最近读的一本关于研究技术和股市泡沫的书，十分呼应分析当下的AI热潮，这本书还是很有帮助的。不得不说作者的总结和分析还是十分有帮助的，值得反复阅读。
此刻正坐在西撒哈拉地区的Dakhla Sur Mer四楼餐厅等着开会，不得不说这样的经历还是十分玄幻的。脚下是一片生机勃勃的土壤，暂新的路面外到处是建设的工地，连海塘都似乎在新修。这片上个月刚从联合国解决完主权归属问题，百公里外依然是蔓延的雷区，我们试图要找的mine和地雷mines竟然是一个词。这个区域的招商政策也十分友好，几乎没有什么税收，是个机会巨大的希望之地，当然这也是叙事。
Bubbles are important, undeniable facts of life for citizens living under entrepreneurial capitalism. However, bubbles are both inefficient (from a strictly economic perspective) and potentially damaging to the individual interests of those who are caught up in them. Our inability to avoid bubbles suggests that our understanding of them is incomplete.
泡沫本身是重要的，反复的泡沫也说明我们不懂泡沫。
It was not merely unknown if and how such metrics would translate into bottom-line profits—it was unknowable. The eToys story epitomizes the interaction of unknowability and consequent narratives that are used to divine the unforeseeable future.
预估新公司的利润转化是个很难的事情.
Counterintuitively, knowing who might profit from an innovation might reduce the likelihood of a bubble. Because all bets are tied up in one firm, the bet is more closely aligned with the success of the technology, as opposed to different segment or monetization strategies associated with the new technology. There is less room for competing narratives to appeal to different populations and thereby drive up the entire sector.
所有新技术的利润如果明确集中到一个公司，尽早形成垄断格局，实际会降低泡沫的形成。这点上其实是创造了一个更伟大的值得投资的平台公司。这个现象要好好关注。目前的AI模型上一定程度与此类似，只是还没集中到单一公司的状态。
The presence of pure-play firms indicates that uncertainty may be exploitable by new entrants. Second, pure plays make good stories. Finally, for a bubble to form, there must be a way for investors to literally buy into the story. If there are no tradable financial assets that closely track the fortunes of the technology, then there can be no market speculation. Without a pure-play investment opportunity in a given technology, it is simply not possible for a speculative bubble to form for that technology. Simply put: a market must already exist for there to be a market bubble.
Pure-Play纯新概念公司的价值是显著的，至少在泡沫形成上有很大的帮助。可以纯讲故事，也是泡沫形成的必要条件，这个角度看纯GPU公司的高估就十分易于理解了。
Some ideas and technologies are better subjects of narratives. It was easier to tell a story about human flight than the world’s first synthetic plastic, Bakelite. The degree to which technologies lend themselves to storytelling is an important factor in driving bubbles.For other technologies, such as polyester or the laser, neither of which generated a boom and bust cycle, we find no associated coordinating event and no set of plausible entrepreneurial narratives.
天生有些领域更适合讲故事，特别是能面向消费者的，比如车、比如手机、互联网等等。塑料这类材料则不行，反而是更扎实的进步。所以在适合讲故事的领域，会讲故事就很重要。
The fourth and final causal factor that contributes to the likelihood of speculation is the presence of novice or unsophisticated investors. Retail-facing innovations may be more likely to grab the attention of a broad set of investors, even when that retail-facing quality is not perfectly correlated with profitability.
新韭菜的推波助澜对于泡沫的形成也十分关键。
we measure their frothiness. Formally, frothiness is the number of standard deviations from the predicted stock or index trend. We classify an episode as a boom and bust when frothiness is greater than 2, that is, 2 standard deviations above the trend, where the “trend” is the predicted stock price looking forward and backward seven years.The dotted line is the bubble line—or the line that at every point in time is exactly 2 standard deviations above the trend.
泡沫通常形成在2个标准差的位置。
Recognizing that uncertainty is a fundamental feature of the entrepreneurial dynamic, we explore how different types of uncertainty accompany the emergence of new industries. Each type of uncertainty—technological, competitive, business model and value chain, and demand—is introduced through a series of technology case studies.
All things being equal, uncertainty can lead to higher asset valuations because of the option effect. The logic is straightforward: all limited liability investments (including investments in corporations) have some of the central features of options; the upside is unlimited, but one can lose only the amount invested.
不确定性是企业的基础特征，特别是对于新行业和新企业。技术、竞争、商业模式、产业链和需求的不确定性都很大。同时不确定性+有限责任公司的特征，也构建出来了期权价值，不确定性越大，价值越大。
an effective technological narrative generates expectations—“statements about the future” that are “performative”; statements about the future can alter behavior of market participants; “they do something.”⁷ One important performative effect is that expectations can be self-fulfilling.
Assuming that the investing public possesses heterogeneous tastes for different narratives, the more technological bids we can construct about each technology, the more people are likely to be interested in potentially investing in a venture intended to create each particular “imagined future.” More potential narratives means more bids, and different bids will appeal to varied investor tastes. By sweeping up a larger, more varied set of investors, a technology that can support more narratives is likely to generate speculation.
故事的角度越多，越好讲，泡沫也越大。
Stratifying the nature of uncertainty—seeing it as a family of types—helps us understand when narratives are more likely to take off; we can more clearly see the various dimensions around which narratives might be constructed. That is, unpacking uncertainty allows us to estimate both the likelihood of the option effect mentioned earlier, as well as the likelihood of fantastic narratives. Both can lead to boom and bust episodes, but only the latter can lead to a bubble.
将叙事中的不确定性分层分析很重要。既能看到期权性价值的机会，也能看到潜在的风险。
In other words, whereas expert investors appreciated correctly the importance of airplanes and air travel, the narrative of inevitability largely drowned out their caution. Technological uncertainty was framed as opportunity, not risk. The market overestimated how quickly the industry would achieve technological viability and profitability.
“必然性”的叙事往往会掩盖很多问题，比如盈利等，会把不确定性当成机会而不是风险，这个需要警惕。
The combination of new rubber trees coming on line and better use of raw materials led rubber prices to plummet 75% between 1910 and 1913, even in the face of increased demand for automobiles. This caused a bust. To this day, the inflation-adjusted price of natural rubber in 1910 remains the historical apex. Free entry into the production of rubber ensured that supranormal profits would never return to the rubber market. What happened then is an important lesson in the role of free entry and market structure in producing broad-based market speculation.
橡胶价格崩盘的故事很有意思，明显低估了供给及其对价格的冲击。
A successful business must position itself in an advantageous part of the market ecosystem, even as that ecosystem is forming. Since the eventual form of this ecosystem is often unknowable, the business and its investors must be both good and lucky. Betting on growth implied a bet as well that demand would be high if a product could be offered that provided sufficient quality at the right price. As we now know, this was a good bet. In contrast, it was a much poorer bet for airplanes in the late 1920s.
好企业必须在一个生态系统中的有利位置。站对位置十分关键，稳定币是个好位置，自动驾驶上提供方案就不是个好位置，搞车可能是，也可能是芯片等等。
We are experiencing a very similar situation today when we consider autonomous vehicles. It is unclear which companies will make money, where the bottleneck will be, and which company, if any, will control the bottleneck. Should we place bets with ride-sharing companies such as Uber or Lyft, legacy automobile manufacturers such as Ford and GM, diversified tech firms like Google and Apple, upstarts such as Tesla, or technology specialists such as Intel, which purchased the leading sensor provider Mobileye?
自动驾驶上的这个问题很有意思，投资哪里最好？这就是不确定性。
To summarize, we have seen that uncertainty is more than a binary description of the state of the world. Rather, uncertainty can be decomposed into components—technological, competitive, business model and value chain, regulatory, and demand uncertainty. Uncertainty is understood only when applied to specific contexts. It is only then that we might consider what is uncertain.
不确定性并非非黑即白，而是可以分成多个方面：技术的、竞争的、商业模式的、产业链的、监管的、以及需求的等几个维度，
First, we are more likely to buy into narratives that draw analogies to things with which we are familiar. That is because in we tend to falsely believe that the past always predicts the future. But often, new opportunities are difficult to understand, and sloppy analogies lead to bad decisions. Experience is about learning to restrain and override these very biases; it is about not constructing specious narratives.
类比是讲故事的重要方式。比如稳定币是数字美元这句话就很直接，机器人替代人也很简单。经验的重要性就是能看透这个故事、看穿这个偏见，而不是编造。
Simile and metaphor are critical to the construction of narratives. The use of analogies is typical in trying to understand an uncertain world. This idea was what Stanford professor Chip Heath and his brother, Dan Heath, of Duke University, would call “sticky.”Even in the face of countervailing evidence, sticky ideas have some or all of the following six attributes: simple, unexpected, concrete, credible, emotional, and (finally) part of a story. Simple means that the idea has a brisk message, even if the underlying reality is complex. The idea of offshore oil appears simple, even if the geology and economics of offshore oil extraction are not. Unexpected is something that makes an idea shiny, interesting. That Israel might have oil was unexpected in the sense that part of the mythos of Zionism is self-reliance and hard work.
Just as sticky ideas about new, emerging technologies can envelop novice investors, so, too, can incomplete information. Nobel laureate Daniel Kahneman points out that stories are easier to concoct when the facts are sparser.¹⁵ The less we know, the more we supply from our own imagination, which is influenced by our biases.
故事越有煽动性越能吸引新韭菜，信息越少，大家就越能编，就会炒的越high。煽动性的故事有明确的特点：简单、意外、具体、可信、感性以及很有故事性。
Importantly, there was not a general expectation of selling shares as a means for investors to earn a profit; rather, the expectation was that shares would yield dividends in the same way that bonds earned interest. For example, investors in early automobile companies expected dividends, not to be able to sell their shares. There was no common way to think about “growth” stocks—indeed, the idea of growth stocks did not emerge until the 1960s. But the new industrials still attracted investors because they were filled with wonder and generated cash.
60年代之前居然没有成长股的概念。最初的股票投资人都是为了赚利润和粉红，而不是从股票买卖中赚钱。即便如此，当时也还有历史上的多次泡沫，南海泡沫等等。
“From 1926 through 1930, New York banks loaned $25 million on margin ($342 million in 2016 dollars), and out-of-state banks loaned $16 million ($218 million in 2016 dollars), but other lenders loaned more than $48 million, equivalent to $641 million in 2016, so more than both other groups combined. Not only were novices investing, we can also conclude that stock market speculation in the late 1920s was also backed by a new and likely inexperienced class of margin lenders. The democratization of investment had worked on more than one level. Money markets allowed industrial firms to park idle capital in the hands of market speculators, even as the firms themselves were ill equipped to understand the risks involved in the practice.
资金和杠杆是炒作起来的重要支持，过去关注大小非等资金面的因素还是很有道理的。
Start-up stories are more interesting than profits or products in particular divisions of existing firms, and—all else equal—publicly traded start-ups are the most interesting of all. Start-ups are more likely to form when entrepreneurs can tell a story about how a new technology makes existing firms’ capabilities obsolete.
Speculation requires pure plays, and pure plays are defined by the way new technologies complement existing firm capabilities. Existing firms tend to have diversified across a range of markets, and new technologies that complement those already in use are likely to become part of the portfolios of diversified firms. Indeed, some technologies on our list, such as nylon, emerged as part of the research and development efforts of an already-diversified firm. We should not be surprised, therefore, that investors were unable to map their investments directly onto the fortunes of many new technologies.
投机需要投资到纯概念股上。不对多个概念股抱有幻想，纯度实际有巨大的炒作价值、期权价值。这也是新上市的纯概念股容易炒到天上的原因，实际是个不错的投资机会，值得适当冒险参与。cricle等已经证明了这点，以后要抓这类机会。
Should we expect a crash in Tesla shares? Our model would predict that this is a likely event: a consumer-facing technology, high uncertainty, and a skilled entrepreneur behind the curtain pulling the levers. We suspect that investors are ignoring a strong response from established automobile manufacturers entering the EV market and the way prices will fall in the face of this competition. We predict that Tesla shares will come down to earth. Call us when they do.
特斯拉的泡沫风险也很高，目前本质上就是叙事支持的。电动车的故事自我实现了，自动驾驶和机器人等还需拭目以待。
A successful narrative can be self-fulfilling. On the other hand, the power of words is not limitless, and often, compelling narratives fail to make money and end in failure.
好的故事是可以自我实现的。但只有故事也不行。
If the answer is affirmative across all six categories, we should be very concerned that we are in a bubble. If one comes to the conclusion that the answer is no across more of these categories, the fear of the god of froth should decline. When only one or two categories are affirmative, it would appear unlikely. （1）Story: Is the story particularly compelling or sticky? （2）Use: Is the story about something that you are familiar with in terms of use or imagination but in an area that you fail to understand the business well? （3）Naïfs: Are there other naive investors in the market? Who is investing in this technology? （4）Pure play: Is there a stock that is believed to track the fortunes of the technology directly? （5）Competition: Does the narrative ignore future competition? （6）Business model: Are there a variety of stories about how money will be made commercializing the new technology? （7）Narrative accelerator: Did something or somebody turbocharge the narrative? （8）Leverage: Are investments significantly leveraged? Do intermediaries play a large role?
泡沫的形成离不开这8个因素，是可以逐项分析的，2-3项符合问题不大，但6项以上符合要小心了。故事是否有煽动性、用途是否易于理解和联想、是否有新韭菜入场、是否有纯概念股上市、是否忽视了未来的竞争、是否可以幻想出多种变现方式、是否有人加速叙事，以及是否有高杠杆的情况。从AI角度看，故事很有煽动性、用途也容易联想、变现方式也可以幻想出多种、低估了未来的竞争、英伟达/OpenAI正在加速叙事，这五点基本具备了；新韭菜目前还看不清、纯概念股目前还没有，估计要等openAI上市了；高杠杆是目前coreweave、甲骨文最近正在做的事情，这么看的话，AI的泡沫无疑正在形成中了。昨天还在分析Cricle的情况，看起来circle的故事、用途、韭菜、纯概念股、变现模式、低估竞争上无疑这六点都具备了，这个可以解释股价从上市后300跌至60，跌掉80%的原因。但未来呢？还需要继续去理解。
Make no mistake, extensive competition leads to large-scale business model experimentation and is often very good for consumers, but it will depress security prices, and investors and policy makers need to be aware. The inability to foresee entry is common and pervasive.
激烈竞争对消费者是好事，但对于股票投资人可能是灾难，最近外卖大战就是如此，受损的是美团股东。
Such statements had real and serious economic consequences as they fueled chicanery associated with bubbles. Policy makers need to exercise great caution when commenting on markets that are filled with novice investors. Their words will be given undue weight and can spark large market movements.
监管者的发言可能会进一步刺激泡沫的形成。
That we get fooled is not surprising; narratives and stories are how we think. But with a better idea of how, when, and for whom these stories become costly, we can better avoid them.
叙事和故事是我们的思考方式，所以被骗就很正常了。所需要做的是熟悉这个套路来避免损失。