HackerTube English?
表示、直觉与反馈 →

档案没有记下来,就说明它没有发生吗?

一位历史学家进了档案馆,却只能看其中一个时期。一位飞行员目睹了异常,流传的却不是那次飞行的录像。从两位证人的经历出发,分清世界里缺了什么,和记录里缺了什么。

本文的问题 · 23
  1. 进得了档案馆,为什么还看不见另一方?
  2. 一段影像是谁的证词?
  3. 名册上有名字,为什么仍难追踪谁在推动?
  4. 分母里还剩下哪些任务?
  5. 档案里有六万份文件,时间线在哪里?
  6. 照片里的标尺和冰退后的树林
  7. 谈话录下来了,还能在哪一步传错?
  8. 照片里少了一截树,怎样让别人复核?
  9. 一段嗡鸣怎样把两个时间轴对齐?
  10. 同一则故事被许多地方引用,究竟增加了几份证据?
  11. 楼梯上的目击者能把先后顺序钉牢吗?
  12. 一首诗在古书里出现,它也一样古老吗?
  13. 数据里没有一次捐赠、一次照护或一位行人,哪种缺失?
  14. 没有收到信号,能知道信号在哪一步消失吗?
  15. 地图上的空白与字典里的空白,谁定了收录规则?
  16. 居民测到很响,到底还需要独立记录什么?
  17. 一封匿名信能把调查推进到哪里?
  18. 同一起校园事件,病毒式传播的片段和后来采访分别能证明什么?
  19. 从管道和运河,能读出决策者的动机吗?
  20. 回过头看得清楚,当时也看得见吗?
  21. 国家排名和工厂关门,记录的是同一笔损失吗?
  22. 被接收的论文,和没有进入故事的投稿
  23. 图例写了一个名字,哪些差异仍要另外查?

进得了档案馆,为什么还看不见另一方?

Paine 讲了一个查史料时才会撞上的细节:她进了俄国外交档案馆,获准查阅的却只有沙皇时期。后来改查美国总统档案,得到了一批新证词,却没有因此听见所有缺席者。把同样的问题交给研究古 DNA 的 Reich:当一个谱系几乎没有取样时,图上的空白可以说明样本稀少,不能直接说明那里没有人。

原文与时间戳 · 3
I actually got into the foreign ministry archives there, but only for the czarist period.
Sarah C. M. Paine — Why dictators keep making the same fatal mistake · ▶ 46:05–46:15
spent two months every summer going to U.S. presidential archives.
Sarah C. M. Paine — Why dictators keep making the same fatal mistake · ▶ 46:34–46:45
we really have no sampling except for the Denisovans and Neanderthals right now.
David Reich — How one small tribe conquered the world 70,000 years ago · ▶ 116:17–116:21

一段影像是谁的证词?

Fravor 给出了足以改变阅读顺序的一句话:他自己那次目击没有视频。后来出动的机组带回了影像,相机还可以改变视场。先画两行:第一行是目击者看见了什么,第二行是另一架飞机的设备记录了什么。两行可以互相提出问题,但若把它们叠成「目击过程的录像」,就凭空多出了一份证据。Paine 换档案馆的经历给了同一个检查办法:换记录者时,从哪一句开始需要重新署名?

原文与时间戳 · 3
There is no video from our event. It was all four sets of eyeballs staring at this thing.
David Fravor: UFOs, Aliens, Fighter Jets, and Aerospace Engineering | Lex Fridman Podcast #122 · ▶ 75:15–75:19
a better zoom capability when you go into TV mode
David Fravor: UFOs, Aliens, Fighter Jets, and Aerospace Engineering | Lex Fridman Podcast #122 · ▶ 94:28–94:31
I actually got into the foreign ministry archives there, but only for the czarist period.
Sarah C. M. Paine — Why dictators keep making the same fatal mistake · ▶ 46:05–46:15

名册上有名字,为什么仍难追踪谁在推动?

费城无线网络一案里,Gurley 描述的是从电信公司的说客到议案的一条路径。但到了会面记录,登记的可能是中间人的名字。Weinstein 则怀疑学术声誉使某些想法较难出现在刊物上,同时也承认,许多偏离共识的人确实走错了路。「漏记」与「被漏掉的人说得对」是两个命题。若要追第一条路径,下一份有用的材料是议案版本和署名的对应关系,而不是另一个同意你的评论。

原文与时间戳 · 3
proposal drafted by lobbyists for the telecommunications companies.
All-In Summit: Bill Gurley presents 2,851 Miles · ▶ 4:37–4:41
They show up as all those people. There's a lot of obfuscation that's out there.
All-In Summit: Bill Gurley presents 2,851 Miles · ▶ 26:00–26:06
do not usually subscribe to the consensus position and almost always lose their way.
Eric Weinstein: Revolutionary Ideas in Science, Math, and Society | Lex Fridman Podcast #16 · ▶ 44:36–44:42

分母里还剩下哪些任务?

Kriesel 画了四个计划停站,两个准点,一个晚点,一个取消。若只计算真正到站的三站,准点率约为 67%;用计划停站作分母,同一班车只有 50% 兑现了准点承诺。把这个算例带回一次工具演示:展示十次得到结果的回答时,再数一数有几次超时、跳过或没有作答。两种分数都能算,但使用工具的人需要的是哪一个?

原文与时间戳 · 2
alle Stopps, die nicht ausgefallen sind
BahnMining - Pünktlichkeit ist eine Zier (David Kriesel) · ▶ 35:51–35:55
Wir zählen alle Stopps, die geplant waren
BahnMining - Pünktlichkeit ist eine Zier (David Kriesel) · ▶ 35:58–36:08

档案里有六万份文件,时间线在哪里?

律师回忆拿到大量企业文件后,还得把时间与事件一一串起来。平台治理访谈则提出相反的困难:复核者可能拿到一条孤立消息,却缺少前面的协调行为。一个材料太多,一个上下文太少,下一步都可以问同一件具体的事:哪条记录能把事件甲和事件乙按时间接起来?

原文与时间戳 · 4
more than 60,000 documents
The Biggest Chemical Coverup In History · ▶ 17:40–17:43
I actually did read all of that
The Biggest Chemical Coverup In History · ▶ 18:02–18:04
there's a choice to appeal any action
Joe Rogan Experience #1258 - Jack Dorsey, Vijaya Gadde & Tim Pool · ▶ 4:30–4:34
our team having a lack of context
Joe Rogan Experience #1258 - Jack Dorsey, Vijaya Gadde & Tim Pool · ▶ 87:59–88:03

照片里的标尺和冰退后的树林

Carlson 在地貌照片里指出一栋三层楼,用来让人看清波痕的尺度。在另一段访谈中,他和 Hancock 谈到冰退后露出的树林和被搬运的巨石。标尺有助于判断现象的大小;它无法给地貌定年。若要把两处地点连成一场事件,下一张更有帮助的图会标出各自独立测出的年代及误差。

原文与时间戳 · 4
Missoula is in a basin
Joe Rogan Experience #606 - Randall Carlson · ▶ 70:15–70:20
a three story building you see right there for scale
Joe Rogan Experience #606 - Randall Carlson · ▶ 84:46–84:51
revealing that there were forests there
Joe Rogan Experience #725 - Graham Hancock & Randall Carlson · ▶ 62:54–62:57
carried in ice
Joe Rogan Experience #725 - Graham Hancock & Randall Carlson · ▶ 104:23–104:28

谈话录下来了,还能在哪一步传错?

Duane 回忆一段没有录音的交谈。到了庭上,警员和当事人各自记得不同的话。补一份录音可以核查「当时说了什么」。但把这个办法带进他后面的反例:若另一个证人真诚地认错了人,完整的谈话录音也没有拍下那次辨认。要解决争议,先把想核查的主语写出来:是谁说了什么,还是谁真的做了什么?

原文与时间戳 · 2
And then there was a conversationthat was not recorded.
Don't Talk to the Police · ▶ 4:17–4:19
even mistaken and unreliable evidence
Don't Talk to the Police · ▶ 22:42–22:44

照片里少了一截树,怎样让别人复核?

拍到远岸的一棵树后,Olson 希望另一个人可以复核低机位到底看不见哪一截。他把能认出来的地标也拍进去,记录相机离水面的高度。这样即使读者怀疑剪辑,仍可以回到同一位置重新观察。影片后来追查网上发帖者的身份时,遇到另一种缺口:原始记录已有部分被删。当年谁写了那条帖子,不能像湖边机位那样靠重新摆一次来核实。读这两段时可以问:如果丢掉眼前这一份证据,我该重做一次观察,还是必须找到当时留下的另一份副本?

原文与时间戳 · 2
include an identifiable landmark in the shots
In Search Of A Flat Earth · ▶ 8:09–8:12
a lot of that evidence was poorly archived and has since been deleted.
In Search Of A Flat Earth · ▶ 48:31–48:36

一段嗡鸣怎样把两个时间轴对齐?

给一段无日期录音标时间,先得有另一个观察者:连续记录电网频率的人。若两条曲线中同一串微小起伏能对上,录音就得到候选时间。在一份样本里分析者转看约 100 赫兹的谐波才找到轨迹。Zoz 的网络案例也依赖分别观察两端;Crenshaw 的调查案例还额外用到重复出现的网名。并排比较时,别只问记录是否相似,还要问候选中有多少条能同样吻合、身份由哪份独立材料确认。

原文与时间戳 · 6
record the grid frequency data
The hidden background noise that can catch criminals · ▶ 1:11–1:16
we'll be able to see the harmonic component around 100 hertz.
The hidden background noise that can catch criminals · ▶ 3:09–3:12
It also shifts the trust model over to the VPN provider
DEF CON 22 - Zoz - Don't Fuck It Up! · ▶ 15:14–15:25
there's no compromise necessary of TOR to acquire the circumstantial evidence
DEF CON 22 - Zoz - Don't Fuck It Up! · ▶ 23:56–24:06
looking around for the silk road on the public Internet
DEF CON 22 - Adrian Crenshaw- Dropping Docs on Darknets: How People Got Caught · ▶ 36:39–36:45
using the same handle in multiple places
DEF CON 22 - Adrian Crenshaw- Dropping Docs on Darknets: How People Got Caught · ▶ 43:43–44:00

同一则故事被许多地方引用,究竟增加了几份证据?

Tom Scott 曾根据博物馆网站讲述消防保险牌匾,后来委托研究者查政策与个案,才把识别投保房屋和决定是否救火两件事分开。英国图书馆的工作人员提醒,连努力收齐出版物都需要主动找到没有自称出版者的人。Hank Green 对微塑料标题的追踪则显示,多次报道可能只是同一区间上限被反复传递。统计三条新闻与找到三个独立测量之间,差了一次来源追溯。

原文与时间戳 · 6
fire marks were an indicator of whether a building was insured
I was wrong (and so was everyone) · ▶ 2:50–2:55
their firefighters did try to fight all fires
I was wrong (and so was everyone) · ▶ 2:58–3:01
people who don't necessarily think of themselves as publishers.
This library has every book ever published. · ▶ 1:14–1:20
we can't decide today what's going to be important in 50 years' time.
This library has every book ever published. · ▶ 4:15–4:19
their own range of 0.1 grams to 5 grams
Are You Eating a Credit Card Every Week? · ▶ 0:56–1:01
They just said 5 grams.
Are You Eating a Credit Card Every Week? · ▶ 1:52–1:54

楼梯上的目击者能把先后顺序钉牢吗?

影片把证人走过楼梯的说法,和她是否在枪响后几秒就经过的说法分开询问。后一个命题比前一个多了时钟,重建时间线时不能偷偷借用前一个命题的确定性。回看电网嗡鸣的案例,会发现两者的差别在哪里:如果没有同时留下的外部时钟,答不出的秒数应当在图上留下一个区间。

原文与时间戳 · 3
the number and origin of the gunshots.
The Kennedy Assassination: Inside the Book Depository · ▶ 33:57–34:00
Did it happen within seconds of the shooting?
The Kennedy Assassination: Inside the Book Depository · ▶ 49:42–49:44
record the grid frequency data
The hidden background noise that can catch criminals · ▶ 1:11–1:16

一首诗在古书里出现,它也一样古老吗?

已知的网页反复转载同一首诗,却不给作者和日期。Grey 找到更老的书页时,发现诗在 1722 年版的编辑脚注里。此时能钉住的,是这位编辑在那一年引了这首诗。若把书的初成时间当作诗的日期,就又把后人的记录塞回了一次较早的事件里,和把后一次航班的影像标成前一次目击是同一类时间错误。

原文与时间戳 · 2
never with the poem's date or an author attached.
Alexander Pope Wrote a 35-Page Takedown of a Historian · ▶ 1:42–1:45
It's a footnote added by the editor of this 1722 edition.
Alexander Pope Wrote a 35-Page Takedown of a Historian · ▶ 7:37–7:41

数据里没有一次捐赠、一次照护或一位行人,哪种缺失?

Feeney 主动隐去名字,使捐助很难从公开报道中数出来。Yang 讨论的无偿照护则有另一种缺失:工作确实存在,衡量市场交易的数字却不直接给它标价。以城市游戏中的步行与停车为第三个例子,先核对模型是否记录这类空间成本,再解释一条无人行走的道路。三种情况分别要追查披露方式、计量规则和模拟边界。

原文与时间戳 · 6
he’d always insisted on remaining anonymous
Secret Billionaire: The Chuck Feeney Story · ▶ 20:04–20:08
his message of giving while living to inspire others
Secret Billionaire: The Chuck Feeney Story · ▶ 51:41–51:46
what is her work valued at in GDP?
Joe Rogan Experience #1245 - Andrew Yang · ▶ 58:29–58:32
evolve from GDP as a measuring stick
Joe Rogan Experience #1245 - Andrew Yang · ▶ 58:42–58:46
the parking lots next to Dodgers Stadium in LA
Cities Without People · ▶ 6:53–6:58
theres a sidewalk nobody ever walks on.
Cities Without People · ▶ 11:35–11:37

没有收到信号,能知道信号在哪一步消失吗?

费米悖论的影片把困难步骤放在不同位置,都可能解释尚无探测结果。另一份演示直接改变可居住行星的寿命上限和困难步骤数,重新计算我们处在多早的位置。观测保持不变,结论却随假设变动。此时更有区分力的做法是问:未来哪种观测,只有其中一组假设才很难解释?

原文与时间戳 · 4
a barrier that is really hard for life to overcome
The Fermi Paradox — Where Are All The Aliens? (1/2) · ▶ 3:34–3:44
We have no evidence that there's any life besides us.
The Fermi Paradox — Where Are All The Aliens? (1/2) · ▶ 5:11–5:15
the number of hard steps, which is 6
Humanity was born way ahead of its time. The reason is grabby aliens. · ▶ 6:28–6:32
maximum planet lifetime: 10 billion years
Humanity was born way ahead of its time. The reason is grabby aliens. · ▶ 6:32–6:35

地图上的空白与字典里的空白,谁定了收录规则?

国界线随制图者采用的承认标准而变,名字的首次出现也随检索者接受的拼法而变。Grey 追踪 Tiffany 时强调,早期英语拼写缺少统一规则。对任一「最早」或「总共多少」的主张,都可以先试一种反查:把一个确有行政活动却未获普遍承认的地方,或一个发音相近但拼写不同的人名,送进同一条检索规则,观察它会被保留还是丢掉。

原文与时间戳 · 4
borders will differ depending on who you got the map from.
Some Countries Exist Only on Paper · ▶ 0:05–0:09
governments that issue passports
Some Countries Exist Only on Paper · ▶ 1:46–1:49
with its exact modern spelling.
Tiffany: The Name That Cheats Death · ▶ 4:31–4:35
Consistent spelling in the English language totally wasn't a thing back then.
Tiffany: The Name That Cheats Death · ▶ 5:16–5:20

居民测到很响,到底还需要独立记录什么?

矿场旁住户展示在家中获得的读数,并描述连续噪声与车流声的差别。下一份有用的记录应同时保留测点、时段、设备设定和背景对照。这样可以把「这台仪器当时显示多少」与「附近居民在不同时间长期暴露于什么」分成两项可核查的问题,避免一个峰值既被拿来代表整天,也被拿来代表整片社区。

原文与时间戳 · 2
It's not like truck traffic
I Live 500 Feet From A Bitcoin Mine. My Life Is Hell. · ▶ 4:31–4:34
they've got a tremendous amount of noise pollution
I Live 500 Feet From A Bitcoin Mine. My Life Is Hell. · ▶ 12:42–12:50

一封匿名信能把调查推进到哪里?

讲者明确把信件当作一个例子,其中一位通信者描述了批评同行时担心的职业后果。保密使外人难以核实特定关系,来信者也不是随机抽到的。因此可以用这封信提出下一步可检验的问题,例如人们担心哪种后果、可用哪些独立记录追踪;要回答「有多普遍」,还需要与来信渠道无关的取样。

原文与时间戳 · 2
To honour the request of confidentiality
I was asked to keep this confidential · ▶ 0:44–0:51
I am reading this to you as an illustration
I was asked to keep this confidential · ▶ 3:52–3:59

同一起校园事件,病毒式传播的片段和后来采访分别能证明什么?

长访谈里的当事人评论、网上流传的片段、记者后来找到且愿意署名或匿名受访的学生,是来源不同的三份材料。一次转发能增加观看次数,却没有因此增加一位现场目击者。重建时间线时给每句话标出它出现于现场、事后访谈还是传播评论;把不能代表全部学生的视角保留为个人观察。

原文与时间戳 · 4
google evergreen state university
Joe Rogan Experience #1006 - Jordan Peterson & Bret Weinstein · ▶ 2:52–2:57
over time it exponentially increases the amount of people that are exposed to
Joe Rogan Experience #1006 - Jordan Peterson & Bret Weinstein · ▶ 96:58–97:05
This is the video, viewed by millions
Campus Argument Goes Viral As Evergreen State Is Caught In Racial Turmoil (HBO) · ▶ 0:47–0:49
they’ve been hesitant to publicly dissent.
Campus Argument Goes Viral As Evergreen State Is Caught In Racial Turmoil (HBO) · ▶ 5:06–5:09

从管道和运河,能读出决策者的动机吗?

RealLifeLore 从输气管道和运河重建乌克兰周边的物质依赖。地图能告诉我们一条路线被截断会影响哪里。Dalio 用八个指标画历史曲线,继而给出储备货币促进借贷的解释。两部视频都可以分两遍看:第一遍画线,列出它们表示什么;第二遍给每个因果判断标出所需的新记录。例如,证明某条运河进入了一次战争决策,要找的是当时谁掌握了什么信息、如何权衡其他方案。

原文与时间戳 · 4
pipelines were built across ukraine
Why Russia is Invading Ukraine · ▶ 11:24–11:26
peninsula from a canal built during the soviet era
Why Russia is Invading Ukraine · ▶ 17:32–17:33
I used eight metrics.
Principles for Dealing with the Changing World Order by Ray Dalio · ▶ 13:25–13:28
it becomes a reserve currency, which leads to even more borrowing.
Principles for Dealing with the Changing World Order by Ray Dalio · ▶ 15:32–15:37

回过头看得清楚,当时也看得见吗?

Jobs 后来能指出,自己在 Macintosh 上用了旁听书法课时学到的东西。他还专门强调,读大学时看不出这条线。把他的故事放到 Dalio 的历史曲线旁边,可以给回顾性解释加一道约束:暂时遮住结局,写下观察者在那个时刻能看见哪些信号。结局能帮助说明过去的一次选择怎样产生了价值;它也可能让人忘了其他当时同样合理的去向。

原文与时间戳 · 3
when we were designing the first Macintosh computer, it all came back to me.
Steve Jobs' 2005 Stanford Commencement Address · ▶ 4:17–4:22
it was impossible to connect the dots looking forward
Steve Jobs' 2005 Stanford Commencement Address · ▶ 4:57–5:02
I used eight metrics.
Principles for Dealing with the Changing World Order by Ray Dalio · ▶ 13:25–13:28

国家排名和工厂关门,记录的是同一笔损失吗?

FT 把边境手续带来的具体成本与其他国家的经济表现并列。Raspberry Pi 访谈把镜头缩到某一种零件:小厂一旦拿不到板,零售需求与工业客户保供会互相竞争。两个案例都可以先跟踪一笔订单,再决定可以外推多远。英国整体表现还叠加了能源和供应链冲击;树莓派零售端的缺货,也不是所有板子产量的计数。

原文与时间戳 · 4
supply chain effects of the Ukraine war
The Brexit effect: how leaving the EU hit the UK | FT Film · ▶ 11:17–11:20
draw a comparison between Britain
The Brexit effect: how leaving the EU hit the UK | FT Film · ▶ 11:26–11:29
more supply, but it's also got more industrial demand.
He started a computing REVOLUTION—then the shortage hit · ▶ 8:50–8:55
they go out of business
He started a computing REVOLUTION—then the shortage hit · ▶ 10:39–10:42

被接收的论文,和没有进入故事的投稿

Boghossian 与 Lindsay 讲了向期刊投稿的过程,谈话中一句「发表了」马上被纠正成「接收了」。先把投稿、送审、返修、接收、发表各画一格,就能看到这次更正改变了哪个事实。若想据此估计更广泛的同行评议质量,还需要完整的被拒样本、期刊选择方法和对照材料。轶事可以提示一个待调查的问题,成功讲出来的个案不能独自给出总体比例。

原文与时间戳 · 2
academic papers for the journals
Joe Rogan Experience #1191 - Peter Boghossian & James Lindsay · ▶ 2:37–2:38
got accepted
Joe Rogan Experience #1191 - Peter Boghossian & James Lindsay · ▶ 9:08–9:10

图例写了一个名字,哪些差异仍要另外查?

Grey 讨论美国原住民称谓时指出,不同部族和个人对统称各有取舍;讲联邦土地时,他又把同一图例颜色分成多种管理目标。两者涉及不同的分类问题。读者可在图例下补一行:要确定社群希望怎样被称呼,应听谁说?要确定一块地允许什么用途,应查哪个机构的文件?

原文与时间戳 · 4
more than 500 federally recognized tribes.
The US Has More Tribes Than the World Has Countries · ▶ 1:58–2:02
while preferred by some
The US Has More Tribes Than the World Has Countries · ▶ 3:06–3:09
Conservation, Recreation, Extraction.
A Third of America Belongs to America · ▶ 5:16–5:18
while it may be in a state, it is not of the state.
A Third of America Belongs to America · ▶ 8:23–8:29

读完再想一步

拿一份你最近用过的日志或图表。写下一个绝不会出现在里面的事件,再写下一个可能被登记到别人名下的事件。你需要哪份独立记录来区分它们?