站长把项目发到 HN 之后，网站被 AI bot 流量冲垮

1. 我在 HN 发帖后，我的网站被 AI BOT 流量淹没了

过去几年，我一直在做 sprinklz.io，这是一个可定制的信息流阅读器，除了其他功能外，还能把你订阅中某些话题的优先级降低。

一个月前，我把它发到了 HN 上，得到了大约一百个赞，真不错！看起来大家确实已经听腻了 LLM 相关的东西，也很高兴能把它从信息流里降级。这很好地演示了为什么这样的工具有用。一切看起来都挺顺利，结果……

我的网站变慢了。真真切切地、痛苦地慢。起初我把它归因于这篇突然爆火的帖子带来的额外关注。和任何一个开发者一样，我既为此庆祝，又发誓以后要把网站做得更稳健。然而奇怪的是，这种卡顿一直持续下去，甚至在我那篇帖子上线好几天之后还越来越严重，而我记录的指标怎么也解释不通。就好比——只有一小撮真正活跃、已登录的用户，服务器却呈现出高得离谱的负载。

于是我去看了下数据。我给每个端点都加了顶层日志（记录每个 url 路径被请求的次数），结果发现大多数请求根本连新闻文章都没有去拉取！参考一下，我的网站是用单页应用方式构建的，这意味着大多数请求甚至连渲染页面所需的 javascript 都没有运行（！）。

进一步深挖，我查了聚合的 NGINX 日志以找出滥用来源，你可能不会相信，结果发现我们将近一半的流量都来自某一个用户。我就不点名了，我查了一下，他们自称为一家服务，会监听网站的变化，然后把它们总结成一封 AI 垃圾新闻邮件。所谓“监听变化”，其实就是每分钟对网站请求很多很多次。已封禁。

其中几个是常见的已知爬虫，并且友好地标明了自己的身份。Claude bot、OpenAI bot、谷歌的搜索索引器以及一些其他的爬虫。我还顺带捕获了一些漏洞扫描器。至少有人在尝试访问 -/wp-content/phpinfo.php（我的网站并不是一个 WordPress 站点）。我添加了 robots.txt 来劝退大多数 AI 扫描器，同时允许 google-bot 和其他与搜索相关的流量，并请 claude 查询所有漏洞路径并模拟扫描器以确保我没有泄露任何信息。我还添加了合理的速率限制，等等，等等。

不过，最重要的教训是：我的网站确实该变得更快了。我希望我的网站能够胜任每天处理 xxx QPS 的真实流量，并且它应该具备比我之前更好的可扩展性，即使在高负载下也是如此。受这次经历的启发，我在过去几周里对底层基础设施做了不少改进，这些改进应该能为所有用户带来相当显著的性能提升。

附：如果你想看看的话，这个网站完全免费，在此访问

简而言之，它允许你定义自己的自定义规则，并且可以补充像来自 Hacker News 的积分总和这类数据。你可以用它来过滤主题，比如 AI 或 LLM，但这些主题完全是任意的，你也可以编写自己的自定义规则。

2. 使用 LLM 让 Emacs 浏览器重焕新生

每当我跟别人提起 Emacs 浏览器，他们的第一个反应往往是“等等，Emacs 居然有浏览器？！”，紧接着就是“嗯……也算是某种浏览器吧”。

Emacs 浏览器被设计用于在基于终端的纯文本环境中渲染网页，它的表现嘛……基本如你所料。它在仅有 HTML、内容较少的网站上表现尚可，但其它类型就不太行了。

问题在于 Emacs 本质上是一个纯文本编辑器，而现代互联网却远非基于纯文本。简单的 HTML，以简单的 1990 年代风格设计，能够清晰地转换为纯文本，但现代的 UI 设计——其视觉样式、厚重的 JavaScript，以及总体上的混乱——就不是这么回事了。

如果能找到一种穿越回过去的方法，Emacs 浏览器倒是个不错的选择……

好了，计划其实很简单。我们把现代网页的源代码改写成老派简单风格的 HTML，然后再把它送到 Emacs 浏览器里。我们只需要找到一种既简单又可扩展的方式来实现这一点。

嗯，你没忘记 LLM 对吧？能够以一种有望减少世界上垃圾内容的方式来使用它们，这感觉很不错。

基本上，这个计划就是写一个程序，它会：

在 firefox 中打开一个网页
加载页面并运行所有 JavaScript
让一个 LLM 生成一个可复用的程序，把当前页面内容转换为简化的文本
将生成的 JavaScript 注入到 firefox 浏览器中
抓取新的页面源码并转发到 emacs 浏览器。

是的，而且我一直在使用它，它跑得非常顺畅。坦白说，以前 eww（emacs web wowser）打不开的很多网页，我现在都可以正常浏览了。

下面是一些前后对比示例：

Clojure 的 mapcat 文档

图 1：（之前）使用 eww-readable

图 2：（之后）使用 simpleweb-simplify-page

它需要一点时间来生成初始的修改器程序，但一旦生成完成，运行速度相当快，在日常浏览中基本不需要额外的 LLM 调用。我自己一直在用，也乐在其中。

源代码在这里：https://github.com/sstraust/simpleweb

它是开源的，可以免费下载安装，等等等等。

2. 如果技术越来越好，为什么一切都越来越糟？

我想我们都知道答案：糟糕的激励机制。

技术让系统更有生产力，但实际上，它并没有让你个人变得更有生产力。那些让计算机更高效的工具，也让人们能更高效地掠夺你的时间、注意力和幸福感。我说的就是神经网络啊宝贝。

我最早在 2018 年注意到这个问题。更准确地说，我注意到我自己出了问题——每天花三个小时看《星际争霸 2》的回放，然后纳闷时间都去哪了，又是怎么又熬到凌晨两点还不睡的。

让我觉得有意思的是，这些问题直到 2010 年代末才开始出现。我至少从 2000 年代中期就有了能上网的电脑，没过多久又有了能上 Twitter、Reddit 之类网站的智能手机。这些年我可能确实变得没那么有韧性了，但我隐隐有种怀疑：是电脑变了。

我也说不清楚是怎么回事，但每次我与某个系统或网站互动时，总有种感觉：我在玩一场对抗游戏——我只是想正常过我的日子，也许读一篇关于电脑的小文章，而他们却在想办法让我在他们的网站上多停留哪怕一秒。

我还有这样一种感觉：随着软件工程从爱好者和发烧友的领域，逐渐变成越来越多由从金融或商业路线晋升的领导者主导的领域，我们曾经喜爱的这些重要工具，已经变得既无比乏味，又无比赚钱。

那我们能做些什么让它停下来呢？

走出户外，去摸点草。说真的，当世界变成这样时，我觉得最好的办法就是离开。这并不容易，尤其是当很多安排都在线上的时候，但如果你能安排好，我向你保证：和朋友在公园里下棋的一天，比你在网上度过的任何时间都要好。
走上街头。我认为抗议是有效的，应该有更多人参与。如果你愿意采取行动，这会给其他人树立榜样，让他们也可以采取行动。
不要再回到同一台机器上。我觉得我们每个人心里都有这种赌徒谬误——第一次没中奖，但有时候会中奖，于是我们又去试一次，但事情根本不是这么运作的。如果你在亏钱，教训就是别再往机器里投硬币了。

编者按：我一直在开发一个免费的可自定义的新闻阅读器，借此发泄一些我对现代互联网的不满。如果你不使用这个，我建议你去关注独立博客和网站，因为完全摆脱这些大平台要好得多。

编者按 2：我还没有把这篇最新的文章发布到我的 RSS XML 里，因为我还没准备好推送通知，会在未来几周内补上。

3. Deft 版本 0.2.0

现在 deft 0.2.0 已经发布了，我觉得正好可以聊聊我做了什么，以及为什么要做它。

Clojure 的「一切皆为 map」工具集的核心理念是，getter 和 setter 都是浪费时间。Clojure 的诞生，在很大程度上，正是为了回应 2008 年那种糟糕的 Java 工作流——那时候你得写 getter、setter 和 builder，背负着没人喜欢的冗余代码。

问题是，如果你想用内置的 defprotocol 和 defrecord 来表示数据，又想定义那些依赖于某个字段是否存在的函数，那么把这种信息编码到 protocol 里的唯一办法，就是给它加上 getter 和 setter 方法。太糟糕了。

你也可以用 protocol + malli schema 的组合来表示，但这样在使用时，你就得指定「这个函数接收的输入既实现了这个 protocol 又实现了这个 Malli schema」。同样糟糕。

我们想要的是这样一种东西：可以像 defprotocol 和 defrecord 那样使用，却默认就给我们所有的 getter、setter 和类 map 行为。关键洞察在于——既然字段只要存在于 record 中就能顺带获得那些好东西，那我们只需要给带必需键的 protocol 加上支持就够了。

Deft v0.2.0 新增了对类型和 protocol 定义中命名空间键的支持。也就是说，你现在可以这样做：

(defp Positioned :required-keys [::pos - ::Point])

以及

(deft Circle [::location/position radius])

(>Circle ::location/position [1 2] :radius 12)

这很有用，因为你现在可以定义依赖于这些字段存在的方法了，例如：

(defp Gremlin :extends [location/Positioned])

(defmethod game-object/get-hitbox ::Gremlin [self]
  (js/Matter.Bodies.rectangle
   (get-in self [::location/pos ::location/x])
   (get-in self [::location/pos ::location/y])
   15 15))

来看看吧！

https://github.com/sstraust/deft

4. 用多少 LLM 才算太多？

我上周读完了《On the Edge》，感觉很糟。不是说不尊重 Nate Silver——我非常尊重他。所以当他这本关于赌博与冒险的标志性著作沦为一团东拉西扯的随笔时，我既惊讶又失望。不过，作为一名完美主义者，我还是一路咬牙读到了最后一页，迎面撞上的是这样一句话：

「ChatGPT 在写作本书时给予了重大帮助，它既是创作灵感上的缪斯，帮我构思章节小标题、隐喻与类比，也帮我厘清那些很可能在它语料库中已有充分体现的技术主题。」

从技术上讲他是亲手写的这本书，从技术上讲他是用键盘一字一字敲上去的，但你也清楚——他真的「写」了吗？他是否把全部的心思、心智与存在都灌注到纸上的每一个字里？也许吧。这让我想起那部绝对经典之作 [Simple Made Easy](https://www.youtube.com/watch?v=SxdOUGdseq4)——那场演讲的核心，有一大半都建立在精确地定义语言、精确地说出你真正想表达的意思之上。当你把这项工作外包出去，你跳过的就不只是遣词造句，而是整个思考过程，你也因此错失了那种有条理的精神清明。

读 Nate 这本书时，你能真切地、明确地感受到那种像是从 Nate 心智结构里自然生长出来的语言，与那些事后才被补上的文字之间的差别。尤其最后几章，似乎缺少了那种能把所有内容凝聚在同一个愿景之下的精神凝练。我感到遗憾，因为它本可以是一部杰作——看得出来下了很大功夫，尤其是在访谈与数据采集上——却感觉就差那么最后一块洞察，没法把它彻底推到终点。

那么，多少 LLM 算太多 LLM？嗯，涉及到人与人的连接时，我的答案是接近零。我不想读你用 AI 生成的邮件，更不想读你用 AI 生成的文章。我宁愿直接看你的提示词，因为我在乎的是你，对，你个人的想法，我想明白你想表达什么。

别误会我，在合适的用途上我很喜欢 LLM。我很愿意把 LLM 用在医生诊断记录上，构建预测流感是否在扩散的模型。但说到写作，尤其是这种写作，它就显得完全多余，因为当你把自己的想法写下来时，我已经得到我想要的东西了，中间再插一道工序简直荒唐透顶。

作为扑克玩家和博弈论研究者，我相信 Nate 会赞同这句话——“要是你懒得写，那我也就懒得读”。你懂的，尽管我相信艺术可以与艺术家分离而独立存在，但在内心深处，这句话对我来说是成立的。并不是说一个人写出的相同文字与 LLM 生成的相同文字本质上有什么不同，而是它会让我脑子里那个小贝叶斯估计器有点失衡。

如果我认识的最聪明的人花了 10,000 小时去写作、编辑、打磨并出版一本 500 页的小说，这告诉我的不只是“我认识的最聪明的人对某个话题有一些想法”，更告诉我他们决定把他们宝贵的时间、他们宝贵的精力花在我身上，来向我传达这个想法。我读它所用的时间和它写它的时间成比例（而且通常更短），这让我感觉很好，因为这表明你珍视创造这次交流所投入的时间和精力。

不仅仅是一整坨 AI 生成的垃圾会撞上这道坎。就算只是在写作里掺一点点 LLM，你也会失去梳理自己思绪和建立人际连接的能力。

5. 用大约 350 行 Clojure 写一个国际象棋

我最近发布了 deft，一个用普通 map 来定义类型和协议的 Clojure 库，所以，为了试一试，我用 Clojure 写了一个国际象棋游戏。

国际象棋是任何类和接口系统的好试金石，因为它规模合理、自成一体（而且类的划分几乎天然明确），但同时它也带着真实应用的那种凌乱——你想写得简洁，却还是得处理边界情况，比如王车易位和吃过路兵。

我亲手画了所有的棋子，并用这段 reagent 小代码片段来显示画面，

[:h-box
 (doall (for [x (range (count tiles))]
                  ^{:key x}
                  [:v-box
                   (doall (for [y (reverse (range (count (nth tiles x))))]
                     ^{:key y}
                     [:p {:style {:background-color (cond
                                                      (= [x y] @currently-selected-location)  "yellow"
                                                      (= (mod (+ x y) 2) 1) "white"
                                                      :else "black")
                                  :height "4rem"
                                  :width "4rem"
                                  }
                           :on-click (fn [] (move-selected-piece
                                                    game-board-atom
                                                    currently-selected-location
                                                    [x y]))}

                      (if (not (nil? (get-in tiles [x y])))
                        [:img {:src (image-path (::piece-type (get-in tiles [x y])))
                               :style {:height "100%"
                                       :width "100%"
                                       :filter (when (= (::color (get-in tiles [x y])) ::black)
                                                 "invert()")}}]
                        "")]))]))]

但真正有意思的，是如何表示棋盘状态以及上面的所有棋子。

我基本上做了最直白的事，先定义了表示棋子的类型，再把棋盘定义为一个由棋子组成的 8x8 网格

(deft ChessPiece [color piece-type])


(deft ChessBoard [tiles - [:vector [:vector [:or :nil ChessPiece]]]
                  curr-move - [:enum ::white ::black]
                  move-history - [:vector ChessMove]])

然后对于棋子行为，我定义了协议，规定棋子如何显示、以及它能如何移动。

(defp PieceType
  (image-path [this]))

(defp MovablePiece
  (is-valid-move-for-piece? [this game-board start-location end-location]))

以骑士为例：

(deft Knight []
  PieceType
  (image-path [this] (image-asset "knight_white.png"))
  MovablePiece
  (is-valid-move-for-piece?
   [this game-board start-location end-location]
   (and
    (let [direction (map abs (map - end-location start-location))]
      (or (and (= (first direction) 1) (= (second direction) 2))
          (and (= (first direction) 2) (= (second direction) 1))))
    (is-valid-end-location? game-board start-location end-location))))

is-valid-move-for-piece? 返回一个真假值，表示“我的棋子能不能走到这一格”。

对大多数棋子来说这样很好用，但对更复杂的走法（比如王车易位，要同时移动多个棋子）就不够了。要处理这种情况，我们再用一个 multimethod

(defmulti special-move-for-piece (fn [x & args] (:type x)))

它只针对特殊走法定义，并返回修改后的棋盘实际状态。

所以对于兵，它看起来大概是这样：

(defmethod special-move-for-piece ::Pawn
  [this game-board start-location end-location]
  (let [curr-piece-color (::color (get-in (::tiles game-board) start-location))
        move-direction (into [] (map - end-location start-location))
        prev-move (last (::move-history game-board))]
    (cond
      ;; en passant
      (and (= (second move-direction)
              (if (= curr-piece-color ::white)
                1 -1))
           (= (abs (first move-direction)) 1)
         (= (:type (::piece-type (::piece prev-move))) ::Pawn)
         (= (apply max (map abs (map - (::end-location prev-move) (::start-location prev-move)))) 2)
         (= end-location
            (into []
                  (map +
                       (map (fn [x] (/ x 2 )) (map - (::end-location prev-move) (::start-location prev-move)))
                       (::start-location prev-move)))))
      (-> game-board
          (update ::tiles assoc-in start-location nil)
          (update ::tiles assoc-in end-location
                  (get-in (::tiles game-board) start-location))
          (update ::tiles assoc-in (::end-location prev-move)
                  nil))

      (and (is-valid-move-for-piece? this game-board start-location end-location)
           (or (= (second end-location) 0)
               (= (second end-location) 7)))
      (-> game-board
          (update ::tiles assoc-in start-location nil)
          (update ::tiles assoc-in end-location
                  (assoc (get-in (::tiles game-board) start-location)
                         ::piece-type (>Queen))))
      :else nil)))

我不会在这篇文章里逐个复制粘贴每个棋子的走法集合,但为所有不同的棋子类型和特殊情况定义相关方法相当直接,我对最终的代码相当满意。当你读这些代码的时候,看起来真的像是有人把国际象棋的所有规则写了下来,这让我感觉很好,也让我觉得我们的做法是井井有条的。

但是现在,某种程度上最重要的问题是(对我来说):

这个库真的有帮助吗?deft 让我们的生活变得更好了吗?

我觉得确实有,理由是:

deft 同时给了我们 protocols/records 和 multimethods/plain-maps 的特性,所以我们不需要决定哪种工具适合我们的用例。这减轻了很多认知负担,因为我们可以用一种工具,并且相信它能把一切做得不错。
特别地,能够为我们的协议定义 :default 方法分发很好,而且能够从 repl 重新加载协议方法的定义而不必重新实例化整个棋盘(并清除当前的局面)真的非常好
能够强制要求在特定类型上必须定义某些方法,这真的很好;以及能够轻松添加 Malli schema,这也真的很好

总的来说,我对最终的结果相当满意,我认为对我们这个库的检验是成功的。

你可以在这里找到

这个游戏的完整代码:https://github.com/sstraust/deft/blob/master/examples/src/examples/chess.cljs

这个库的代码:https://github.com/sstraust/deft

以及详细的视频讲解在这里:https://www.youtube.com/watch?v=lrofmFI1UJY

6. 宣布 Deft:一个用于 Clojure 和 ClojureScript 的类和接口系统

今天我要发布 deft 了!这是一个用普通 map 和 Malli schema 来取代 Clojure 现有类系统的库。

我花了大约 6 个月才真正把它想透彻,并把设计做到恰到好处,所以能最终发布感觉很好。

这个库:https://github.com/sstraust/deft 视频教程:https://www.youtube.com/watch?v=dlW6YzwUZ-M Hackernews 讨论:https://news.ycombinator.com/item?id=46590958 (请给仓库加 star 吧!)

简介: 这个库的诞生源于我们对 Clojure 现有类型生态的不满。defrecord/defprotocol,以及 plain maps+multimethods,它们目前的使用方式,使得 Clojure 程序从根本上难以组织和重构。

问题在于:

Clojure record 使用单冒号(未命名空间化)的关键字访问来查找 record 字段
Clojure record 不允许你定义抽象类,或依赖于复杂分发行为的东西
Clojure map 非常难以确定并保持稳定。即使有 Malli 这样的库,也很难声明能够说明"我可以用这个东西做什么"的接口
除非你严格地针对每个用例都定义 Malli schema,否则在解构过程中很容易出现脱节,因为漏掉的字段查找会返回 nil

我们的目标是提供一个仍然用起来愉悦的东西,但在需要的时候可以真正把行为钉死(并且仍然符合人体工程学、有趣)。

另一个目标是让这个东西可靠且稳定,并且(希望)容易在现有代码库中采用。来看看吧!

8. 如何在 Wordle 中获胜

这周我玩了第一局 Wordle，4 步就猜出来了！

好吧，我确实有一点点运气，但这并不全是运气。在玩的过程中我有时间思考一些策略，也看到了一些我认为别人可能没注意到的陷阱。

8.1. 不要重复猜测已经为绿色的字母

除非你是想走最后一步猜测来赢得游戏，否则你不应该重复猜测同一位置上的同一个字母。你已经试过了，再来一次不会给你新的信息。

你的"猜测"一定会是错的，但当你真正去解这道谜题时，你又排除（或找到！）了一个可能的字母。来看一个例子：

假设你有 B - A - _ - E - S，全部为绿。你只需要猜出最后一个字母，但剩下的选项还很多。在你没有猜过这个位置的所有字母中，D、K、N、R、S 都可以。你可以逐个字母去猜，但你很可能不会一次就猜对，最坏情况要花五步。相反，如果你猜 D - R - I - N - K，一个你明知是错的单词，你相当于一次"猜"完了所有字母，这样保证最多 2 步就能解出来（如果没有黄色就直接猜 S）。

8.2. 猜测新字母时，把它们放在常出现的位置

试一下，把 Wordle 当成二分搜索来想。你的目标是通过每次猜测获得尽可能多的信息。如果你的猜测能把问题的解空间一分为二（也就是说，如果显示绿色你就排除了半数可能的单词，显示黄色也排除了半数单词），那么你（平均）每次获得的信息量都是最大的。

现在——你只能在某一个位置上猜一个字母，要么它就在你选的那个特定位置，要么就在另外四个位置中的某一个。如果你想让这两种结果发生的概率大致相等（或尽可能接近相等），那么你选的那个位置最好确实非常可能命中，才能弥补另外 4 个槽位的不足。

同样是去了解字母 'Y'，通过 R - A - I - N - Y 这种猜测，你（平均）学到的信息远远多于 Y - E - A - R - N。

8.3. 先猜常用字母

基本逻辑同上。答案要么是你猜的 5 个特定字母之一，要么是其余 21 个字母之一。你要选那 5 个字母，让问题的解空间尽可能接近对半分。

更进一步，一旦你已经猜对了几行的字母，在已知字母的前提下，你应当去猜那些最常见／最能把剩余解空间均匀分开的字母。Q 作为首猜并不好，但如果你已经确定了一个 U，那情况可能就不一样了……

8.4. 每一步都过一遍可能的单词列表

记住，你的目标是让每一次猜测都把解空间"对半分"。如果你对当前解空间里的单词有一个大致的感知／随机采样，就更容易挑出能最精确地平分这个列表的字母／位置。

9. 《习惯的力量》（书评第一部分）

Charles Duhigg 曾是《纽约时报》的记者，这一点你能从他的文字中感受到。这是一份现代科学报告，写法耸动、励志，但常常具有高度误导性。

这本书应有尽有：从电视迷到超级明星的自我改造、星巴克的成功、小马队的战绩，一切都简单地用习惯的力量来解释。当然，习惯在每一件事中都只扮演了很小的角色。当然，每一场橄榄球比赛都多少要靠点运气（甚至还要看风和天气）。但是——有人养成了好习惯，后来成功了——砰——因果关系——搞定——问题解决。

不过，撇开那些离谱的结论不谈，确实有一些真正的（而且是最近的！）科学研究正在进行，我们对习惯的认识也在不断变化，这既有趣又有力量。对我而言，关键要点是：

习惯绕过了我们正常的推理过程（并且不储存在与计划和短期记忆相同的大脑区域）习惯可以被建模为触发——行动——奖励的序列。如果你缺少其中任何一环，你就无法触发（或学会）这个习惯循环即使我们不再关注某个习惯，它也可以潜伏很长时间。不过别误会，这本书的目标是带来改变。你能做些什么来改变你的人生，就像佩顿·曼宁（Peyton Manning）那样？同样，这里的科学结论是喜忧参半的，但作者的观点似乎是：

习惯很难戒除，但如果你把它们拆解成触发——行动——奖励的循环，就会容易一些。如果你能换掉行动环节，同时保留相同的触发和奖励，就能"劫持"你的坏习惯，从而学会好的行为。如果你想要做出颠覆性的改变，最简单的办法是把全部注意力集中在一个习惯上，其他的习惯很可能随之而来。习惯可以是件好事。思考是困难的，决定做什么需要付出努力。如果某件事（比如决定晚餐做什么）需要耗费心力，你就更不可能去做它，因为这种心力付出和实际做饭一样累。试着养成习惯，并提前想好你要做什么，这样到了关键时刻，就不需要费力了。

10. 现代货币理论与你

和任何政治议程一样，现代货币理论并没有被精确定义，人们常常根据自己的目标来裁剪它的定义。据我所知，它大致是这样的：

如果你是一个控制本国货币的国家，那么举债是可以的，只要不引发通货膨胀，你想印多少钞票都可以。

有些人认为，凭借现代货币理论的力量，我们有资金去支撑所有最疯狂的希望和梦想，只要有人有勇气启动印钞机。免费大学、贷款偿还、免费住房，我们可以一键支付所有费用，而且不会产生负面的长期后果。只要我们不引发通货膨胀，就真的没什么可担心的。

10.1. 那个大前提

问题是，那可是一个相当大的"如果"。如果你从一个错误的假设出发，你可以"证明"任何你想要的东西。如果我们能无限印钞，又不会引发通胀，那可真是太棒了，但这种事真的可能吗？我们怎么会走到这一步，为什么有人相信我们能做到？

理论上，经济学建立在供求关系之上。如果每个人都想要苹果，但市面上没有苹果可卖，苹果的价格就会上涨。如果没人想要苹果，但每个公民都囤积了 100 个苹果，那么苹果的价格（以及苹果的隐含价值）就会下降。

同样的道理,如果美国每个人都神奇地拥有了 1200 万美元,他们就会相互竞争,拿着这些钱去抢购同一批数量有限的商品。很多人想要三居室的房子,但房子并不会神奇地变多。仍然是同一拨人在相互竞争,抢着买同一栋房子。想要在竞价中赢过一个持有 1200 万美元、真心想要某样东西的人,你同样也得花掉几百万美元。

从根本上说,人们想要钱,是因为可以用它来购买商品和服务。从全球范围来看,如果每个政府都加印货币,但全球可用的商品和服务总量并没有相应增加,那么实际上并没有提高每个人的真实购买力,因为大家仍然在争抢同样有限的商品和服务。只有当生产出来的商品和服务数量也增加时,真实财富才会增加。

10.2. 我们是如何走到这一步的

所以,如果美国明天就开始大规模印钞、大规模举债,我们会预期所有东西的价格飞涨。我们会预期政府债券变得远不如以前值钱,对吧?对吧?

结果发现,在 2008 年金融危机期间,美国确实印了很多钱,也确实背负了大量额外债务,人人都预期通胀会发生,但不知怎么,通胀就是……没有发生。市场很难讲清楚道理,也没有人真正明白为什么会出现这种情况,但人们提出了几种理论:

美国被视为一个值得信赖的机构,人们需要一个安全的地方存放自己的钱。国际金融体系对美国美元的投资程度已经深到美国可以为所欲为。刺激法案带动了如此多的经济增长,以至于抵消了通胀。

10.3. 美国作为值得信赖的机构

好吧,以前我们也背负了大量债务,结果也没出什么事。我们印了数量惊人到离谱的钞票,人们依然信任我们是一个稳定、可靠的机构。所有人都相信美国政府会偿还债务,而且偿还的美元会保持它的价值。

即便这场危机是由美国经济引发的,美国看起来仍然是一个安全的资金存放地,因为过去美国一直表现得如此负责任、如此可靠。

那么——如果我们突然开始不负责任,比如疯狂地累积巨额债务,再印钱来偿还这些债务,人们就会一直借给我们钱,对吧?对吧??如果我们什么都不生产,就能一直印钱,人们也就会一直把东西给我们,对吧?

有种说法认为,因为我们借债是用美元计价的,而美元又是事实上的世界货币,所以我们永远不会有麻烦,因为我们可以直接印更多的钱来填补亏空。但是,如果你换到的实物商品和服务上,美国并没有给出任何真正有意义的东西,你认为人们还会一直借给我们吗?如果美国政府买走了你的所有东西,却只给你一纸凭证,没有任何实物回报,你认为人们还会继续把美元当作默认货币吗?

从技术上说,我们可以通过疯狂印钞来还清债务,但从投资的角度看,这本质上和违约没什么区别。我们做出了承诺,但在商品和服务层面,我们并没有提供任何真正的价值来兑现这个承诺。

拖欠贷款非常危险,因为它会毁掉你的信用。一旦你破产,再想获得贷款就难上加难,因为人们不再信任你了。

同样的道理,即使美国在技术上并未真正拖欠债务,但如果它靠印更多钱来掩盖,本质上也会带来同样的后果,因为人们不会再信任我们了。

我们可以"兑现"当下的信任和影响力,但一旦破坏了这份信任,就再也无法像以前那样依赖它了。

10.4. 劫富济贫

这种论点大致是这样说的:

富人有钱,穷人没钱。所以,如果我们额外印一些钱,发给穷人,就会压低富人手中钱的价值,从而把富人的购买力实质性地转移并重新分配给穷人。

它以一种干净利落的方式达成了人们本就追求的目标。

问题是——富人持有的财富并不以美元形式存在。那些真正的富豪深知美元并不一定保值,他们持有房产、企业股票,甚至珍贵的艺术品。美元一贬值,对很多富人影响并不大,但却会伤害其他人。

真正受到伤害的是那些受 15 美元(或更低)固定最低工资约束的人,以及那些不太懂投资、把所有钱都放在储蓄账户里的中下阶层。它确实能帮到负债的人,也可能带来其他一些正面效果,但对许多人来说,它更像是一种破坏稳定的因素,而非真正的帮助。

10.5. 刺激与经济增长

好吧,这里有一个真正站得住脚的论点。其思路是——印钱没关系,因为我们可以创造更多商品和服务。如果我们把钱花在刀刃上——比如修建桥梁、改善宽带——就能为原本失业的人创造就业。更重要的是,这些新的基础设施和支出将进一步刺激经济增长。

如果我们每花 1 美元,就能产生 $N 美元的经济活动(这怎么可能可靠地算出来?),那我们印的就不仅仅是钱,而是价值。

这是我喜欢的一种论点,因为它在逻辑上自洽。如果我们生产出更多商品和服务,那么即使流通中的美元更多,也有更多东西可以用这些钱去购买。如果它真的能带来那么多增长,这招就管用。

这里的问题在于它过于乐观。你的政府真有那么聪明、那么有组织,能创造出价值 10 万亿美元的东西吗?也许吧。我们有过辉煌的时刻,我希望我们能让它发挥作用,但大多数时候事情并不会这样发展。

11. 完美的相机

每年临近假期,我都会奢侈一把,给自己买一件真正的好东西。疫情期间被困在家,没什么可花钱的地方,但我玩风景摄影玩得不亦乐乎。我们还是没去酒吧和餐厅,但可以开车到处转转,也一直在探索新泽西州内外的公园。

今年我下定决心要买一台新相机。要比手头那台再好一点的那种,说不定还能用来打印照片和大尺寸海报。

我是在 2018 年 12 月买了第一台相机。我记得当时花了很长时间去挑选，最后选定了 Nikon Z6。我其实很想要一台 D850，但加上镜头价格太高，根本不在考虑范围内。我也看过 D750，尽管我很喜欢光学取景器的概念，但无反似乎是未来的趋势，而且一台老旧的单反会让我变成"恐龙"。我也看过其他品牌，但 Canon 当时还没有很棒的无反选择，而我对 Sony 这家公司也没什么好感。

时光飞逝，两年后，我又站在了同样的位置，花很长时间去挑选相机。

其间，我也有机会试用了一台老式的单反（D610），我彻底爱上了它。画质更好（至少直出就更好）。光学取景器用起来令人舒心。更重要的是，整台相机的设计都经过深思熟虑，做工精良。每个按键都在恰到好处的位置，而且在我需要的时候总能找到对应的按键。像白平衡这种在 Z6 上要埋藏在菜单深处的功能，在这里只需一次点击就能调出。它们甚至还有一颗非常好用的辅助对焦灯，基本上解决了我在弱光环境下对焦的所有问题。感觉无论我需要什么，伸手就能在 D610 上找到答案。

这次疫情期间我花钱花得很少，所以我决定是时候入手一台真正好的相机了。所有的品牌和类型都纳入了考虑范围。我决定要买最好的相机，这样就不用再想着升级的事了，至少很长一段时间内都不用。

当然，我很快发现事情没那么简单。就算你有全世界的钱，不同的相机也有不同的取舍。画质最好的那些往往又大又笨重。有些相机动态范围技术上偏低，但成像却漂亮自然。无反相机拥有人脸和眼部识别，但又缺少那种出色的光学和机械质感。

事实上，每台相机都带来不同的体验，不同相机在不同的场景下才是"最好的"。如果拍自拍，你想要一块可翻转的屏幕。如果带相机去聚餐，你想要一台时尚轻便的。如果拍摄活动，你需要最可靠的机身和最棒的对焦追踪。

我当然想要一台小巧的相机，能带去和朋友们的桌游之夜；但我也想要一台（可能又大又重的）能拍出顶级画质的风光机。我很快就意识到，在不同的场景下我需要不同的配置，没有哪台相机能完美满足我所有的需求。

综合所有因素，我把通常拍摄的场景大致归为三类：

摄影散步。我专门为了拍摄优美的风景照而出门。我希望获得愉快的拍摄体验和极致的画质。轻便固然好，但并不十分在意相机是否更重，或是对焦是否更慢。亲友合影。这是和家人、朋友出去玩迷你高尔夫时我会带在身边的相机。我不需要极致的画质，但希望它不引人注目，并且非常可靠。我在意的是在各种场景下都能拍到一张可用的照片。如果在埃菲尔铁塔前的全家福拍糊了，我们不会再去一次，也不会花两个小时重新调整构图去追求完美出片。这里也很重要的一点是：让别人来帮忙拍，成片也应该好看。人像/自拍——介于上述两者之间。你有更多时间来布置环境、带上合适的镜头，但仍希望有相当可靠的脸部/眼部识别。 4.1 候选机型

4.1.1 佳能

佳能生产专业的单反和无反相机。

他们最新的专业无反相机 R5 是一款实力均衡的全能机型——据我所知，它在几乎所有场景下都表现出色。它的速度足以拍摄体育和野生动物，面部/眼部识别非常出色，方便用来拍摄亲友合影，镜头素质也非常好（尽管昂贵），可实现极致的画质。一个很大的加分项是它配备了翻转屏——这对于专业摄影师来说也许没那么有用，但在拍摄全家福时却非常实用，能把每个人都收进画面。

R5 唯一的真正缺点是它非常昂贵，以及在拍摄视频时会过热（虽然我不拍 8K 视频，但拥有一台可能过热的相机还是会让我有些紧张）。其中一个难题在于第三方 RF 镜头不多，所以你只能用佳能自己非常好、但也非常昂贵的镜头。佳能的色彩以优秀著称，但我还听说 R5/R6 相比早期的佳能单反出片风格更平淡，这意味着要在后期花更多功夫才能得到非常好的图像。

他们也生产一款专业单反相机，比较受欢迎，直出 (SOOC) 效果扎实，不过动态范围较差，相比尼康的对应机型也没有什么实质性的性能提升。

佳能还生产 RP，它是 R5 的更小巧、性能更低的版本

4.1.2 索尼

索尼的 A7R4 在所有无反相机中拥有最高的分辨率和最佳的面部追踪能力。它被认为是一款技术上非常精密的相机，但在一些易用性方面被认为有所欠缺，比如 SOOC JPEG 和人体工学。现在佳能在面部追踪和技术规格上基本追了上来，我觉得 A7R4 变得更难推荐了，尤其是因为我本身也不太喜欢索尼。不过它要便宜得多，特别是把镜头的成本考虑进去时。

索尼还有 A7C，搭配蔡司 35 f/2.8 镜头，是你能买到的最小巧的组合之一，画质却非常好。还有 Rx1R，分辨率更高，但缺少翻转屏和机身防抖 (IBIS)；以及 RX100，体积更小，但 1 英寸传感器也意味着缺少一些你对专用相机的画质预期。

4.1.3 尼康

我有一台 Z6，它的面部/眼睛对焦简直糟糕透顶。糟糕到我完全无法信赖它。我拍过一些看起来很糟糕的照片，后来才意识到是相机脱焦了。我甚至无法使用单点自动对焦，因为摇杆移动对焦点太慢了。

我还注意到，与我用 D610 拍的照片相比，它的色彩平衡略有偏差，特别是在使用带转接环的 F 卡口镜头时，会让同一支镜头拍出的同一张照片显得没那么讨喜。也许可以通过后期调回来，但我觉得 D610 拍的照片好看得多，也更容易处理。

也许尼康在 Mark II 中修复了对焦问题。也许用上最新最好的 Z 卡口镜头并做出调整后，颜色就能正常了。也许不用转接环对焦会更好。但到了这个份上，我就是不再信任他们了。我确实很喜欢 Z 卡口镜头，只是希望它背后能有一台更好的相机。

相比之下，我使用 D610 的体验非常棒。这台十年前的相机，性能依然能跟上今天的标准。按钮布局和人体工学完全符合我的需求。它算不上什么全新理念，而是一款经过千锤百炼、坚固耐用的专业相机。它让我觉得我所需要的一切都被周全考虑并提供了。最重要的是：直出的画质近乎完美。

D850 作为尼康最新的 DSLR，是 D610 的下一代产品（严格来说是好几代）。它保留了大部分相同的设计和做工，使用同样的 F 卡口镜头，并且在我看下来，呈现图像的方式依然非常讨喜。它其实比设计精良的 D610 少了一些功能（比如更精细的白平衡调整和内置闪光灯），但为了这些小小的牺牲，你得到的是分辨率高得多的传感器、更好的动态范围、更好的对焦点、推拉画面时更好的色彩保留等等。

与 Z6 不同的是，我信任 D610。我知道如果买一台基本相同配置的 D850，我会得到基本相同的出色照片（尽管有望再好那么一点点）。

我一直等着/盼着尼康推出 D850 的继任者，但根据较新的 D780 以及其他近期推出的相机，我猜测：

他们可能不会大幅提升分辨率。至少，发布一台分辨率明显高于他们最新无反产品的 DSLR 是不合理的。他们很可能把无反相机上同样糟糕的面部和眼睛对焦照搬到实时取景模式中。其实我更喜欢 D610 上的对比度检测对焦。它可能很慢，但至少可靠。加入相位检测也意味着传感器上多了些可能影响画质的元件。他们可能会加入机身防抖，我确实很想要机身防抖，但对我来说不是特别大的事，因为我的很多镜头本身就有防抖。 4.1.4 徕卡

我觉得徕卡就是那种品牌——人们买来试用，就是为了弄清买一台徕卡到底值不值。开法拉利真的会顺畅很多吗？除非亲自试试，否则你永远不会知道。

玩笑归玩笑，我想到了几个人们可能想买的真实理由：

它们渲染效果不错。这一点很难量化，尤其是当技术图表并不总是支持这些说法时，但有些人喜欢徕卡图像的整体渲染/美学观感。它们外观好看。好吧，对于摄影师来说，这似乎不应该重要，但它确实重要，因为它影响你是否愿意随身携带它。如果你拿着 D5 出现在皇室婚礼上，你会看起来像狗仔队，但徕卡则时尚而优雅。你甚至可以把它当作穿搭的一部分佩戴，只是为了看起来有点文艺范儿。它们体积小巧。你见过巨大的尼康 Noct 0.95 镜头吗？你把它和徕卡 noctilux 比较过吗？Noct 稍好一点，但你能把通常很大的全画幅定焦镜头塞进一个非常小的机身里。这意味着你可以很随意地携带，在日常环境中也更方便。它们是旁轴相机。如果你想要旁轴相机的体验，徕卡基本上是你唯一的选择。我个人对此不太感兴趣，但如果你需要一台手动对焦旁轴，徕卡就是你的归宿。我喜欢徕卡相机非常小巧。对我来说，能随身携带、不用时也不碍事很重要。不背一台巨大的单反出去吃饭会让人更轻松。

问题是，徕卡的自动对焦并不出色（即便有的话），而且在"非专业用户的易用性"方面完全缺失。它们是一台很棒的轻便度假相机，适合趣味摄影散步，但如果我去一个美好的家庭度假，我想要一张美好的全家福照片，而徕卡相机就是无法胜任。（Q2 算是勉强可以）。

4.1.5 Fujifilm

他们的 APS-C 相机系列很扎实（而且外观漂亮！），但我觉得它们的体积不够小，无法与 A7C 这样的机身在画质上的差异相抵。他们也生产 X100V，其旁轴风格很吸引人，但有更小、更轻、画质更好的相机可选。

GFX 系统是一场冒险，因为它有点不便，但无可否认地，它是比任何其他实用系统表现更好的机型。当其他厂商推出 1 亿像素时，图像可能看起来颗粒感更重，但用更大的图像传感器做到这一点，你就真的多获得了 50% 的空间。不仅如此，我能看出刚刚发布的 GFX 100S 是与众不同的。它不仅修正了初代 GFX 100 的人体工学，不仅把性能塞进了一个小巧的、单反大小的机身，而且它的价格只有 GFX 100 的 60%，绝对算得上物超所值。

我想入手一台，因为我知道这种性能突破的相机能够长期保值。就像他们造出了一台比自家绝对旗舰中画幅相机还要好得多的相机，而且定价不仅低于自家旗舰，也低于全画幅竞争对手的旗舰。D850 发布时是 4 年前的一件大事，但这是一件当下正在发生的大事，并且会在未来很长一段时间内继续保持重要意义。

我还不能完全为富士的色彩科学背书，因为我还没有像用我的尼康那样把它们彻底试用一遍，但我听说过他们色彩科学的诸多好评，他们的胶片模拟看起来也真的很吸引人，至少从我浏览过的在线测试照片来看是这样。

GFX 系统最大的问题是可用的镜头并不多。他们已有的镜头品质惊艳，有些甚至被认为显著优于对应的单反版本，但你必须使用富士提供的镜头，如果你需要特定规格的镜头，比如 14-24 2.8，那就只能自认倒霉了。这是一个你真正在购买机身的同时也在购买镜头的系统，所以不能只看相机本身，还要看它搭配你实际要用的镜头时的表现。

如果我入手 GFX 系统，我想要 32-64 f/4 和 45 f/2.8 这两款镜头，它们都是性能很高的镜头，不过我确实希望他们能有一款真正的 24-70 f/2.8 等效镜头，以及更多焦段选择。

4.2 决策时刻

我知道，由于疫情，我在几个月到一年的时间里基本无法参加正常的社交活动，所以随身带一台又大又重的单反也无所谓。我也不想成为那种为了不同场景而口袋里装着一堆不同相机的人。这里的关键是用手机自拍也没问题。你可以拥有一台好相机，但并不是每张照片都需要顶级的画质。如果你只是想留个美好的晚餐做煎饼的回忆，比起画质，更重要的是不要打断你试图捕捉的瞬间。也许如果我买一堆相机，我会买一台真正小巧的，但目前我并不需要，手机对于非摄影类照片来说已经是很可靠的替代方案。我也有些警惕定焦相机，因为一旦裁切画面，画质上的所有优势都会丧失。

我本想亲自把每台相机都试一遍，但同样由于疫情，我没法去相机店。我可以在线上看一些样张，但总是很难判断有多少是经过后期的，又有多少是来自场景光线和拍摄者的技术。

基于以上考虑，我只关注大型、高性能的可换镜头相机，缩小了搜索范围。Nikon Z 和 A7R 系列被排除在外，因为我认为它们已被 Canon R5 超越。5D Mark 4 也被排除，因为被认为已被 D850 超越。这样剩下的选择就是：Canon R5、Nikon D850 和 Fujifilm GFX 100s。

R5 拥有出色的 RF 镜头群，是一款非常全面的相机，配有翻转屏幕，并具备非常好用的人脸和眼部识别功能。此外，佳能一直以制造好用、易于上手的相机而闻名。问题在于，在镜头方面，腾龙的镜头要便宜得多，而且性能也非常出色（特别是 35/1.4！），而 D610 的实时取景人脸识别实际上非常准确且可用，只是对快速移动的主体来说太慢了。R5 直出的色彩也被认为比它的单反版本略显平淡。R5 上我真正最怀念的其实是翻转屏，不过我觉得如果真的很需要拍合影，我们可以请别人帮我们拍。相比 D850，它也缺少了大量物理按键和光学取景器的体验。

GFX 100S 是我考虑最多的一台相机。搭配 45 f/2.8 或 110 f/2，它能拍出任何实用相机中最好的画质，知道自己拥有最好的画质会让人感觉很好。如果你拍的照片不如网上找到的样张好看，知道不是因为选错了相机会让人安心。在编程中，我们把调试看作"找到破坏你软件的错误源头"，而拥有最好的设备能让你的系统更简单，因为你少了一个需要排查的错误来源。你不必再回到同一个地方用另一台相机重新尝试，因为你知道那不是相机的错。它让你不再纠结于器材，转而专注于那些让一张照片变得优秀的其他因素。

虽然 D850 与 R5 相比仍有竞争力，但它也是一台已有 4 年历史的相机。以技术发展的速度来看，一台相机在淘汰前能用 10 年已经算幸运了，而它已经走过了生命周期的中段。相比之下，GFX 100s 不仅是新的，而且是开创性的，是一项似乎可以长久立足的、巨大的技术飞跃。你购买的是一个庞大、新颖且快速成长的系统，而不是一个正在慢慢被淘汰的系统。它也已发展到第四代，并且真正获得了对细节的关注，从而被打磨成一款成熟的产品。它缺少竖拍手柄，但大多数情况下它已经非常完善，并没有什么重大缺陷或缺失。

在 D850 和 GFX 100s 之间，抉择是最困难的，因为 GFX 100s 实在是一个非常诱人的选择。

但最终……

我选择了 D850。对我来说最重要的是我喜欢使用 D610，而且不仅如此，我信任它。我不知道 GFX 100s 是否好玩，也不知道它拍出的图像是否是我喜欢的那种风格，但我知道 D850 一定会，因为它和 D610 完全一样。我记得两年前我买 Z6 之前，一些评测者告诉我它成像不错而且自动对焦可用，但在我看来，他们错得不能再错了。

GFX 系统可用镜头的缺乏对我来说也是一个很大的问题。如果你用 35mm 拍摄，然后再裁剪到 85mm，你就无法获得中画幅传感器的任何画质优势。如果你使用 GFX 系统，你只能获得可怜的 2 倍变焦范围，这意味着除非构图恰好与你手中仅有的那支镜头完全吻合，否则你拍不到想要的画面。

富士的一些镜头确实很棒，我不是说它们不好，但并非所有镜头都是完美的，如果我需要的那支镜头不存在，我就没办法获得它。45mm、110mm 和 32-64mm 被大多数评测者（以及技术图表）认为近乎完美，但其他焦段的情况并非如此。要是我想要一支性能超群的 35-150 呢？要是我真的需要我那支 24-70 f/2.8 的长焦端呢？你不仅只能购买昂贵的富士玻璃镜片，还无法选择最适合自己使用场景的镜头。

即便是相同的焦段，选择也很重要。富士只有一支真正的饼干镜头，但它并不是我想要的那种用途。我想要一支 28mm 饼干镜头，以及一支更大、性能更高的 50mm，但富士并不提供这些。不仅如此，有些镜头的成像风格可能不是我喜欢的。腾龙和适马都推出了优秀的 35 1.4，但我更喜欢腾龙的成像风格，我不想为了转入富士系统而放弃这种风格。

我在开头说过，我愿意换系统（如果当时有 Z6 可买，我肯定会换），但不必为了入手新镜头而折价出售旧镜头，这点很好；更重要的是，我已经做过功课，知道自己喜欢哪些镜头，哪些镜头适合我需要拍摄的题材。

4.3 终章

我的 D850 在几周前送到我手中，正好是我想要的样子。它和 D610 几乎一样，只是对焦更利落，分辨率也更高（成像看起来也更锐利一些）。作为一项工程设计，它考虑周全、设计精良。机身结实得像坦克，看起来能扛住任何考验。我真正怀念 D610 的只有：自动白平衡里更细化的色温调整、回放模式下独立的润饰按钮，以及内置闪光灯。我也更喜欢把白平衡按键放在机身侧面，而不是像 D850 那样放在顶部靠近指令拨盘的位置，尽管 D850 上的布局依然非常合理。传感器分辨率更高，画面偶尔会有少许颗粒感，但即便降采样到与 D610 公平比较的水平，D850 仍然胜过 D610。

但真正让我惊艳、也出乎意料的是，D850 的文件在后期调整时表现多么出色。把 D610 的文件提亮 2 档，画面会显得发灰，必须提亮曝光并加强对比来补偿，而 D850 上完全不会出现这种情况。两者的直出画质大致相当，但一旦进入后期，D850 的色彩明显更讨喜，也更容易调到自己想要的风格。

画质方面，我觉得我买到的是一台 D610，只是各方面都更好了。呈现出来的画面正是我想要的样子，比从前更讨喜。它与我熟悉、已经非常满意的镜头配合得天衣无缝。我喜欢通过光学取景器取景拍摄。

这是一台 4 年前发布的相机，但奇怪的是，我觉得它很长时间内都不会落伍。佳能和索尼最新最强的产品在画质上几乎没有提升（实际上我更喜欢尼康的画质），使用起来也没有明显更顺手或更愉快。我不认为索尼会很快让自己全新发布的 Alpha One 显得过时，腾龙 35/1.4 也与最顶尖的无反镜头不分伯仲。这是一台 4 年前发布的相机，但它是我的最爱，我认为它会在很长时间内保持竞争力。

12. 相机镜头：我的包里有什么

相机镜头价格不菲。如果你持有 DSLR，你的镜头花费多半比机身本身还贵。这是因为镜头往往是画质的瓶颈。如果你隔着一扇模糊的窗户往外看，传感器再灵敏也无济于事。

此外，相机镜头形形色色、大小各异。受限于物理原理，不可能出现一支镜头包打天下。如果某支镜头的变焦范围极大，那么它很可能在画质上有所妥协。你需要根据拍摄场景挑选合适的镜头。

如果想找到适合自己的镜头，应该先想清楚自己常在哪些场景下拍照，再挑一支适合该场景的镜头。除非你在影棚里，或者拍摄对象非常配合，否则多半不会想在拍摄过程中频繁换镜头，所以需要一支能覆盖整个拍摄场景的镜头。

我的包里有什么？

5.1 人像镜头：腾龙 35-150mm f/2.8-4

大多数专业摄影师会随身携带两支变焦镜头，一支 24-70 f/2.8 和一支 70-200 f/2.8。但对人像而言，这两支都不能完全满足你的需求。85mm 被普遍认为是“标准”的人像焦段，不过 50mm 甚至 35mm 也很常见。问题在于，人像常用的焦段范围呈现出以 70mm 为中心的正态分布。为了拍到合适的照片，你得不停地在两支镜头之间来回切换。我用 24-70 拍过人像，发现自己总是处在变焦范围的长焦端，常常希望再多一点焦距。

腾龙 35-150 之所以完美，是因为它的变焦范围中心正好也是我拍人像所使用的焦段中心。它的变焦倍率比标准的 24-70 稍大一些（5 倍对比 3 倍），但它并不是那种在画质上做出妥协的超大变焦镜头。大多数实验室测试都把它和某些老款 70-200 f/2.8 镜头评为同一水平，所以相比更标准的变焦，你并不会损失太多。

此外，它的速度也相当快。它采用可变光圈，所以在变焦范围的长焦端会损失一些速度，但 35mm 下的 f/2.8 和标准的 24-70 一样快。这意味着，如果你在室内没有闪光灯，它有时仍然可以使用。

它并不是那种“把一切功能都塞进去”的镜头，但对我来说已经非常接近了。如果我要去某处一日游、不确定会拍什么，我常常会带上这支镜头。缺点是它体积很大（尤其是在加装 FTZ 转接环时），对焦速度比一些原厂 Z 卡口镜头稍慢、稳定性也稍逊，而且相比一支优秀定焦镜头，它的锐度和速度都略逊一筹。

5.2 暗光拍摄：50mm f/1.8 G

Ok, well, I'm currently looking to upgrade my 50mm to a different lens of the same type, either the 50mm f/1.8 Z or the 50mm f/1.4 Sigma Art. I bought it because it was cheap, and did something none of my other lenses had. It's not even sharper than my zoom lenses at it's only focal length of 50mm, but it does have one thing those zoom lenses do not: a maximum aperature of f/1.8. If it's dark, and you need light, a fast prime is your only option. Once you're already cranking your camera's ISO the extra stop over an f/2.8 makes a huge difference. A fixed length prime is the least flexible, but it's important when you need it. Also, most fixed length lenses are sharper than their zoomy counterparts, but not this one :(.

5.3 Lightweight Travel/Group Photos: 24-70mm f/4 S

I'm gonna be honest, the only reason why I own this lens is it was half off when it came with my camera. Don't get me wrong, it's still a good lens, but I probably wouldn't have gone out and purchased it on my own. What I like about this lens is that it's super reliable. It's a native Z mount lens on a Z body camera, and the autofocus just works better than any of the other lenses on the same camera. The image quality is also consistently good at f/4 throughout the zoom range, so much so that I don't have to think about it. If I'm going out for the day with casual people, this is the lens that I take. Most people don't want to wait while you fiddle with your settings, and most group photos don't care about artistic framing. The people care that you got the shot, the one where everyoen was smiling, and it didn't come out weird. If it's blurry that's really bad and you likely don't have that many do-overs. It's also super small on the Z body, and feels nicely balanced. Most travel photos I take are wide-angle anyway, because I want to capture the environment, and you're often in those situations where you have to pull the shot quickly and not mess it up. The lack of light can be inconvenient sometimes, but so far it hasn't been worth it to me to shell out for the much better (and much more expensive) 24-70mm f/2.8 S lens.

13. Why Many Primary Forecasts are Wrong

Ok, wrong might be a little bit too severe, but at the very least unfairly biased. I think most models underestimate Warren and Sander's chances of winning the primary because it evaluates the voting in every state as a single event, rather than viewing the primaries as a sequential selection process. In past elections, while the later states still matter, the early voting states have had special significance because they narrow down the pool of candidates by eliminating out candidates which seem unlikely to win.

Let's say you're a Democratic moderate. Even if you really really like Tulsi Gabbard, you're probably not going to vote for her, because you know that if you do, your vote won't matter. Since she's not going to run anyway, you might as well vote for Mayor Pete or Joe Biden. When the votes from the early states come in, they tell the people what candidates are unlikely to win. When a candidate does badly in an early state, people become less willing to vote for them and they become even more unlikely to win.

While eventually the race usally comes down to two or three, often closely split candidates, it's important to remember that primary elections are an unstable equillibrium, and it can be easy for things to snowball. Right now Biden is by most polling clear favorite to win the election, with roughly double the odds of the next major candidate. But it's important to remember that this is because Elizabeth Warren and Bernie Sanders split the progressive vote, and their policies and supporters are in many ways very similar. If one or the other starts to outpreform in some of the early states, they're going to start picking up all of the progressive votes, and start to become competitive in the states where Biden was heavily favored. While Biden could also gain some ground from early victories, he has less room to grow because the moderate vote is less split, and the other candidates to gain from like Mayor Pete, have some significant ideological differences.

14. The Wonky Math behind the Democratic Primary

Oh man, if you thought the electoral college was bad, you're in for a real treat.

On paper, it looks like primaries are mostly the same process. Every state gets "electoral college delegates", and the number of delegates for each state is the same as in the presidential election: two plus some extra porportionate to the state's population.

The thing is, the number of delegates is porportional to the state's TOTAL population, not the number of people that actually vote in the election. If California has 50 million citizens, but only 5 people actually vote, those 5 people have tremendous influence. In primaries, certain states are dominated by particular parties. That means even though Tennessee has 5 million elligible citizens, it only has 400 thousand voting in the Democratic primary. That means if you're a Democrat in Tenessee, you have much more influence than a Democrat from California. The state is considered large even though the number of Democrats actually voting is small.

7.1 Biden is Winning in States with Low Voter Turnout

Ever wonder why Biden seems so unpopular with so many voters, yet forecasts frequently predict him to win the election? Biden's best state's, according to 538: Alabama, Mississippi ,Delaware ,South Carolina, North Carolina, and Louisiana, are ALL low Democratic voter turnout states. Biden isn't actually that popular with the Democratic party overall, but it doesn't matter because he's disporportionately popular among the subsets of people that have the most voting power.

By contrast, take a look at the best states for Bernie Sanders: Vermont, Utah, Washington, California, Colorado, New Hampshire, all heavily Democratic states. Because he's appealing to more liberal states, and because his message resonates with certain groups of people, he's being disproportionately hurt by the electoral college relative to his popularity.

https://fivethirtyeight.com/features/biden-is-the-front-runner-but-there-is-no-clear-favorite/

15. What's Going on With The Impeachment

The impeachment trial, as it stands, is a little bit of a mess, with what feels like many muddled and ever-changing accusations, against a president who keeps doing new things to be accused of. Right now Nancy Pelosi orchestrated an impeachment vote in the house, but has stalled refusing to send a vote to the senate. Given all the confusion I'd like to take a step back to understand how we got here, and why it all happened the way it did.

Before anything, we need to understand that this is first and foremost a popularity contest, with a jury of the American public. There are legal rules, sure, but the people who are voting on impeachment are congressmen, and congressmen represent the desires of their constituents. You can see this in the correllation between congressmen's willingness to defend Trump and his popularity within their home state, as well as how their treatment of him changed as he rose in popularity within the republican party. In order to get him impeached, the Democrats need to convince the people that he should be removed from office.

8.1 The Russia Investiagtion

On November 8, 2016 Democrats were absolutely stunned. Despite all the polls from trustworthy sources like the New York Times*, an unthinkable candidate had won the election. There had to have been a mistake. There had to have been something deeply wrong, Democrats everywhere searched for an explanation.

Not believing that such a candidate could have won fairly, they blamed James Comey for unfairly reporting on Hillary's e-mails just days before the election. They blamed misleading polls discouraging Democrats from voting in an already "certain" election. And finally, they blamed Russia for interfering in our Democratic process, certain that for this to happen, rules must have been deeply and influentially broken.

There was a lot of evidence suggesting wrongdoing. Trump always seemed to have a soft spot when he spoke of Russia. There were the meetings in Trump tower. And Trump seemed to be doing everything in his power to kill the investigation.

The first weeks of Trump's administration were hectic to say the least. Trump picked fights with anyone and everyone. He struggled to assemble a cabinet and wound up leaving many key government positions unfilled. He seemed hell-bent on ignoring all rules and structure, as much as he possibly could. And finally, in a move that shocked many, he fired the person investigating him, FBI director James Comey.

Things moved quickly. Deputy Attorney Rod Rosenstien appointed a special counsel to replace Comey in the Russia investigation. Mueller seemed like just the man for the job, and over the coming weeks, his team gained momentum. Democrats saw not only mountains of evidence piling up, but real criminal offenses, with Mueller and his team reaching indictment after indictment. And Trump, although powerless to fire Mueller seemed to be doing everything in his power to kill the investigation.

The Democrats saw the way the investigation was going, and they saw the indictments lining up, and they made a crucial tactical error. Instead of zeroing in on the firing of Comey, the offense they knew he had committed, they threw all their efforts behind the Russia investigation, doing everything possible to stoke public excitement for a dramatic Mueller testimony.

They had a precedent from the Nixon investigation, one of the few issues on which Trump lacked positive sentiment within the Republican party, and what looked to be an undeniable act of treason. They thought the Mueller report was going to be a bombshell. That they were going to catch Trump red-handed, and that it would suddenly and dramatically sway public opinion. They were expecting something so severe that it would be impossible to claim impeachment was a purely partisan investigation. Even many Republicans were holding their words, waiting to see what the report described.

A week before the report's release, the Democrats made another tactical error. Instead of demanding Mueller first testify to the public directly, they allowed his report to pass, largely unquestioned, through Trump's justice department, and Attorney General William Barr. I don't know if they could have stopped Barr from publishing his letter, but they could have kept the tension, pushing people to wait for the bombshell that was the final report. Instead, they let the singular news story, the one they were building up for months, be not Mueller's report itself, but Barr's report on Mueller's report.

8.2 The Mueller Report

On March 24, 2019, Barr issued his letter to Congress. The verdict was "no collusion", i.e. the investigation did not establish coordination between Trump and the Russian government. Importantly, despite all the evidence in the report, Barr himself decided that Trump had not committed obstruction of justice, and attached his own findings, instead of the report's in his letter. Mueller himself described Barr's letter as a mischaracterization of his report, and in another mistake, Mueller expressed his discontent privately instead of immediately going to the public.

It was only after the full report was released, nearly a month later, that people began to realize what happened. Most neutral media organizations and fact checkers described Barr's letter to be intentionally misleading. Instead of "totally exhonerating" the president, Mueller had carefully layed out the case for obstruction of justice, and seemed only to not reach a verdict because it was not in the mandate for his report. In addition, instead of finding no evidence of direct collusion, he did suggest there was questionable behavior towards Russia.

By this point the moment had long since passed. Democrats had been hyping up a jarring news story, and what they got was a drawn out fizzle. There was a lot of dirt in the Mueller report, but Democrats didn't get the conclusive proof they claimed was lurking. They tried to pivot and push further, but having failed the first time, it looks like they were just grasping at ways to accuse the president.

If they had focused in on obstruction of justice, they would have gotten the president on an act they knew he had committed, and bad news from the Russia investigation wouldn't have broken their momentum. On the other hand, if they had spent months arguing that Trump had broken rules by firing Comey, and real evidence was found from the Russia investigation it still would have been the bombshell they needed to sway public sentiment and press further. This is again an instance where Democratic leadership was too confident and failed to plan effectively to minimize their downside.

8.3 Rebuilding a Case

In the months after the Barr's letter, the Democrats were frustrated and disorganized. They sensed Trump was doing something wrong still, and pushing the boundries, but they lacked a single rallying cry. If they had done so from the beginning, their strongest play would have been to rally around the firing of James Comey, but by now years had passed, and it would be hard to pick back up again.

Despite the lack of a central rallying cry Democratic voters were angry, and they were growing more angry with each new presidential offense. More and more moderate Democrats, long evasive, now publically voiced their support for impeachment. As primary season rolled in, all eyes turned to the candidates, who felt rising pressure from the left-most wing of the party. Unable to avoid the question in debate after debate, the candidates all lined up in support of impeachment. Only Nancy Pelosi and her closest allies held their breath.

8.4 The Ukraine Scandal

Finally, in September, news broke that President Trump asked the Ukranian government to investigate Hunter Biden the son of his chief political rival. Abuse of power to investigate a political oponent drew such a parallel to Watergate that it immediately set off alarm bells in politician's minds. Not only did they have more evidence of wrongdoing, but their anger had a central focus once again. Pelosi, sensing a shift in momentum, announced that she would begin the impeachment process. After an announcement from their extremely cautious, yet trusted leader, the rest of the Democratic party followed suit.

Why was Pelosi cautious for so long? I think it's because you only get one shot to announce articles of impeachment (or at least, if your first attempt fails, it makes your second attempt harder). I think she was saving it, waiting for a news story like this one to break, so that when something like this did come out, she could investigate it and apply pressure on it. Starting an impeachment proceding draws national news coverage, and she wanted an event that the American people would see as definitely wrong, something that was worth spending her news coverage on.

The impeachment process also makes it easier for the Democrats to carry out a formal investigation. This means that the democrats got to learn about the Ukraine scandal in more detail, and potentially uncover even more damaging information. In addition, waiting for so long gave some kind of additional credibility to the investigation. It sends the message that Pelosi feels "we don't want to do this, and we won't unless we really have to, but as it turns out we really have to."

In addition, Pelosi learned from the mistakes of the Russia investigation. Despite gaining such a powerful rallying cry, she didn't make the impeachment entirely reliant on proving coordination with Ukraine. She included an article "Obstruction of Congress", about Trump's efforts to block investigation, which relies almost entirely on publically available knowledge. As such, it can't be easily disproven, so Democrats can keep pushing in that direction no matter what other information comes out.

8.5 The Threshold of Popular Support

Despite all this, Nancy Pelosi is stuck. One of the reasons why she stalled on starting the impeachment process for so long is she knew that she wouldn't be able to carry it through to completion. At least not through the Reupblican controlled Senate, where Mitch McConnel has promised a speedy acquittal. And yet, she still wanted to start the impeachment process so that she could draw public attention and promote an investigation. So, she drafted the articles of impeachment, got them passed in the house, and then, did the only thing that she could do. She stalled. Hearing Mitch McConnel's vow, she refused to send the articles of impeachment to the senate. This means that Mitch can't kill the impeachment, and she gets to keep up the pressure until the tide turns in her favor.

What gives? Trump is an unpopular president. Just a month after he took office, over 50% of people disapproved of his presidency, according to polling tracked by 538. As of this moment, more Americans support impeaching Trump than those who are against it. Congressional votes are largely determined by the opinions of their voter-base. So why can't Pelosi get the votes she needs?

Well, impeachment is very popular among Democratic voters, but as you might have guessed, it's not very popular among Republican voters. And the thing is, it's Republican senators that need to vote for impeachment for the thing to pass. 50% of people might want Trump impeached, but that breaks down to like 80% of Democrats and only 10% of Republicans. Republican senators likely won't support impeachment unless a majority of their own party supports it. That means Trump has to be really unpopular.

Given the unwavering support of his base, it's not clear what it would take to move the needle on impeachment. Polls show an uptick in support for impeachment after news of the Ukraine scandal broke, but only a small uptick, and one that still gives impeachment less than 10% support among Republicans. If anything, the history behind the impeachment proceedings, and the haphazard way in which it all went down made it much harder to impeach Trump because it gave wiggle room for his most ferverent supporters to believe it was all just a partisan mess. I don't think it'll go anywhere in the near future, because if Pelosi is smart, she'll hold her cards and continue to prevent a Senate vote, and if Trump doesn't make any big mistakes, he likely won't lose enough popular support to be impeached.

I think going forward, the stalemate will continue. Pelosi doesn't have the votes, and shouldn't risk a clean shot at impeaching the president. Senate Republicans might try to entice her with the promise of additional testimony in the house trial, but I think Pelosi would be stupid to take it. I think Pelosi is going to wait for the next big bombshell, and then pull the trigger whenever the momentum swings.

16. Treating Cancer with Chemo (Part 3)

In this section, we now look at the specific drugs we would need to put the ideas in Part 2 into practice. We have a tremendous advantage here because we are not developing new chemicals. All of the drugs listed have already seen clinical use. What we'd like to do is, send a signal that tells all the cells in your body to start dividing, and then send a chemotherapy drug which kills any cells that are still dividing. Your body has many types of signals which tell cells when to grow and divide,

notes: platelts come from Megakaryocytes which come from hematopoietic stem cells

I found the paper that has this idea https://www.ncbi.nlm.nih.gov/pubmed/8839398

17. Treating Cancer with Chemo (Part 2)

In the last section we looked into what types of cancer treatments are out there, and, from a research perspective, what areas hopefully have the most low hanging fruit. Due to cancer heterogeneity, and the problem of targeting undetectable or microscopic tumors, black-box chemotherapy methods seemed like they would yield the best results. I wrote down the definition of our "black box" and narrowed in on one property, that cancer cells divide regardless of external signaling.

Many drugs already target this pathway. Certain drugs attack cells which divide more rapidly, intentionally causing damage to cells during the process of cell division. In theory this will cause cancer cells to die because they will be more likely to die than the normal cells.

The problem is, this also kills healthy cells. Some cells, such as bone marrow cells, divide rapidly and, some cancers don't even grow that quickly. Some cancers grow at a normal rate for cells, they just don't stop growing when they're supposed to. The problem is that what you want to measure is not the cancer's rate of growth, but it's responsiveness to external signals within the body.

So how can we target this? How do we identify cells which don't respond to signals?

Well, instead of targeting cancer cells, we can send signals to identify healthy cells. We know that the healthy cells will respond to the signals and the cancer cells won't. By targeting everything else, we can measure when something doesn't respond to the signals that it's supposed to, and then go after it. Instead of attacking cells that divide quickly, we can tell all of the cells to stop dividing, and then kill all the cells that still divide. The cancer cells won't listen, but the normal cells will, so they won't get the damage that usually comes with intensive chemotherapy.

What's completely wild about this to me is that people are doing this already, but not on the cells that actually matter. We treat hair loss. We can help people not lose all their hair who are going through chemo by having them apply ice packs to the top of their head while they're receiving treatment. This works by slowing the division of the cells around your hair follicles, so they won't actually get killed and fall out while you're getting treatment.

In the same way that we treat hair loss, we can tell the cells in your bone marrow to stop dividing as quickly while the chemo is active and thus drastically reduce the side effects. Furthermore, we'd be able to use drugs that weren't previously available because even though they killed the cancer they were otherwise too toxic to the human body. Even better, we already have the tools to do this. We already have drugs which slow the growth of bone marrow cells. We just have to apply them in the right way.

18. Treating Cancer with Chemo (Part 1)

Cancer treatment is something I've thought about a lot recently, and the more I've studied it, the more I've seen that it's an interesting and complex problem. At it's core a cancer is a group of cells whose genetic code has mutated such that they divide uncontrollably, in a way that cannot be properly regulated by the body. Cancer cells, in contrast to benign tumors, also have the ability to sustain their growth, and to invade other parts of the body. If the cancer cells grow too much, they can "crowd out" other bodily functions, by having large tumors that press against vital organs, or "replace" other healthy cells.

The challenge with treatment is that cancer cells, unlike bacteria or viruses, look remarkably like healthy human cells. It's difficult to kill the cancer cells without accidentally killing the normal cells. As a result, the main treatment options for cancer are:

Surgery: Cancers typically grow as a cluster of cells in the same physical location, called a tumor. One way to treat this is to just cut out the tumor. This is good because it can quickly and fully remove the entire tumor, but bad because it might not be in a place you can safely reach with a knife, and bad because it doesn't catch it if the cancer spread without your knowledge.
Radiation: Radiation zaps the cancer with high intensity beams of light. Think a really powerful microwave pointed directly at your tumor. This is good because it kills cells in places that aren't easy to reach with a knife, and can cover a wider area than would be safe to cut out, but bad because it might not kill everything in its beam and again, it won't catch the cancer if it spreads.
Chemotherapy: This is a broad term that refers to using chemicals to kill, or otherwise harm the cancer cells. Since this refers to many different types of chemicals, which act in many different types of ways, it's hard to give a one-size-fits all explanation. One type of chemical which is fairly common essentially tries to kill cells which are dividing very quickly or haphazardly, hoping to target the cancer cells unctonrollable growth. Chemotherapy is great because unlike radiation or surgery, it can attack cancer cells anywhere in the body, meaning that if the tumor had spread to a small undetectable cluster of cells, chemotherapy would still fight it. The downside of chemotherapy is it's not always very good at determining which cells are cancer cells. This means it can kill healthy cells all over your body, most notably in the bone marrow, which tends to replenish quickly.

There are a few other types of treatments, such as immunotherapy, which I'll skip for now as largely niche (but growing) fields.

I'd like to focus here on chemotherapy because, while currently surgery and radiation are often necessary, because they are unable to attack small groups of cells, I don't believe that in the future, anything more than incremental progress will be made. Whatever the solution is will likely have to be some sort of medicine because we can't physically target something we don't have the ability to detect.

Medicines that treat cancer (even including immunotherapy) are hard to make because it's hard for anything to classify whether or not a cell is cancerous. So, in order to treat cancer, we need to solve the classification problem, build a model to predict whether or not a cell is cancerous, and then encode it on a macroscopic scale. The more accurate our model, the more likely that the drug will kill the cancer without being otherwise fatal to the human body.

We can approach this classification as a "black box", i.e. the cell is a generic thing with certain properties which we can't see into. Or, we can think of it as an open box, i.e. the mutated DNA of a cancer cell causes these specific chemical reactions which, eventually cause the cell to divide uncontrollably. With open box models, we have a tremendous amount more information that we can use to classify the cell as dangerous, and it seems like this would be intrinsically easier, but this information comes at a cost. Cancer is a name for a disease that can occur in many different ways and for many different reasons, often exhibiting significant heterogeneity within a single tumor. If we treat one specific pathway of chemical reactions, we'll only solve the problem for those specific cells, and we'll miss out on all the different types of cancer that are possible.

I'd like to think about this problem as a black box problem because in addition to the curse of heterogeneity, it's also just better suited to my background. Treating the cell as an open box generally means lab work with equipment I don't have access to, and at best a 5-10 year trial period before a new chemical gets into production. It's my hypothesis that there are enough building blocks that are already well studied, that if we put them together in just the right way, we'll get a better result.

So let's take a look at the black box. I took a look at the definition of a cancer cell. It boils down to roughly:

The cells grow and divide regardless of what body signals them to do
They don't die when they're supposed to
They can invade other parts of the body
They get the rest of the body to support their growth (encourage blood vessel construction)

I'd say 2 is almost impossible to go after as a black box, since you can't ask a cell whether it will die without killing it. 3 seems similarly difficult because at best it can only attack spread, and not the original tumor, and because it's hard to ask a cell where it originally came from. 4 seems promising, but again has the problem that it likely won't cure tumors of smaller than a certain size. So, all things considered, I'd like to take a closer look at 1, that cancer cells grow and divide regardless of the body's signals.

19. The Importance of Doing Nothing

This morning I woke up, a little on the later side, and didn't have anything important to do. I didn't have work, and I didn't have chores, just had to eat breakfast, and be out the door by noon. I had 2 and a half hours, so it should have been easy.

With so much time, and nothing, really, to do, I started browsing the front page of reddit. I thought maybe the act of reading, or looking at something would wake me up a little. 30 minutes passed. This was no big deal, as I had all morning, really, but I felt like I wasn't quite done yet, so I kept going.

By 1130, I was still searching for something initeresting, and I still hadn't found what I was looking for. I knew I was wasting time, and I knew that I wasn't going to feel fully satisfied at the end of it, but it always felt like the next two or three minutes would make me feel a little bit better, and that little time wouldn't be a deal breaker. By the end of it, I would click on a YouTube video and only watch the last thirty seconds of it, because in truth I didn't really want to watch the video, but I wanted to feel like I was going to find something new and entertaining. I didn't really feel like I was thinking at all, or even that I was enjoying what I was doing, but I wanted to see something. You can probably guess that I rushed the quickest breakfast (read: two bananas), and was out the door 20 minutes late.

Whenever I mess something up, I ask myself how I could of done better. In this case it seems pretty obvious. I could have just not wasted 3 hours scrolling around on the internet. I could have looked at Reddit for only an hour and a half, made myself a delicious breakfast, and still left on time.

But in practice, it's not that simple. That's because what you do with your time affects your ability to think and make decisions. If, after 90 minutes, I took a break to really sit, and think clearly about what I wanted to be doing, I would have realized I still wanted to breakfast, and that I wanted to be on time. But because I was watching videos, and because I was constantly blasting my mind with different signals, I never had a moment to really process what was going on.

Not only that, I noticed even after I left my apartment, after I had already stopped looking at a screen, I was still left with that same mentality. While I was walking to the train, I still felt this impulse to play chess or read the news. I wasn't thinking about the rest of the day, or about things I wanted to learn, or things I really cared about. For most of the afternoon, my head was still overloaded, and when I went to think, what came out were random memes and stories. I had given myself so much new information to process, that even after I was done looking at it, it still crowded out my other thoughts. It's not just about the moment, wasting your time does damage, and it's important to recognize what damage it's doing.

Maybe the videos made me feel happy for a half hour or so. Maybe I even learned a tiny bit, about news, or life advice, or something that might make good conversation. But I think I would have been better off if I had done, really, absolutely nothing. I would have been better off if I just sat down in a chair, and just waited. When I do nothing, it gives me the mental space to process longer-term thoughts, to prepare for my day, and to figure out what I really, genuinely want to be doing.

20. Why I Support Citizens United

Most people I know really don't like Citizens United. For those who aren't familiar, Citizens United was a supreme court decision that essentially let corporations and wealthy individuals funnel unlimited amounts of money into American politics. Not only that, it let people donate to certain types of political causes with little to no transparency or oversight. The people I know who don't like it view it as letting big money buy elections and spread corruption in American politics.

We have anti-bribery laws (which the courts have upheld). We have campaign finance laws (which the courts have upheld). So, what makes the Citizens United ruling different?

The Citizens United ruling applies only to campaign advertisements (not general campaign contributions), and only to money that is spent "independently", i.e. without talking directly to the candidate or political party. This means that if I'm Exxon Mobile, I still can't write Mitch McConnell a check for 10 million dollars. I also can't work with Mitch McConnel to film a TV ad for his upcoming election. But if I decide that I really like Mitch McConnel, and I don't talk to anybody, and I just pay for an ad telling everyone how great he is, then nobody can stop me.

As experience has shown, advertising can make a huge difference in the outcome of an election. If someone is allowed to spend an infinite amount of money on advertising, they, as an individual can have a massive influence on the outcome of an election. We have anti-bribery laws, but if someone can use money to ensure candidates with certain political views are elected, aren't the effects the same?

I like to think of this by considering the role of a news organization. We need some news organizations discussing American politics in order to keep voters informed, so that they can better understand the issues and decide who to vote for. Along these lines, it's even important to have news organizations that debate different topics, and to have editorials which discuss their own opinions because it's important for people to understand the reasoning behind the legislation, and get a sense of why or why not something is a good idea. Especially for traditional media outlets like newspapers and television, It's nearly impossible to publish editorials or simply report the news without incurring some sort of bias.

If you wanted to start a new cable news channel, you'd have to pay for a studio, for cable companies to distribute your broadcast, and to advertise your broadcast to others. Your news broadcast might be extremely biased, because, as we've already discussed, it's impossible to regulate bias in news organizations. You might not want a long-form news broadcast, so instead of paying for an entire television channel for an entire day, you might want to only purchase a slot that's the length of your show, or the message you want to publish. Separately, you might also want to be able to advertise your show (as news organizations are able to advertise), and in doing so accurately describe what's in it.

So, the question is, where do you draw the line? What's the difference between paying to broadcast an editorial piece and donating directly to a political campaign? Well, I think making sure that the broadcast isn't part of (organized by) the political campaign itself is a pretty fair place to draw the line.

21. Camera Settings

When I got my first DSLR, I opened up the settings menu, and found it loaded with options: active-d lighting, vignette control, auto-bracketing, ambient lighting, and much more. It took a while to sift through all the options, but after assigning some reasonable defaults (use full resolution, turn on vibration reduction when available, etc.), I found that despite all of the options, there were really only four dials that effected the raw image. The rest were either only in post-processing, or just different ways of twisting those original four dials.

Despite all the complexity, the only four dials you have on your camera are:

Exposure

Your camera has a sensor in it. When you take a picture, the camera opens up and lets light into the sensor. The longer the camera opens up, the more time light has to hit the sensor, and the brighter the overall image.

This comes from a time before people had digital cameras. When people used film, each wave of light would cause a chemical reaction on the film, so letting in waves of light for a longer period of time would cause more chemical reactions, and all of the individual waves of light would "add together" to form the overall image. That's why if you move a camera around while the film is exposed, the image comes out blurry.
Focus Distance

In front of the sensor, your camera has a lens. The job of the lens is to take in all of the light from the outside world, and bend it, so that the light is pointed at the sensor. Typically the outside world is larger than the size of the sensor, so the lens bends the light toward the center, with the light kind of following the shape of a cone. If the lens bends the light correctly, the square of the outside world that you want to capture will hit the image like a bunch of parallel beams of light, and you'll project a much bigger image onto a much smaller sensor. However, if the cone of light is too short, or too long, instead of hitting the image like parallel lines, the beams of light will get criss-crossed. Instead of one exact point on the outside world corresponding to one exact spot on the sensor, it might miss, and send some of the light kind of nearby, causing the image to be blurry. The correct focal distance is different for objects at different distances from the camera, so if you make nearby objects in focus, you make faraway objects out of focus.

Long story short, if a part of an image is not in focus, it looks blurry. You get to choose a distance away from the camera such that objects at that distance are in focus. The further away an object is from the focus you chose, the more blurry it looks in the photo.
Aperture

Typically, there's a physical shutter that opens up when you take a picture to let light onto the film or sensor. Aperature refers to how big of an opening there is when the shutter is open. If there's a really big hole, it lets in a lot of light, but if there's really small hole, it only lets in a little bit of light. Also, if there's a small hole, the cone of light (from the focus section) is skinnier, so the beams don't get criss-crossed as much. This means that objects that aren't at the focus distance don't wind up being as blurry, because there's less interference. If there's a big hole, the cone is wider, so only objects at a specific distance are in focus, and everything else comes out blurry. This can be useful in taking portraits when you want to emphasize an in-focus person against an out-of-focus background.
ISO

How sensitive your camera sensors are to light. High ISO means your camera registers a lot of input from a small amount of light. Most cameras are only able to get accurate readings up to a certain level of sensitivity, so bumping your ISO too high can result in noisy/grainy images.

So why do cameras have so many other options? Well, taking photos in the real world it isn't always easy to quickly and accurately adjust these four dials to set up for the shot you want. Want to get that monkey in focus? The one that's swinging across the jungle? Good luck twisting your focus knob and trying to keep up with it. You'll probably want your camera's autofocus tracking. Want to take many photos in a row, with different levels of exposure? There's a setting for that. It's not enough to just have access to those dials, you also want convenience tools that let you quickly choose how and when to take a photo.

In addition, even though it doesn't effect the raw image, many cameras have built-in post-processing to adjust for things like ambient lighting. While for more serious shots, you'll probably edit the raw files in a program on your computer like photoshop or lightroom, it is convenient, visually to get quick feedback on what the image might look like, and for those of us that are lazy out there, can be a quick way to produce reasonable quality images.