this year, according to Dune Analytics data, Polymarket set an all-time high of more than $387 million in trading volume in a single month, involving more than 40,000 active traders. In contrast, the site had only 1,200 monthly active traders in the same period last year, with a betting amount of $6 million.

In fact, there are many other online betting platforms around the world, and there are also a variety of content on the door, including the probability of aliens coming. But Shayne Coplan said that Polymarket is different from them, mainly because Polymarket's odds are constantly determined by a group of bettors, rather than by centralized odds makers, and users can buy and sell shares of future event results.

Unlike traditional platforms, Polymarket runs on the blockchain, uses smart contracts to settle and pay bets, is one of the fastest-growing on-chain applications, and uses automated market makers (AMMs) to provide liquidity to the market. One of the platform's radical ideas is to break the shackles of traditional prediction tools through decentralized market mechanisms.

They believe that when predictors have money to invest, they will work very hard to find the right answer.

"Prediction markets are more accurate than expert models or surveys in predicting the future." Polymarket once claimed in a blog post, giving two basic reasons.

The platform said that first, when people have money to invest, they will integrate all the best analysis they can find, consulting polls, models, expert comments and other factors that may not be reflected in popular indicators.

The second reason is that prediction markets allow people to express not only their opinions, but also their confidence in these opinions. This is a difference that opinion polls or simple pros and cons systems cannot match, allowing a small number of very confident people to guide the market to reflect their views through their bets even when most people believe in different views.

Polymarket posted an example: For example, in a certain election, suppose there are a hundred people who believe that a certain candidate will win, but their certainty is only 80%. Then suppose there is one person with superior knowledge who is 100% sure that this candidate will lose, perhaps because he has a proprietary analysis of early mail ballots and clearly sees that others have misread the data.

In a rational free market, this person will make a very large bet, and the market will shift to reflect his point of view. In other words, even if only one person has better information, he can overcome the "misinformation" that guides the majority by betting large enough.

Polymarket claims: "In an unrestricted rational free market, the best answer wins: the market odds will shift to reflect the views of those with the best information."

Because it operates on a blockchain network, information about its trading volume is public, and the platform says this makes it a better source of real-time public sentiment data than polls or traditional media.

They also feel that this is very different from social media, where the messages that get amplified are the most popular messages, not the most accurate messages.

What are the issues discussed on social media? “One problem is that, because of the way algorithms and product features on platforms like Facebook and Twitter work, opinions and predictions on social media tend to be optimized for maximum engagement. Well-researched, thoughtful predictions shared on these platforms are often replaced by other information that lacks analysis or evidence but is deliberately extreme or provocative, or that simply repeats what the audience already wants to believe. In addition, many social media platforms are rife with bots, fake accounts, state-sponsored manipulation, and paid promotional content. Understanding the world through social media is bound to produce a fairly distorted picture of reality.”

Instead, Polymarket claims that prediction markets are only used to facilitate transactions between individuals who believe they have enough information and analysis to predict future events. Furthermore, prediction markets reward those who make successful predictions, free from the distortions of a large amount of online activity, and expand global access to unbiased information.

The platform claims that their platform is “good for society” because businesses, policymakers, and organizations can use the predictive insights generated by prediction markets to accurately understand the world and plan accordingly. “It doesn’t take a lot of people to participate in a particular market to generate valuable information that has the potential to help people around the world better plan their future.”

“Polymarket is a decentralized global prediction market platform that enables predictors to earn money by speculating on world events, giving everyone a clearer view of the future. Prediction markets are one of the most effective tools to improve society’s ability to predict the future and make decisions. Polymarket’s mission is to drive adoption of prediction markets to reward accuracy and repair the information ecosystem.” The platform claims.