Concise Summary简洁概述
Science is powerful precisely because it suspends judgment until experimental evidence arrives — but that strength is also a weakness. When a question has huge future consequences yet cannot be tested right now (cryonics, long-range AI risk, certain evolutionary psychology hypotheses), Science's honest verdict is merely "not proven," which tells you almost nothing. Yudkowsky argues that Bayesian reasoning must fill the gap: absence of evidence is only evidence of absence when you would reasonably expect evidence to appear. Dismissing cryonics as "unscientific" — when reviving a frozen mammal is simply beyond today's technology — is a category error. A rational agent must make the calculated bet before the experiment is even possible.
科学之所以强大，恰恰在于它暂缓判断，直到实验证据到来——但这一优势同时也是弱点。当某个问题具有巨大的未来后果，却暂时无法通过实验检验（如人体冷冻、长远 AI 风险、部分进化心理学假说），科学的诚实裁决仅仅是「尚未证明」，几乎什么信息都给不了你。Yudkowsky 认为，贝叶斯推理必须填补这一空缺：「缺乏证据」只在你本有理由期待证据出现时，才构成「不存在的证据」。以「不科学」为由驳斥人体冷冻——而复活冰冻哺乳动物仅仅超出了当今技术能力——是一种范畴错误。理性行动者必须在实验甚至还不可行之前，就做出理性的押注。
Infographic信息图
Science's great strength is also its weakness
科学最大的优点也是它的弱点
Science says "go test it" without caring where your theory came from — which is powerful but leaves you stranded when no test is currently feasible.
科学说「去测试吧」，不关心你的理论从哪里来——这固然强大，却在测试暂时不可行时让你无所适从。
Cryonics as the key example
人体冷冻作为核心案例
Dismissing cryonics as "unscientific" misapplies the rules: Science's real verdict is "not proven," not "disproven." Absence of evidence is weak here because revival is technologically impossible right now.
以「不科学」驳斥人体冷冻是误用规则：科学真正的裁决是「尚未证明」，而非「已证伪」。因为复活目前技术上不可行，缺乏证据在这里只是弱证据。
Connected theoretical webs
相互连接的理论网络
Evolutionary psychology hypotheses that are untestable in isolation can be "indirectly supported" as part of a productive theoretical mesh — Science's binary verdicts miss this.
孤立地看无法检验的进化心理学假说，作为有效理论网络的一部分，可以获得「间接支持」——科学的二元裁决对此无能为力。
The calculated bet
理性的押注
When experimental verdict is decades away and stakes are mortal, a rational agent must reason to the best available estimate rather than wait for proof that arrives too late.
当实验裁决要几十年后才能到来、赌注事关生死时，理性行动者必须推理出当下最佳估计，而不是等待来得太晚的证据。
The ambulance analogy
救护车类比
Refusing cryonics until someone is actually revived is like refusing to board an ambulance until it is already at the hospital — Mike Li's analogy catches the absurdity.
拒绝人体冷冻，直到真有人被复活——就像拒绝上救护车，直到它已经到达医院——Mike Li 的类比一针见血地揭示了这一荒谬。
Detailed Summary详细概述
Yudkowsky opens with a personal confession: at 18 he held a stupid theory — consciousness caused by closed timelike curves in quantum gravity — and he followed every rule he had been told was part of Traditional Rationality. He believed testable things (microtubules supporting quantum coherence), and he would have been willing to say "I was wrong" after a negative result. Science would have been fine with this. But he did not want to waste ten years.
Science's Structural Limitation
Science, Yudkowsky argues, cannot tell you which theory is right in advance of experimental test. It says "go test it" without caring where the theory comes from. This is Science's great strength — it stops arguments by demanding evidence — and its great weakness: it offers no traction when the test is not yet available.
The Bayesian Correction
A commenter asks why Yudkowsky cares about untestable questions. His answer: questions that are easily and immediately testable are hard for Science to get wrong; the problem is with consequential questions where the test is delayed or extremely expensive.
The essay introduces the key Bayesian principle: absence of evidence is only evidence of absence to the degree that evidence would reasonably be expected to appear. Snake oil cures for cancer should produce detectable case reports if true; therefore the absence of case reports is strong evidence against them. But rare fossils are not expected even if intermediate species exist — so gaps in the fossil record are weak evidence against evolution.
Three Cases
Macroscopic quantum superpositions are in principle testable — just not cheaply or now. They establish the pattern: important consequences, delayed test.
Evolutionary psychology introduces a subtler problem. Some hypotheses are testable and some are not, but the untestable ones form the theoretical scaffolding that lets you generate the testable ones. Science gives a binary "proven/not proven" verdict that cannot handle "indirectly supported."
Cryonics is the culminating example — 150,000 people die per day, the future consequences are enormous, and we cannot revive a frozen mammal with today's technology. The Bayesian point: even if future nanotechnology could successfully revive cryonics patients, you would not expect a successful revival right now. So "no one has been revived" is not strong evidence against cryonics. As Mike Li observes, demanding revival before you sign up is like refusing to board an ambulance until it is already at the hospital.
The Deeper Point
Yudkowsky distinguishes two errors. First, some people confuse "not proven" with "disproven" — this is a step outside science, not a misstep within it. Second, some confuse "no feasible experiment" with "strong absence of evidence" — this is the deeper error he wants to correct.
The essay ends with a stark reminder: sometimes the disconfirming experimental result for civilizational bets is that your entire species has just been wiped out. You do not get to say "interesting data; I'll revise my theory now." You have to do the thing Science does not trust anyone to do — reason carefully before you are clubbed over the head with the answer.
Yudkowsky 以一则私人告白开篇：他在 18 岁时持有一个愚蠢的理论——意识由量子引力中的闭合类时曲线所导致——而他遵循了所有被告知属于「传统理性」的规则。他相信可检验的东西（神经元微管支持量子相干性），并且愿意在负面结果之后说「我错了」。从科学的角度，这完全没问题。但他不想浪费十年。
科学的结构性局限
Yudkowsky 认为，科学无法在实验测试之前告诉你哪个理论是正确的。它说「去测试吧」，并不在乎理论从何而来。这是科学最大的优势——它通过要求证据来终止争论——也是它最大的弱点：当测试暂时无法进行时，它毫无着力点。
贝叶斯修正
一位评论者问 Yudkowsky 为何关心不可检验的问题。他的回答是：容易且立即可检验的问题，科学几乎不会搞错；问题在于测试被延迟或代价极高的重大问题。
文章引入了关键的贝叶斯原则：「缺乏证据」只在证据本该合理出现的情况下，才构成「不存在的证据」。 如果蛇油真能治愈癌症，应该会出现可检测的案例报告；因此案例报告的缺席是强有力的反证。但即便中间物种真实存在，稀有化石也不必然存在——所以化石记录的空缺对进化论只是弱证据。
三个案例
宏观量子叠加在原则上是可测试的——只是当下太昂贵或不可行。它确立了一种模式：重大后果，测试延迟。
进化心理学引入了更微妙的问题。部分假说可检验，部分不可检验，但不可检验的那些构成了让你产生可检验假说的理论脚手架。科学给出的「已证明/未证明」二元裁决，无法处理「间接支持」这种情况。
人体冷冻是压轴案例——每天有 15 万人死亡，未来后果巨大，而我们无法用今天的技术复活冰冻的哺乳动物。贝叶斯的要点是：即便未来的纳米技术真的能成功复活冷冻患者，你也不该期待现在就看到复活成功。所以「从未有人被复活」不是反对人体冷冻的强证据。正如 Mike Li 所说，要等到真的有人被复活才愿意签约，就像拒绝上救护车、要等它已经到达医院才肯上一样。
更深层的要点
Yudkowsky 区分了两种错误。第一种：一些人把「未证明」混同于「已证伪」——这是迈出了科学之外的一步，而非科学内部的失误。第二种：一些人把「没有可行的实验」混同于「证据强烈缺席」——这是他真正想纠正的更深层错误。
文章以一个严峻的提醒作结：对于文明层面的押注，否定性的实验结果有时意味着你的整个物种刚刚被抹去。你没有机会说「有趣的数据，我现在来修正我的理论」。你必须做科学不信任任何人去做的那件事——在被答案砸中脑袋之前，仔细地推理。
FAQ常见问答
Is Yudkowsky saying we should ignore science and just reason from first principles?Yudkowsky 是在说我们应该忽视科学、只靠第一原理推理吗？
No. He explicitly says that when "definite unmistakable experimental evidence" is available, you should go with it — "why on Earth wouldn't you?" His point is narrower: when no feasible experiment exists, Science returns an uninformative "not proven" verdict and rational Bayesian reasoning must fill the gap.
不是。他明确表示，当「确定无误的实验证据」存在时，你就应该依据它——「你有什么理由不这么做呢？」他的观点更为具体：当没有可行的实验时，科学给出的是毫无信息量的「尚未证明」裁决，此时理性的贝叶斯推理必须填补这一空缺。
Why is the cryonics example specifically so important to him?为什么人体冷冻这个例子对他尤为重要？
Cryonics combines three features that make the Science-can't-help problem maximally vivid: enormous daily stakes (150,000 deaths), a clear path through which it could in principle work (preserving synaptic patterns until future nanotechnology), and a complete absence of feasible experiments right now. Most casual dismissals of cryonics commit the error he diagnoses: treating 'not proven' as 'disproven.'
人体冷冻结合了三个特征，使「科学无能为力」的问题变得最为鲜明：巨大的每日赌注（每天15万人死亡）、原则上可行的运作路径（保存突触模式直至未来纳米技术就绪），以及当下完全没有可行实验。大多数对人体冷冻的随手驳斥，都犯了他所指出的那个错误：把「尚未证明」当作「已证伪」。
What does "absence of evidence is evidence of absence" actually mean in Bayesian terms?「缺乏证据是不存在的证据」在贝叶斯意义上究竟是什么意思？
It means: failing to observe X lowers your credence that X exists — but only proportionally to how likely X would be to produce observable evidence. If a phenomenon would reliably generate detectable traces, no traces is damning. If a phenomenon would rarely leave traces even if real (rare fossils, a technology not yet invented), no traces is nearly uninformative. The formula is a likelihood ratio, not a binary rule.
意思是：没有观察到 X 会降低你对 X 存在的置信度——但仅按 X 产生可观测证据的可能性比例来降低。如果某个现象能可靠地产生可检测痕迹，没有痕迹就是强有力的反证。如果某个现象即便真实存在也很少留下痕迹（稀有化石、尚未发明的技术），没有痕迹几乎不包含信息。这个公式是一个似然比，而不是一条二元规则。
What is the 'indirectly supported' verdict that Science lacks?科学缺少的「间接支持」裁决是什么意思？
In evolutionary psychology, many hypotheses are mutually supporting: hypothesis A predicts which experiments you would run to test hypothesis B, and B's confirmation boosts A's credibility even though A was never directly tested. Science's binary verdicts — proven or not proven — cannot capture this 'part of a productive theoretical web' status. Yudkowsky suggests we need a third category.
在进化心理学中，许多假说相互支持：假说 A 预测了你该用什么实验来检验假说 B，B 的证实提升了 A 的可信度，即便 A 从未被直接检验。科学的二元裁决——已证明或未证明——无法捕捉这种「有效理论网络的组成部分」的状态。Yudkowsky 认为我们需要第三个类别。
What is the essay's practical takeaway for an individual?这篇文章对个人有什么实用启示？
When you face a question with large personal consequences but no feasible experiment, do not hide behind "the science isn't settled." Instead, (1) apply Bayesian reasoning to ask what evidence would appear if each hypothesis were true, (2) calibrate how much the absence of current evidence actually tells you, and (3) make the best-reasoned bet available rather than waiting for certainty that may never come — or may arrive too late.
当你面对一个具有重大个人后果却没有可行实验的问题时，不要躲在「科学尚无定论」后面。而是：（1）用贝叶斯推理追问：如果每个假说为真，会出现什么证据；（2）校准当前证据的缺席究竟能告诉你多少；（3）做出当下能做的最佳理性押注，而不是等待可能永远不会来、或来得太晚的确定性。
How does this essay relate to the earlier piece 'Science Doesn't Trust Your Rationality'?这篇文章与之前的「科学不信任你的理性」有何关联？
That essay argued that Science institutionally mistrusts individual judgment and demands replicable external tests as a corrective. This essay accepts that diagnosis but reveals its flip side: the institutional mistrust leaves agents helpless on questions where tests are delayed or impossible. Together they frame the core tension of the Sequences: Science is the right gold standard when available, but rationality must act without waiting for it.
那篇文章认为，科学从制度上不信任个人判断，要求可重复的外部测试作为纠偏机制。这篇文章接受这一诊断，但揭示了它的另一面：当测试被延迟或不可能时，这种制度性不信任让行动者陷入无所适从的困境。两篇文章共同勾勒出系列的核心张力：科学在可用时是正确的黄金标准，但理性必须在不等待它的情况下行动。
In-depth Analysis · Pros & Cons深入解读 · 优缺点
This essay sits at the intersection of philosophy of science and practical decision theory. Yudkowsky's target is a widespread and underexamined failure mode: using "the science isn't settled" as a reason for inaction on questions that, by their nature, cannot settle until it is too late.
这篇文章处于科学哲学与实践决策理论的交叉点。Yudkowsky 的靶子是一种普遍而未被充分审视的失败模式：以「科学尚无定论」为由，对那些按其本质在为时已晚之前都无法定论的问题拒绝行动。
- Precise diagnosis of a real failure mode对真实失败模式的精准诊断The essay correctly identifies that 'not proven' does not equal 'disproven,' and that this conflation is common and costly. The cryonics ambulance analogy crystallizes the error vividly.文章正确指出「尚未证明」不等于「已证伪」，这一混淆普遍存在且代价高昂。人体冷冻的救护车类比以极为生动的方式定格了这一错误。
- The likelihood-ratio framing of absence of evidence「缺乏证据」的似然比框架The fossil/snake-oil contrast is a clean and memorable way to teach conditional reasoning: what you would expect to observe given each hypothesis, not just whether you observed it.化石/蛇油的对比是一种简洁而难忘的方式来教授条件推理：关键在于给定每个假说时你预期会观察到什么，而不仅仅是你是否观察到了。
- Self-implicating opening自我牵涉的开篇Starting from his own youthful error — a testable but wrong theory he faithfully followed Scientific norms on — grounds the abstract argument in a concrete, honest narrative.从他自己年轻时的错误出发——一个可检验但错误的理论，他忠实地遵循了科学规范——把抽象论证锚定在一个具体、诚实的叙事中。
- The civilizational-stakes closer文明赌注的结尾The final paragraph — in which the disconfirming experimental result is species extinction — is unusually forceful rhetoric that correctly scales the stakes beyond individual life decisions to existential risk.最后一段——否定性实验结果就是物种灭绝——是异常有力的修辞，正确地将赌注从个人生死决策扩展到了存在性风险的量级。
- The cryonics argument assumes a specific mechanism人体冷冻论证预设了特定机制Yudkowsky's Bayesian defense of cryonics hinges on whether current freezing methods actually preserve the synaptic patterns he argues are identity-preserving. This is a contested empirical question he treats as settled by reference to other essays, not demonstrated here.Yudkowsky 对人体冷冻的贝叶斯辩护，关键在于当前的冷冻方法是否真的保存了他所认为构成身份认同的突触模式。这是一个有争议的实证问题，他通过参引其他文章将其视为已解决，而非在这里加以论证。
- Underdetermination problem for rational pre-experiment reasoning理性先验推理的欠定性问题The essay tells us to reason carefully when experiments are not available, but does not address how to avoid the trap it opens its own example with: the 18-year-old Eliezer's theory was also arrived at by careful reasoning. Bayesian priors for novel empirical claims are notoriously hard to set without experimental anchoring.文章告诉我们在实验不可行时要仔细推理，却没有解决它开篇自己案例所揭示的陷阱：18岁的 Eliezer 的理论也是经过仔细推理得出的。对于新颖实证主张，在没有实验锚定的情况下设定贝叶斯先验是出了名的困难。
- The 'indirectly supported' category is underspecified「间接支持」类别缺乏规范Yudkowsky gestures at the need for a third verdict ('indirectly supported') but provides no criteria for when a hypothesis earns this status versus when it is mere speculation dressed as theory. Without guardrails, 'part of a connected web' could excuse almost anything.Yudkowsky 点到了需要第三种裁决（「间接支持」），但没有给出一个假说在何时赢得这一地位、何时只是披着理论外衣的纯粹推测的判断标准。没有护栏，「相互连接的网络的一部分」几乎可以为任何东西背书。
- Selection bias in which pre-experimental bets to make选择做哪些先验押注时存在选择偏差The essay argues powerfully for making rational bets before experiments are possible, but does not address how to handle the proliferation of such bets. Rational Bayesian agents placing many pre-experimental bets will still be wrong on most of them; the essay offers no method for prioritizing which bets are worth making.文章有力地论证了在实验可行之前做出理性押注，但没有讨论如何应对这类押注的泛滥。进行大量先验押注的理性贝叶斯行动者，仍然会在大多数押注上出错；文章没有提供任何方法来排定哪些押注值得做出。
A sharp and practically important essay that correctly identifies a gap in how most people apply scientific norms. Its core argument — that 'not proven' should not be read as 'disproven' when the absence of evidence is expected even under the favorable hypothesis — is both Bayesian-correct and underappreciated. The main gap is that it diagnoses the problem clearly but offers limited guidance on how to actually conduct pre-experimental reasoning without falling into the trap the author himself nearly fell into at 18.
一篇敏锐而具有实践意义的文章，正确地指出了大多数人在应用科学规范时存在的盲区。其核心论证——当有利假说下证据的缺席本就在预期之中时，「尚未证明」不应被解读为「已证伪」——既符合贝叶斯原则，又长期被低估。主要缺口在于：文章清晰地诊断了问题，却对如何在不陷入作者本人18岁时险些落入的陷阱的前提下实际开展先验推理，提供了有限的指引。
Original Text原文
Once upon a time, a younger Eliezer had a stupid theory. Let's say that Eliezer~18~'s stupid theory was that consciousness was caused by closed timelike curves hiding in quantum gravity. This isn't the whole story, not even close, but it will do for a start.
And there came a point where I looked back, and realized:
- I had carefully followed everything I'd been told was Traditionally Rational, in the course of going astray. For example, I'd been careful to only believe in stupid theories that made novel experimental predictions, e.g., that neuronal microtubules would be found to support coherent quantum states.
- Science would have been perfectly fine with my spending ten years trying to test my stupid theory, only to get a negative experimental result, so long as I then said, "Oh, well, I guess my theory was wrong."
From Science's perspective, that is how things are supposed to work—happy fun for everyone. You admitted your error! Good for you! Isn't that what Science is all about?
But what if I didn't want to waste ten years?
Well... Science didn't have much to say about that. How could Science say which theory was right, in advance of the experimental test? Science doesn't care where your theory comes from—it just says, "Go test it."
This is the great strength of Science, and also its great weakness.
Eliezer, why are you concerned with untestable questions?
Because questions that are easily immediately tested are hard for Science to get wrong.
I mean, sure, when there's already definite unmistakable experimental evidence available, go with it. Why on Earth wouldn't you?
But sometimes a question will have very large, very definite experimental consequences in your future—but you can't easily test it experimentally right now—and yet there is a strong rational argument.
Macroscopic quantum superpositions are readily testable: It would just take nanotechnologic precision, very low temperatures, and a nice clear area of interstellar space. Oh, sure, you can't do it right now, because it's too expensive or impossible for today's technology or something like that—but in theory, sure! Why, maybe someday they'll run whole civilizations on macroscopically superposed quantum computers, way out in a well-swept volume of a Great Void. (Asking what quantum non-realism says about the status of any observers inside these computers, helps to reveal the underspecification of quantum non-realism.)
This doesn't seem immediately pragmatically relevant to your life, I'm guessing, but it establishes the pattern: Not everything with future consequences is cheap to test now.
Evolutionary psychology is another example of a case where rationality has to take over from science. While theories of evolutionary psychology form a connected whole, only some of those theories are readily testable experimentally. But you still need the other parts of the theory, because they form a connected web that helps you to form the hypotheses that are actually testable—and then the helper hypotheses are supported in a Bayesian sense, but not supported experimentally. Science would render a verdict of "not proven" on individual parts of a connected theoretical mesh that is experimentally productive as a whole. We'd need a new kind of verdict for that, something like "indirectly supported".
Or what about cryonics?
Cryonics is an archetypal example of an extremely important issue (150,000 people die per day) that will have huge consequences in the foreseeable future, but doesn't offer definite unmistakable experimental evidence that we can get right now.
So do you say, "I don't believe in cryonics because it hasn't been experimentally proven, and you shouldn't believe in things that haven't been experimentally proven?"
Well, from a Bayesian perspective, that's incorrect. Absence of evidence is evidence of absence only to the degree that we could reasonably expect the evidence to appear. If someone is trumpeting that snake oil cures cancer, you can reasonably expect that, if the snake oil was actually curing cancer, some scientist would be performing a controlled study to verify it—that, at the least, doctors would be reporting case studies of amazing recoveries—and so the absence of this evidence is strong evidence of absence. But "gaps in the fossil record" are not strong evidence against evolution; fossils form only rarely, and even if an intermediate species did in fact exist, you cannot expect with high probability that Nature will obligingly fossilize it and that the fossil will be discovered.
Reviving a cryonically frozen mammal is just not something you'd expect to be able to do with modern technology, even if future nanotechnologies could in fact perform a successful revival. That's how I see Bayes seeing it.
Oh, and as for the actual arguments for cryonics—I'm not going to go into those at the moment. But if you followed the physics and anti-Zombie sequences, it should now seem a lot more plausible, that whatever preserves the pattern of synapses, preserves as much of "you" as is preserved from one night's sleep to morning's waking.
Now, to be fair, someone who says, "I don't believe in cryonics because it hasn't been proven experimentally" is misapplying the rules of Science; this is not a case where science actually gives the wrong answer. In the absence of a definite experimental test, the verdict of science here is "Not proven". Anyone who interprets that as a rejection is taking an extra step outside of science, not a misstep within science.
John McCarthy's Wikiquotes page has him saying, "Your statements amount to saying that if AI is possible, it should be easy. Why is that?" The Wikiquotes page doesn't say what McCarthy was responding to, but I could venture a guess.
The general mistake probably arises because there are cases where the absence of scientific proof is strong evidence—because an experiment would be readily performable, and so failure to perform it is itself suspicious. (Though not as suspicious as I used to think—with all the strangely varied anecdotal evidence coming in from respected sources, why the hell isn't anyone testing Seth Roberts's theory of appetite suppression?)
Another confusion factor may be that if you test Pharmaceutical X on 1000 subjects and find that 56% of the control group and 57% of the experimental group recover, some people will call that a verdict of "Not proven". I would call it an experimental verdict of "Pharmaceutical X doesn't work well, if at all". Just because this verdict is theoretically retractable in the face of new evidence, doesn't make it ambiguous.
In any case, right now you've got people dismissing cryonics out of hand as "not scientific", like it was some kind of pharmaceutical you could easily administer to 1000 patients and see what happened. "Call me when cryonicists actually revive someone," they say; which, as Mike Li observes, is like saying "I refuse to get into this ambulance; call me when it's actually at the hospital". Maybe Martin Gardner warned them against believing in strange things without experimental evidence. So they wait for the definite unmistakable verdict of Science, while their family and friends and 150,000 people per day are dying right now, and might or might not be savable—
—a calculated bet you could only make rationally.
The drive of Science is to obtain a mountain of evidence so huge that not even fallible human scientists can misread it. But even that sometimes goes wrong, when people become confused about which theory predicts what, or bake extremely-hard-to-test components into an early version of their theory. And sometimes you just can't get clear experimental evidence at all.
Either way, you have to try to do the thing that Science doesn't trust anyone to do—think rationally, and figure out the answer before you get clubbed over the head with it.
(Oh, and sometimes a disconfirming experimental result looks like: "Your entire species has just been wiped out! You are now scientifically required to relinquish your theory. If you publicly recant, good for you! Remember, it takes a strong mind to give up strongly held beliefs. Feel free to try another hypothesis next time!")
曾经，一个年轻的 Eliezer 持有一个愚蠢的理论。姑且说 Eliezer~18~ 的愚蠢理论是：意识由隐藏在量子引力中的闭合类时曲线所引发。这并不是故事的全部，甚至远不是，但作为开头已经够了。
后来有一天，我回头审视，意识到：
- 在走偏的过程中，我仔细遵循了所有被告知属于传统理性的东西。例如，我很小心地只相信能做出新颖实验预测的愚蠢理论，比如预测神经元微管将被发现能支持相干量子态。
- 科学对我花十年时间去检验我的愚蠢理论、最后得到一个否定结果、然后说「哦，好吧，我猜我的理论是错的」，完全没有意见。
从科学的角度来看，这正是事情应该运作的方式——皆大欢喜。你承认了错误！太好了！这不正是科学的精髓所在吗？
但如果我不想浪费十年呢？
嗯……科学对这件事没有太多可说的。科学怎么能在实验测试之前判断哪个理论是正确的呢？科学不关心你的理论从哪里来——它只说「去测试吧」。
这是科学最大的优势，也是它最大的弱点。
Eliezer，你为什么关心不可检验的问题？
因为容易立即可检验的问题，科学很难搞错。
我是说，当然了，当已经有明确无误的实验证据时，就依据它。你有什么理由不这么做呢？
但有时一个问题在你的未来会产生非常巨大、非常确定的实验后果——但你现在无法轻易地实验检验它——然而确实存在一个强有力的理性论证。
宏观量子叠加是完全可以检验的：只需要纳米技术精度、极低温，以及一大片清空的星际空间。当然，你现在做不到，因为代价太高，或者是当今技术不可及之类的原因——但在理论上，当然可以！说不定有一天，他们会在某个大虚空的净空区域，用宏观叠加的量子计算机运转整个文明。（问量子非实在论对这些计算机内部观察者的地位有何说法，有助于揭示量子非实在论的欠规范之处。）
我猜这对你的日常生活并不立即有实用价值，但它确立了一个规律：并非所有具有未来后果的事物，现在检验都是便宜的。
进化心理学是另一个理性必须接替科学的案例。虽然进化心理学的理论形成一个相互连接的整体，但其中只有部分理论容易通过实验检验。但你仍然需要那些不可直接检验的部分，因为它们构成了相互关联的网络，帮助你形成那些真正可检验的假说——随后那些辅助性假说在贝叶斯意义上得到支持，但并非得到实验支持。科学对一个相互连接的理论网络中单独的一部分会给出「未获证明」的裁决，哪怕整个网络作为整体具有实验生产力。我们需要一种新的裁决类别，比如「间接支持」。
人体冷冻呢？
人体冷冻是一个极端重要问题（每天有 15 万人死亡）的典型案例，它在可预见的未来将产生巨大后果，却无法提供我们现在就能获得的明确无误的实验证据。
那么你是否该说：「我不相信人体冷冻，因为它没有得到实验证明，而你不应该相信未经实验证明的东西？」
从贝叶斯的角度来看，这是错误的。缺乏证据是不存在的证据，但仅限于我们可以合理预期证据出现的程度。如果有人大肆鼓吹蛇油能治愈癌症，你完全可以合理预期：如果蛇油真的在治愈癌症，一定会有科学家在进行对照研究来验证它——至少，医生会在报告神奇康复的案例研究——因此，这类证据的缺席是强有力的不存在证据。但「化石记录中的空缺」并不是反对进化论的强证据；化石只是偶然形成的，即便一个过渡物种真实存在，你也不能以高概率期待大自然会乖乖把它化石化，而且那块化石会被发现。
用现代技术复活一只低温冷冻的哺乳动物，就不是你可以合理预期做到的事，即便未来的纳米技术真的能成功复活也一样。这就是我理解贝叶斯是怎么看这件事的。
哦，至于支持人体冷冻的实际论证——我现在不打算展开。但如果你读过物理学和反僵尸序列，你现在应该觉得这件事可信得多了：无论是什么东西保存了突触模式，它就保存了你从一晚睡眠到次日清晨醒来所保存的那么多的「你」。
现在，平心而论，一个说「我不相信人体冷冻，因为它没有得到实验证明」的人是在误用科学的规则；这不是科学真的给出了错误答案的情况。在没有确定实验测试的情况下，科学在这里的裁决是「未获证明」。任何把这解读为拒绝的人，都在科学之外多走了一步，而非在科学内部走错了一步。
约翰·麦卡锡的维基语录页面引用了他的话：「你的话等于在说：如果人工智能是可能的，它应该很容易。为什么会这样呢？」维基语录页面没有说麦卡锡在回应什么，但我可以猜个大概。
这种普遍性错误大概源于：确实存在「缺乏科学证明是强证据」的情况——因为实验本来可以轻易进行，所以没有进行实验本身就令人生疑。（虽然不像我以前认为的那么可疑——各方受人尊敬的来源不断涌入各种奇怪的轶事证据，为什么就是没有人测试Seth Roberts 的食欲抑制理论？）
另一个混淆因素可能是：如果你对 1000 名受试者测试药物 X，发现对照组 56% 康复，实验组 57% 康复，有些人会称这为「未获证明」的裁决。我会称之为「药物 X 效果不佳（如果有效的话）」的实验性裁决。仅仅因为这个裁决在面对新证据时理论上可以撤销，并不意味着它是模糊的。
无论如何，现在有人将人体冷冻随手驳斥为「不科学」，就好像它是某种你可以轻易给 1000 名患者服用然后观察结果的药物一样。「等冷冻人体的人真的复活了再来找我，」他们说；正如 Mike Li 所观察到的，这就像是说「我不上这辆救护车；等它真的到了医院再来找我」。也许马丁·加德纳曾经警告他们不要在没有实验证据的情况下相信奇怪的事情。所以他们等待科学那确定无误的裁决，而与此同时，他们的家人和朋友以及每天 15 万人正在此刻死去，也许能被救、也许不能——
——这是一个只能通过理性才能做出的押注。
科学的驱动力是积累一座巨大到连容易犯错的人类科学家都无法误读的证据之山。但即便如此，有时也会出错，当人们对哪个理论预测了什么感到困惑时，或者当极难检验的成分被烘进了理论的早期版本时。而有时你根本就得不到清晰的实验证据。
无论如何，你都必须尝试去做那件科学不信任任何人去做的事情——理性地思考，在被答案砸中脑袋之前找出答案。
（哦，有时候一个否定性的实验结果是这样的：「你的整个物种刚刚被抹去！你现在在科学上必须放弃你的理论。如果你公开认错，太好了！记住，放弃坚定的信念需要强大的心智。下次随意再试另一个假说吧！」）