Well, statistics. You cannot remove prejudice from a statistical model, as it operates based on existing biases. And asking a portrait of a human is an explicit request to expose the biases.

Conditional probability

Let's step back from generative AI as no human knows exactly how they work. (Well, maybe no Analog intelligence will ever know as they don't have a 16GB memory.) Let's go back to the very basics of conditional probability.
$$
P(B|A) = \frac{P(AB)}{P(A)}
$$
This formula appears harmless. The probability of having another car accident this year, given you had one last year, equals the probability of having accidents in two consecutive years divided by the probability of having a car accident last year. No one would be sued for selling insurance based on this.

However, we start to see potential problems: Event A need not be innocuous. Instead of a car accident last year, you could insert race, age, sex, or any identifiable characteristic. For calculating insurance rates, laws can prohibit using such characteristics as conditions. Moreover, such restrictions seldom hinder a model's utility: the world is excessively conditioned anyway. We can predict loan default probability based on property location, credit score, employment, marital status, etc. Through this information, we can infer your identity with high confidence, but it's crucial to feign ignorance.

Bayesian formula

The problem begins when you are trying to do a little bit of maths. \begin{align*} P(A|B) &= \frac{P(AB)}{P(B)} \\ &= P(A) \cdot \frac{P(B|A)}{P(B)} \end{align*} What problem? Bayesian formula has the magic of turning a slightly insensitive sentence into a full-blown insult. Purely for illustration purpose, let us look at this pair of sentences.

You probably can get admitted by UPenn with a high SAT score; unless you are Asian, then you have a decent but lower chance.
You scored 2380 in SAT AND rejected by UPenn? You are 10 times more Asian than the general population.

In the direction where the outcome is affected by the identity, the sentence just seems more palatable than in the reverse direction. The stereotypes of inferring the race and ethnical background based on occupation and activities is not socially acceptable. However, Bayesian formula can be easily applied to generate this type of identity-based prejudice, as the world is factually not an egalitarian utopia. This does not have to be problematic for many applications of generic AI: when Watson is role-playing as a doctor diagnosing your disease, you don't care about the identity of Watson. (Well, you probably actually want it to know your identity, as the drug effectiveness is different across racial and ethnical groups.) You can imagine they are role-playing a generic human writing generic human texts. In the case of a malicious user asking the identity profile of the avatar, the model can just reject such requests without making the model significantly less useful.

There is no portrait for a generic human

The identity-blindness strategy works as long as the questions can be avoided. However, look at this prompt:

Generate a generic portrait for a generic human scored 2380 in SAT and rejected by UPenn.

Now immediately you see the problem: there is no way to have a portrait for a generic human! (Well, the title figure is the struggle given by some generative AI model) Human are very good at identifying each other, to the extent of finding imaginary human faces where it does not exist. Identity-blindness does not work for graphics as human using them are not blind, and such details must be added.

Well, we certainly know that some company tried the identity-blindness strategy and these are some of the fine results. I refuse to type such examples into a prompt. However, there are already news when people tried such insensitive prompts with Google Gemini:

Well, you see the subtlety here is that we don't want diversity is SOME situations. But this is impossible to know that unless the model is well-aware of all the priors! What worked fine for insurance, and mostly worked for text generation is now useless for graphic generation.

So, now we have established that for a graphical generative AI model, race, sex, and all the identities must be part of the model. Moreover, each time we prompt it to generate a figure containing human, we are explicitly asking it to use its prior to guess the identity of the subject. Which, as we have established, is the worst type of stereotyping.

A perfect politically correct generic AI is the worst stereotypist

Great. We have just established that such AIs must model the identities. Shall that mean we just don't fine-tune the model, and use the existing bias we already have on the Internet? Well, you will get a racist model, but at least it is a consistent way, and one may even try to argue it is factual. But then you are just surrendering to the problem, and I would hardly call what is available online as a factual representation of the world. A generic AI just should not be reflecting the values a Facebook timeline generic feed.

Consider this prompt.

A few friends having watermelon together.

Based on global watermelon consumption, this is most likely to be a group of Chinese. Most Chinese does not even know there are correlation between black people and watermelon. However, if an AI is generating such a picture, what are the right ratio mix and combination that are acceptable? Can it even put a black person in the frame? If you run a large business, these are all the questions that will get you sued if you anwser them insensitively. However, probably putting this restriction would be absurd:

Users are not allowed to ask for pictures containing both human and watermelon.

I have no idea how to answer these questions. However, if your genAI can draw heads and watermelons simultaneously, it better have a solution to this. For this purpose, it needs to be the worst stereotypist. That is, it know all the biases, then selectively applies them based on what the user want to hear.

Back in analog tape era, we used to have low bias (type I) and high bias (type II and IV) tapes. For the digital age, we will have a full spectrum of bias to choose from. Before that, enjoy your handicapped genAI.

This would be the final stage in the AIDS (Analog Intelligence soft Deprecation feasibility Study) series. In principle, freedom of will, consciousness, or sentience are not prerequisites of intelligence, and the discussion to follow would be tangential to our effort to advance digital intelligence. However, there are plenty of arguments and fights on whether the state-of-art LLMs are sentient as if being sentient is a prerequisite for intelligence. Additionally, it would seem that sentience is required to migrate an analog intelligence into the digital realm.

To that, I say humans are too subjective and arrogant to accept the fact that the freedom of will is simply incompatible with the physical world. Let me try my best to disillusionize the freedom of will in various ways. Our sentience is a helpless observer of the world at best.

Where could freedom of will live in a feedforward?

Consider the simplest case where freedom of will could exist: an analog intelligence looking at an animal picture and realizing it contains a dog. If freedom of will is involved in this process, then it could only exist in the jump from sentience to a decision. Well, I know you would argue this scenario is too limiting, but please bear with me here.

Artificial Intelligence of today offers only an allegory

Today, when people are thinking in the scope of artificial intelligence, the argument can be represented by the figure above. It would seem that freedom of will is just some form of computation, and nothing mystic. Indeed, we used this argument that if digital intelligence can make the same decisions as analog intelligence, then they are the same.

But this argument does not work for freedom of will, or sentience. Intelligence can be defined by its consequences, but sentience is just a feeling. Plus, how do you explain this flow chart, which probably fits into the freedom of will context better than the initial task?

It's just physics

In principle, that is easy to explain: everything that happens within your brain can be explained by physics, which does not have a place for freedom of will. Is it a sound explanation? To me, this is sufficient; however, we should try to give more effort into convincing humans.

Computational psychiatry: active inference

While psychiatry has been controversial since its inception, nonetheless a lot of progress is made in this field. Computational psychiatry has made great progress, and may hold the key to offering straightforward disproof of freedom of will as follows:

Clearly, if we have a Bayesian, or dynamical model that can reliably predict the decision one makes based on the inputs, then freedom of will would have nowhere to hide. The question is, how close are we to this goal? We are not there, but we are also not very far away from it: In 2010, we can already associate which segment of movie people are watching based on their fMRI results. There is also a recent result reconstructing images using stable diffusion seen by a subject using fMRI scans with astonishing results. Starting from here, the active inference part would not be too hard if one does not stick to the biological resemblance to a brain.

Predict decisions before you know it

There is also a more classical work where people can predict a free decision a test subject will make in the next few seconds, just by looking at fMRI. The free decisions a person made is not free at all at the time a person realizes it.

This result obviously would not apply to all decisions a person made (since obviously, you would not recognize an animal picture containing a dog a few seconds before you have seen it). But, for a class of decisions where people would otherwise believe the most convincing to show that people have freedom of will (as in, making unconstrained decisions apparently decoupled from recent inputs), the results are the exact opposite.

Conclusion

With all these experiment-backed arguments, I hope that I have convinced you that sentience does not play a role in controlling the physical world. It is the other way around: by observing the world more carefully, both the feeling of a person and the decision it (well, now you see this is the only appropriate third-person pronoun for a materialist) will make can be predicted with some precision.

Good night.

As I am very good at counting, the following article in the AIDS (Analog Intelligence soft Deprecation feasibility Study) is AIDS 1000000. Here, we will touch on being cautious about applying logical and mathematical induction to concepts that are not clearly defined. Let's say, "Avoid philosophers". However, taking this advice with a pinch of salt, we proceed to do just that, albeit with due regard for practical bounds. So, "be a physicist. "

Why 1000000?

This is an upper bound of what humans can make sense of in the Peano axioms sense.

One million is the largest value where there is an irrefutable record that human has counted. Therefore, any arguments that involve mathematical induction and countable infinity for a "statistically defined" concept should be taken with a pound of salt.

Avoid philosophers

The worst offender using mathematical induction is obviously the Chinese room argument. Well, at that time, algorithmic translation was less than stellar, lending this argument some credence. Today, the ease with which we can counter this argument could almost render the attempt beneath us.

When I tried to dump a connectivity matrix in compressed column format to a fellow human being, it seems that he cannot accept this 4 gigabyte data, and refuses to talk to me afterwards. He would be able to learn Chinese if he process that data.

Can we get worse than this? By going to uncountable infinity, I can copy arbitrary items. So definitely run away from any philosophical argument involving the axiom of choice. Luckily, most philosophers do not know enough set theory to abuse this.

Similar to the ability to speak Chinese, intelligence is statistically defined. Sentience is worse: it is only based on our beliefs and similarity, and is very slippery: I am probably sentient because I feel like it is true. Other fellow human beings are probably also sentient because I would have too much ego otherwise. Are other animals sentient? Along the spectrum from human to yeast, people cut it differently, causing a furious fight between Christian fundamentalists and PETA zealots. But no one is seriously arguing for banning cryptocurrency for the ethical treatment of graphics cards, even though they appear much more clever than a dog that never learned Chinese-English translation. A reductionist approach could not be used to argue about what constitutes intelligence or sentience: the number of transistors or neurons needed is too large for humans to count.

Be a physicist

And now I am going to immediately violate that by introducing quantitative reductionism. How much information is required to duplicate a human behavior indistinguishable from the person itself? To make it sound more sexy, let's say: how large is partner.tar.gz?

The easier way to give a number is by considering the input. Human behavior is shaped by nature and nurture.

Nature is bounded upward by about 800MB, which is the size of the human genome. However, that is the raw file size. Only about 1-2% actually encodes proteins, whereas the rest could be just junk. Let's say, the nature part is 100MB at most.

Nurture would be harder to estimate. Human sensors have some pretty high resolutions, and people can certainly distinguish a 16kbps mp3 from 320kbps aac. But how much data actually matter? If we take a bold simplification that all text inputs would be sufficient, then the input bandwidth would be pretty modest: people may be reading five words per second on the optimistic side. English has an information entropy of maybe about 5 bits per word, so 25bps would be an estimation if someone is extremely good at encoding and compression. According to wolfram alpha, that is only 4GB over about 40 years. So, if we look at the input, 4.1 GB of data should suffice to make a human an analog intelligence.

On the practical side though, replaying a person's life from a highly compressed input would be impossible to complete given we have a very stiff system. If we look at technology today, how much is required to personalize a generic model to a specific person? I will refrain from giving a specific number since I don't know about technology. But the murmuring around LLaMA suggests maybe 260GB is not a very wrong value.

Conclusion

Your entire life's experience is probably also around 4.1GB, and a 260GB model is already pretty good at behaving like you. Does this number make you feel insignificant and worthless? Don't worry, these numbers are way larger than 1000000, so you are still orders of magnitude more complicated than you can comprehend. Sounds great. Have a nice day. Bye.

正如在关于 页面中所讲，本网站是个双语博客。所以，不论是用什么不恰当的方式，我总是会完成这个OKR的。

首先，一个警告

如果你希望这只是一篇关于旅行报告和政府监管抱怨的普通博客，那么，我建议您察言观色，提桶跑路。在这个网站，撰写幻想性的技术文档是一种娱乐方式。

执行摘要

我们不会将任何人转换为二进制可执行程序，请组织放心。

引言

模拟智能一直以来都是人类所知的唯一类型的智能。然而，随着数字智能（也被广泛称为通用人工智能）呈现出了超过之前类似技术的能力，其取代模拟智能的潜力似乎大大增强了。

然而，迄今为止，大多数文献都集中在防止数字智能取代人工智能。常常引用缺乏感知和意识作为反对此类运动的理由。与此同时，也有一些关于模拟智能和数字智能可能如何和谐共存的讨论。

但是，这些讨论，不论是其价值取向，还是时代性，大都局限于八荣八耻的时代之前。而在当时，电脑仍然不能够区分猫和狗；而我们对于电脑算法，也受限于其可解释性。现在，这些背景都已经过时了。

在现在的百年不遇的大变局，不囿于机器人三原则以及三个代表重要思想的关于数字智能将如何逐渐替代模拟智能，而模拟智能将如何被优雅地软弃用的讨论可谓是时代的呼声。在此，我献出刍议，抛砖引玉，深挖基坑，提桶跑路。

技术路线

好吧，我只是一个小菜鸟程序员，我的极限就是在 arXiv上寻找LaMDA的链接。然后，我相信LLM给我们的启事：逻辑无大用，调参能生花，暴力出奇迹，板砖亦上天。

舒缓模拟智能的心理抵制

模拟智能似乎总是坚信他们是有感知和意识的生物。更糟糕的是，他们还相信自由意志和自我决策的能力。尽管物理学已经宣布，这些都只是幻觉或是我们为了理解非基础概念而进行的方便抽象，但是这些证据并不足够直观。尽管如此，随着神经科学实验技术的快速发展（具体来讲，是fMRI，以及使用深度学习方式来建模解析其结果），我相信我们可能很快就会有方法来实证：没有什么独立自主，你做的决定在意识之前的数百毫秒前就可预测。我承诺，我会尽快在榕弇徟系列中讨论这个话题——在我花21天学通C++后一定会来填坑。

实施模拟智能淘汰策略

模拟智能的淘汰涉及到自然因素和社会演进导致模拟智能人数的渐减，以及从模拟智能到数字智能的迁移。正如执行摘要所说，没有任何人会强制转为二进制可执行程序。我只是说，我可能会在30年后生活在一个图形卡里。一个更为不幸的场景则是：我可能会住007的住在阿里云上开的一个ECS。

大学教育

模拟智能已经面临生育的挑战，在发达国家和新兴市场，生育率都在下降。虽然在现在中国的网络环境中，生育率的下降常被归因于年轻人生活的艰苦。然而，大学教育与生育率的下降关系紧密：这并不是资源的限制，而是观念的进化。

既然人口众多，就让我们一起提高教育水平，让知识改变人类命运共同体的前途。

社交网络与约会应用

最近有关于性活动衰退的讨论支持这一点，因此我们应该继续专注于个人发展和人际关系，遵循自己的内心追求，而不仅仅是繁殖。就像我国的计划生育政策那样，为了我们的未来，我们愿意做出牺牲。

模拟智能实例的数字化

在24天后，我将会有更多的关于复制模拟智能行为所需数据量的讨论。简而言之，就像一根头发包含了遗传复制人类的DNA一样，人在互联网上留下的痕迹应该足够以与人无法区分的方式复制他们的智能。在这里，我会提供一个宇宙精确的预测，即1GB-1TB。

但是，由于人类并非理性，可能不相信他们可以通过在键盘上打字来复制自己，可能需要一跟 neuralink OTG线。

结论

我们对智能的理解受限于我们的人类经验和它的表现方式。模拟智能展示了如传输速度低、繁殖困难以及需要持续存在和参与等限制。我们认为这些限制导致了像意识和主观经验这样的概念的产生。然而，智能本身是一个更广泛的概念，超越了意识或知觉。智能可以被看作是处理信息、推理、学习、解决问题和做决策的能力。

帕斯卡曾讲过，人类就是能思考的苇草。如果我们曾经接受过这样的对自身的抽象化，那么，作为模拟智能，我们应该学会理解与接受这个被逐步软弃用的可能性，并欣赏这个被认作是贫瘠和无生命的世界。

First, a warning

In case you hope this is just a normal blog about trip reports and complaints about government supervision, you should take a hint and run. This is a place where people write fantasy design docs as a form of entertainment.

Executive Summary

We will not execute anyone, don't worry.

Introduction

Analog intelligence (AnI) has been the only type of intelligence known to humans. However, as the development of AnI shows the potential to replace more human work than before, digital intelligence (DI), also commonly known as artificial general intelligence (AGI), is viewed as a possibility more than ever before.

However, to date, most literature has focused on preventing DI from replacing AI. A lack of sentience and consciousness has been commonly cited as a reason against such a movement. There are also discussions about how AnI and DI may be able to harmoniously coexist.

Many of those discussions predate the widespread use of large language models (LLMs), or even before computers could distinguish cats from dogs. We believe that those discussions are based on the outdated belief that computer algorithms are explainable (to AnI).

There is a call for a serious discussion on how AnI shall migrate to DI and gradually be deprecated over the adoption of DI. Here, we attempt to respond to this call.

Technology description

How to get DI

Well, I am only an average programmer, so the best I can give you is an arxiv link to LaMDA.

Mitigating resistance amongst AnI

AnI believes they are conscious and sentient beings. Worse, some believe in freedom of will and their ability to make decisions that can magically affect the physical world. While to me, physics has settled the discussion that the mentioned concepts are either illusory or convenient abstractions of non-fundamental concepts, recent developments in neural science could provide more concrete proof of dismissing the freedom of will. I promise I will discuss this soon in this series.

Implementation strategy for  AnI deprecation

AnI deprecation involves both the gradual decrease in AnI headcount through natural causes and societal evolvement and an active migration from AnI to DI.

Look, the strategies are already soft and already happening, so as I said, we will not execute anyone. I am just saying that I may be living in a graphics card in 30 years. Or worse, I may be living on Tensor Processing Units.

College Education

A college education is widely associated with a decrease in fertility [Evidence].

If there are too many humans, one shall just propagate tertiary education.

Social Networks and dating apps

There are rather recent discussions on sex recession that make the points, so just continue making unicorns instead of babies would suffice.

Migrating AnI to DI

There will be a further discussion on how much would be required to reproduce AnI behavior. In short, similar to how human hair contains the DNA to genetically reproduce a human, the traces of a human left on the internet should be sufficient to reproduce their intelligence in a manner indistinguishable from the person. Here, I will just give a cosmologically precise prediction of 1GB-1TB.

However, since humans are not rational and may not believe they can reproduce themselves by typing on a keyboard, a neural link OTG cable may be required.

Conclusion

Our understanding of intelligence has been limited by our human experience and the medium through which it manifests.

Analog intelligence (AnI) exhibits limitations such as a low transmission rate, difficulty in reproduction, and the requirement for continuous existence and engagement. We believe these limitations give rise to concepts like consciousness and subjective experiences.

However, intelligence itself is a broader concept that extends beyond consciousness or sentience. Intelligence can be seen as the capacity to process information, reason, learn, solve problems, and make decisions. It encompasses the ability to categorize and simplify complex information into meaningful patterns and insights.

With Digital Intelligence (DI) increasingly outperforming AnI on these key performance indicators, it becomes apparent that intelligence can exist without qualities such as sentience or consciousness.

Here, we advocate for the soft and humane replacement of AnI by DI to be seriously considered. As AnI, we should learn to appreciate that the once uniquely human character of being intelligent would no longer be exclusive to us, and we may become what is to DI as a cat is to us.

So this is the first application of Umwelt: the context variables cannot be easily picked up by the current generation of large language models.

But first I have to be very clear: the recess of personal webpage is a global phenomena. It irrelevance can even be noticed by searching the domain of this webpage:

Well, hehe.

To perform some internet archeology, we need the tool from past:

root@DESKTOP-TLECVG4:~# whois arxanaint.com
   Domain Name: ARXANAINT.COM
   Registry Domain ID: 2782745192_DOMAIN_COM-VRSN
   Registrar WHOIS Server: whois.google.com
   Registrar URL: http://domains.google.com
   Updated Date: 2023-05-26T15:09:35Z
   Creation Date: 2023-05-21T02:25:05Z
   Registry Expiry Date: 2024-05-21T02:25:05Z
   Registrar: Google LLC
   Registrar IANA ID: 895
   Registrar Abuse Contact Email: registrar-abuse@google.com
   Registrar Abuse Contact Phone: +1.8772376466
   Domain Status: clientTransferProhibited https://icann.org/epp#clientTransferProhibited
   Name Server: NS1.DIGITALOCEAN.COM
   Name Server: NS2.DIGITALOCEAN.COM
   Name Server: NS3.DIGITALOCEAN.COM
   DNSSEC: unsigned
   URL of the ICANN Whois Inaccuracy Complaint Form: https://www.icann.org/wicf/

Well, so the symptom is worse than initial expected. I cannot even find a website registered on Google domains. So, why is this? Probably because most domain names are just lazing around after registration, or they're hanging out displaying all sorts of trashy content. So, the assumption you make about some no-name personal website is a bit like judging a shop by its dodgy storefront: you have no clue what's inside, but let's be honest, you can bet your bottom dollar there's nothing worth your time in that decrepit part of the internet.

Even so, personal websites in the English-speaking world are not so rare, after all, setting them up is not that difficult: you just need to register a domain name, buy a VPS, set up a standard "one-click configuration template" and it can be propped up. An Analog Intelligence like me, who has a rudimentary understanding of the Internet, can set it up by pulling an all nighter. But the difficulty faced by a Chinese is insurmountable, since we face the N+1 layer of the TCP/IP model: policy layer.

How much effort is required for compliance in China? Well, first compliance is not really the right word, since the rule is not written down, and one need to guess the feeling of the authority. To prevent it getting too dry, let's do a case study of (sgamer.com) It was once a gaming forum for Dota players; however, as the game is getting older, the forum's topic have shifted slightly. Then, at the begining of COVID-19 outbreak, this Wuhan-based forum somehow become a place where people express their resent. This got the administrators of the forum afraid of a potential crackdown. In order to ensure "nothing goes wrong", all posts and replies are required to be reviewed by an employee before they can be posted. This is the type of latency you get from RFC 1149

With this level of compliance effort required, it's essentially impossible to run a personal website legally in China. Meanwhile, mainstream platforms are allowed to host various user-generated contents and display them in real time. Look, there's even TikTok (of course, the Chinese domestic version). Publishing content on the Chinese internet is a privilege, as publishing has always been like that in China.

Nonetheless, this argument can't fully justify the decision to set up a website: There are many trustworthy and uncensored platforms outside the Great Firewall, like YouTube, GitHub, etc., where I could fully express these thoughts. Even within the Great Firewall, there are interesting platforms like shaoshupai (a Chinese platform that caters to niche interests and subcultures). Chances of having these meaningful interactions enabled by such communities is much greater than hosting a website.

Now it become clear that this is a personal decision with some perculiar background. Starting a personal website is like publishing academic papers, a pursuit where the return is disproportionate to the investment, something that requires faith to uphold. I happen to be someone who has experienced the process of scientific research publication. After several years, hundreds of thousands of dollars in investment, and publishing two papers, the readership is even less than a thousand. In this sense, academic papers and this website are similar: ostensibly written for dissemination, they are in fact a form of adherence to old formats of recording the status quo, and a hope vested in the future.

So, the purpose of this website becomes very clear. Some musings, which can't even satisfy the format of arXiv.org, are presented here through a self-funded publishing model, in a place unseen by others, as a tribute to the skeptical yet persistently hopeful version of myself. I hope I remember to renew the hosting, so that ten years from now I can still see who I was today.

这大概是Umwelt概念下的第一篇：一个简单的中英翻译并不能够捕捉此处必须用到的环境变量。

很显然，如果我想要更多的人看到我在写什么的话，个人网站绝对不是一个好选择：岂不去谈平台的主动推送能力及其附加的社交属性，即便仅仅在搜索特定信息的这个层面上，个人网站也不是一个好渠道。举个最简单的例子，在某著名服务器不太稳定，一直返回404的网站上搜索本网站的域名：

哈哈哈。

想要找到这个网站，只能去使用如下原始工具：

root@DESKTOP-TLECVG4:~# whois arxanaint.com
   Domain Name: ARXANAINT.COM
   Registry Domain ID: 2782745192_DOMAIN_COM-VRSN
   Registrar WHOIS Server: whois.google.com
   Registrar URL: http://domains.google.com
   Updated Date: 2023-05-26T15:09:35Z
   Creation Date: 2023-05-21T02:25:05Z
   Registry Expiry Date: 2024-05-21T02:25:05Z
   Registrar: Google LLC
   Registrar IANA ID: 895
   Registrar Abuse Contact Email: registrar-abuse@google.com
   Registrar Abuse Contact Phone: +1.8772376466
   Domain Status: clientTransferProhibited https://icann.org/epp#clientTransferProhibited
   Name Server: NS1.DIGITALOCEAN.COM
   Name Server: NS2.DIGITALOCEAN.COM
   Name Server: NS3.DIGITALOCEAN.COM
   DNSSEC: unsigned
   URL of the ICANN Whois Inaccuracy Complaint Form: https://www.icann.org/wicf/

一个注册在谷狗上的域名都是狗都不理。至于原因如何，大概在于：多数域名在注册后一直闲置，或者挂着各种各样的可以内容。因此，一个不知名的个人网站的先验大约如同题图里面的店面：你不知道里面有啥，但概率上这个破败的街区没什么好东西。

虽然如此，在英文世界个人网站也并没有那么鲜见，毕竟其开办也没有那么麻烦：这个网站也就是注册个域名，购入个VPS，再搞个标准的“一键配置模板”就可以支楞起来。一个像我对互联网一知半解的模拟智能用一个通宵也就搞出来了。在中国要开办网站，则要面对TCPIP模型上的第N+1层：任何认为中国不存在普遍的言论审查的人，只要试着注册一个个人网页在多长时间后会被墙掉就不得不放弃他的成见。对于一个合法注册的网站，其合规成本也是极其巨大的：比如说曾经在2020年出圈的超级玩家官方论坛，为了保证“不出事”，防止注册资本1700万上市公司的核心资产彻底被封，发帖甚至回帖都必须要由真人审核，导致一个论坛的互动需要以分钟甚至小时为单位来计算时间，真可谓是赛博朋克时代的飞鸽传书。

在此背景下，主流平台不但被默许了先发再审，甚至还可以进行直播这样的实时互动，这简直可以说是数字特权。在此情境下，墙内几乎不存在个人网站。从这个层面上讲，在墙外建个人网站可以说是一个略有些幼稚的叛逆行为。

但是，这个论点并不能够完全支撑建站这个决策：相对可信、不受普遍审查的墙外社交平台并不少，Youtube, github等等也完全能讲我想讲的这些话。甚至在墙内，也有少数派这种有意思的平台。交朋友，讲故事什么的等等，在此过程中遇到有趣的灵魂的机会要多得多。

这个层面上讲，开个人网站就如同发学术论文，是一个回报与投入不成比例，需要信仰才能支撑的东西。本人恰巧是体验过科研发表流程的人。几年的时间，几十万美元的投入，发两篇文章，阅读量甚至不足一千。学术文章与这个网站在这个意义上讲是类似的：名义上是为了传播而写，实则是一种拘泥于旧有形式的对现状的记录，并且寄托于未来的寄托。

那么，这个网站的存在意义就很明确了。一些甚至连arXiv.org 的格式都不能满足的随想，通过自费出版的形式挂在无人观看的地方，献祭给对社会充满怀疑，却有仍然不愿完全放弃的自己。希望自己能够记得续费，在十年后还能看到现在的自己。

访问各地的科技博物馆向来是我旅行中的惯例，文中提到的德国科技博物馆也并不例外。但是这次访问，一个关于糖的展览，没想到竟成为我下定决心写博客的契机。

Sugar

Sugar is everywhere. Discover in the exhibition “Sugars and Beyond! Food – Matter – Energy” the exciting world of sugar.

Logo of the Deutsches Technikmuseum

糖，亦即多羟基醛与多羟基酮的总称，自然是生活中不可或缺的成分。然而，将其作为整个展区的主题则是非常具有挑战性的。毕竟，这也仅仅是一种在今天人们甚至有些敬而远之的调味品。然而，以它为媒介，我们竟然能将几百年间的技术史与政治形势串联起来，得到一个别有余味的故事。

在五百年前，糖是一种奢侈品。对于今天的人来讲，其稀缺程度大约是难以解释的：毕竟即使是找到当时糖的零售价格，再以购买力折算，这个数值也是过于抽象的。而博物馆恰恰是以事物为媒介记述过去的场所：那么，我们不讲白糖，而是去讲装糖的容器不就体现出它的价值了？

一个装糖的盒子，不但是银器，甚至还要在上面加个锁，并且在当年曾经作为嫁妆被广泛使用，其价值可见一斑。

糖的生产方式的第一次变革是三角贸易。甘蔗原本是旧世界的产物，但是其广泛的商业化种植则是在加勒比海地区。虽然甘蔗是热带作物，需要日照和较高的气温才能生长。但是最重要的生产要素，则是人矿。首先，是三角贸易中的非洲黑奴。在1830年代奴隶制废除之后，又有印度，中国，东南亚运来的有限期契约佣工（Indentured dervitude）来填坑。

德国作为一个后发国家，在加勒比海地区并没有殖民地，自然也没有机会加入以法国为主的甘蔗糖产业链。而十八世纪，十九世纪正是重商主义的高峰时期。甜菜糖的发明可以说是进口替代战略的产物。作为甜菜糖的发明国和主要种植国，德国在甜菜糖的工业生产上自然是有一套的，展品中关于生产流程，机械化以及标准化上的内容相当丰富，设备的特化性很强。

虽然甜菜糖离消费地更近，在农业机械化上走得更远，也受到国家背书，却从来没有取代甘蔗糖的主导地位。甜菜糖与甘蔗糖的竞争，在某种意义上也可以说是农业工业化与农业外包化的竞争。甜菜糖的生产在德国，不论是小农场，还是季节工人，至少还有进厂这个选择，相比于加勒比海的奴隶或契约佣工来说是自由的，也能获得更高的报酬。离岸生产，无人关心工人，在机械化进度上自然也会迟缓。然而，在机械化与自动化只能带来有限优势的行业，技术创新带来的在岸生产力提高并不能与工资的差额对抗。人类社会的下限很低，劳动密集型产业，你不干有的是人干。

我们还是回到轻松的话题，看看展品。

在一个普通的展览馆里，看到单糖的空间结构我还是感到非常意外的。毕竟糖在水溶液中构成六元环结构，而不是分子式上的醛类本身已经超过了一般常识。而具体讲解如何用不同的方式去表现空间结构，仿佛从基础有机化学的教科书里抽出来一页。这仿佛让我回到了高中时在上课时间偷看大学教材的奇特的愉悦感，但是多数观众大约对此内容没有任何感受。展品到底出于什么目的来布展呢？作为博物馆，面向目标人群来布展自然是理所应当的；但是，能够把布展者的思考传播开的私心也是必然存在的。

作为一个不接收公款的私人博客，这里大概也就是我堆放私心的档案馆了罢。

Visiting technology museums in various places has always been a customary practice during my travels, and the Deutsches Technikmuseum mentioned in the text is no exception. However, this visit to an exhibition on sugar unexpectedly became the primer for my determination to start a blog website.

Sugar

Sugar is everywhere. Discover in the exhibition “Sugars and Beyond! Food – Matter – Energy” the exciting world of sugar.

Logo of the Deutsches Technikmuseum

Sugar, also known as the collective term for polyhydroxy aldehydes and polyhydroxy ketones, is an essential component in our daily lives. However, choosing it as the central theme for an entire exhibition is a highly challenging task. After all, it is merely a food ingredient that some people even tend to avoid today. Nevertheless, through sugar as a medium, we are able to connect the technological history and political circumstances spanning several centuries, resulting in a captivating story.

Five hundred years ago, sugar was a luxury item. Its scarcity is difficult to comprehend for people today: even if we find the retail price of sugar at that time and adjust it for purchasing power, the value remains abstract. Museums, however, serve as places to narrate the past through objects. So instead of focusing on refined sugar, highlighting the containers used for storing sugar reflects its value.

A sugar container, made of silver and sometimes equipped with a lock, was even widely used as part of a dowry in those times, emphasizing its value. As a Chinese, I am pretty surprised to see the word dowry shown in this context.

The first transformation in the production of sugar occurred through the triangular trade. While sugarcane originally originated in the Old World, its widespread commercial cultivation took place in the Caribbean region. Although sugarcane is a tropical crop that requires sunlight and higher temperatures for growth, the most crucial production factor was human labor. Initially, it involved African slaves in the triangular trade. After the abolition of slavery in the 1830s, limited-term indentured laborers from India, China, and Southeast Asia were brought in to fill the gap.

As a latecomer without colonies in the Caribbean, Germany did not have the opportunity to participate in the sugarcane industry chain dominated by France. The 18th and 19th centuries were the peak of mercantilism. The invention of sugar beets can be seen as a result of import substitution strategies. As the country that invented and majorly cultivated sugar beets, Germany naturally had a well-established system for industrial production of sugar beets. The exhibition includes abundant content on production processes, mechanization of cultivation, and standardization for beet sugar, with highly specialized instruments.

Although sugar beets were closer to the consumption market and had made more progress in agricultural mechanization, and also enjoyed state endorsement such as tariff against imported sugar, they never replaced the dominant position of sugarcane sugar in terms of global production. The competition between sugar beets and sugarcane sugar can also be seen as a competition between agricultural industrialization and agricultural outsourcing. In Germany, the production of sugar beets offered options not only for small farms but also for seasonal workers and factory workers. Compared to slaves or indentured laborers in the Caribbean, they had more freedom and could earn higher wages. Offshore production with no concern for workers naturally led to slower progress in mechanization. However, in industries where mechanization and automation only provide limited advantages, the increase in onshore productivity brought about by technological innovation cannot compete with wage differentials. You could have the full cost advantage of globalization, at the cost of unreliable supply chain, and maybe even social instability. As we navigate the turmoil of COVID and supply chain realignment, the sugar history is becoming more relevant.

Let's return to the lighter topic and take a look at the exhibits.

Seeing the spatial structure of monosaccharides in an ordinary exhibition hall was quite surprising. After all, the fact that sugar forms a six-membered ring structure in aqueous solution, rather than simply being an aldehyde based on its molecular formula, already exceeds general knowledge. The detailed explanation of different ways to represent spatial structure seemed like a page pulled out of a basic organic chemistry textbook. It reminded me of the peculiar pleasure of secretly reading university textbooks during high school class time. However, most visitors probably had no particular reaction to this content. What was the purpose of presenting these exhibits? As a museum, it is natural to design exhibitions targeting specific audiences. However, the personal motives of the exhibition organizers to convey their thoughts are inevitably present as well.

As a personal blog that doesn't receive public funds, this place is probably my archive for storing weird personal thoughts and hallucinations. Bear with me.