「拜读了您的文章，非常受益，能不能约时间向您请教？」——这句话的实际内容是：请你给我一小时。他每周收十条，全一样；而你发出去那一刻自己就不舒服——你知道你什么都没带。
「你上个月写的那套排班法，我在十二个人的班组照着跑了两周。有个跟你预期相反的结果：夜班那组反而更省人，因为交接被合并了。数据在这，也许有用。」——你带的是他拿不到的一手观察。这一条里不要有任何请求，要留到下一次。
EN Most people avoid networking not out of laziness but because it makes them feel slightly dirty — and that feeling has a specific trigger. Research on instrumental networking found that people asked to build ties for professional gain reported measurably more moral discomfort and did less of it, with the effect strongest among those in lower-power positions. The useful reading is that the discomfort isn't a character flaw, it's an accurate read on a lopsided exchange: if all you bring is a request, you are the first person to classify the interaction as using someone, and that classification leaks into your posture and your wording. The fix isn't a thicker skin, it's arriving with something. It doesn't have to be equivalent — five minutes of value is enough — and the four cheapest currencies are usually available to a beginner: a first-hand observation from the front line that a senior person can't get, public and specific credit, an introduction where both sides were asked first, and real feedback that proves you actually used the thing. Giving also isn't pure cost: donating money raises activity in the ventral striatum, the core of the reward system, which is why the give-first version is the one you can keep doing.
不适感是信号，不是缺陷。一次接触里你只取不给，你自己会先给它贴上「利用」的标签——标签一贴，姿态和措辞全都变形，对方接到的是紧绷和算计。带上一样真能给的东西，等于把这次接触重新归类，报警就解除。而且给予不是单纯的成本：它自己走奖赏回路，所以「先给」是少数能长期坚持的打法。
边界：先给不等于讨好。给了就在心里记账等着提，对方能感觉到——那只是把索取延后。判据：如果他永远没法回报你，这件事你还做吗？答不出「做」，就别用「先给」包装，直接开口要还更体面。
"No ask attached — just thought you'd want to see this." —— 明说这次不要东西，可信度立刻不同。
"I actually ran it, and here's what surprised me." —— 用「我真的做了」区别于「我读了」。
"Happy to make the intro if you want it." —— 牵线之前先问，别自作主张。
「你那边有没有什么机会？」——他们和你在同一个群、看同样的消息、认识同一批人。你问到的多半是你已经知道的。问完还会误以为「市场上没机会」。
翻通讯录，挑出一年以上没联系、但当年真共事过的人——前同事的前同事、三年前的合作方、上一份工作的客户。发给其中一个：
「好久没联系。我在往 A 方向走，想找做 B 的团队。不用帮我介绍，我只想知道：这两年这块最活跃的是谁？」——把「有机会吗」换成他答得出的具体问题，回复率完全不同。
EN Your three closest contacts sit in the same information pool you do: they read the same feeds, know the same people, and the opportunities they know about are largely ones you've already heard of. New information arrives across bridges — the single link connecting you to a cluster you're not part of. That's the classic weak-tie argument, and it now has causal backing rather than just correlational: five years of randomised experiments on roughly twenty million LinkedIn users found that moderately weak ties produced the largest increase in job mobility, while the strongest ties helped least. The important refinement is that the curve is an inverted U, not a slope. Complete strangers can't help you either — they don't know you well enough to vouch. The valuable band is people who worked with you once and still remember you: roughly the "haven't spoken in a year but it wouldn't be strange to write" distance. And don't liquidate your strong ties in the process — weak ties supply information, but strong ties are what supply effort, vouching, and help when something goes wrong.
机制是信息冗余，不是人数。你和密友泡在同一个信息池，他们能给你的边际信息接近零。跨到另一个圈子的那条线（结构上叫「桥」）才是新信息入口——不是那个人更厉害，是他站在你看不到的地方。但也不是越弱越好：太弱的人不了解你，没法替你说话。值钱的是中间那一档：还记得你、知道你什么水平，但已不在你的信息池里。
边界：弱连接值钱的前提是它真的跨圈。同一个行业群里加的两百个人不是弱连接，是同一池子里的陌生人——信息一样重复，还没交情。另外这条最稳的场景是找工作与找信息；要人真出力时，该回到强连接。
"You're the only person I know who'd know this." —— 点明你为什么问他，比群发有效得多。
"Not asking for an intro — just who's doing interesting work here?" —— 先要信息，不要人情。
"It's been a while, and that's on me." —— 长期没联系时最好用的一句开场。
「哈喽在忙吗～最近怎么样呀？」——两句寒暄之后拐到请求，对方全程都在等那个「其实是想问你」。掩饰断联比断联本身更伤：连这几句热络都成了工序。
「好久没联系，这点是我的问题。前几天我们在复盘上线流程，我把你当年那套灰度顺序原样搬过来用了，省了一堆事，想起来该跟你说一声。你现在在做什么方向？」
——承认间隔 + 具名认可（存入）+ 一个他答得出的问题（不是请求）。要帮的忙下次再提，这一条只为把账户重新打开。
EN Other people's willingness to help you behaves like an account balance, and showing up only when you need something is a pure withdrawal every time. What most people miss is how cheap the deposits are: public, specific credit that names what you actually used; an introduction where you asked both sides first; a single message that says "saw this, thought of you" and attaches something genuinely useful. None of these cost you an hour. The most underrated asset on that ledger is the dormant contact — someone you worked with once and haven't spoken to in years. In one study, executives asked to reconnect with contacts they hadn't spoken to in three years or more rated the advice they got as more valuable and more novel than advice from their currently active contacts, which makes sense structurally: the old trust survives without needing to be rebuilt, while three years of separate lives means their information no longer overlaps with yours. The one rule when reconnecting is not to paper over the gap. Name it, give a concrete memory anchor, and don't attach the ask to the first message.
这是重复博弈，对方评估的不是这次的请求，是你这个人的模式。只在有事时出现的人模式清清楚楚，第三次就没人回——不是他小气，是他的时间在被单向消耗。低成本存入改变的不是人情账，是你在他脑子里的检索位置：他跟别人聊到某件事时会不会想起你。休眠连接则占了两个好处——旧信任不用重建，而分开这几年让你们的信息不再重叠。
边界：账户这个比喻有个危险用法——把每次给予记成债权等着收。真记了账，味道就变了，对方能闻出来。更稳的是给的时候就当没回报，回报来了算意外。另外别把存入做成骚扰：每周一次没有实质内容的问候，消耗的是同一个余额。
"It's been three years and that's on me." —— 承认间隔，比装作没断过可信。
"I used your thing and it worked — wanted you to know." —— 存入的最小单位。
"No ask this time." —— 明确写出来，对方才会真的放松。
「不知道方不方便，能不能麻烦您有空帮我看看？非常感谢！」——他没法直接回答：看多久？看什么？什么时候？于是这条被留到「等我有空再说」，然后永远没有那一天。而你会把沉默读成拒绝。
「想占你 20 分钟。就三个具体问题，我先写在这：①…… ②…… ③……这周四或下周二都行，你挑。
忙的话直接说不方便就好，我完全不会多想。
聊完我把这几家的对比表整理好发你，你那边应该也用得上。」
EN The reason people don't ask isn't temperament, it's a measurable estimation error. Across five experiments, participants predicted they would have to approach about twenty strangers to get five to agree to a request; the real number was closer to ten. The asymmetry has a clean explanation: when you imagine the request you compute what saying yes costs them, and you leave out what saying no costs them — the awkwardness of refusing a person to their face, which is often the larger of the two. That same asymmetry explains why channel matters so much. A request made in person, where declining has to be done out loud, succeeds at a far higher rate than the identical request sent by email, where ignoring it is free; and people reliably overestimate how well email works. So: say it in person or by voice when you can, and turn three dials on the ask itself — make it concrete enough to answer in one line (how long, what exactly, when), hand them a dignified way out and mean it, and state what you'll send back. And when someone does decline, thank them for saying so directly. That one sentence is what makes it possible to ask again in six months.
你算的是一套不完整的账：看得见他答应要付的时间，看不见他拒绝要付的东西——当面说「不」的社交代价，对多数人比帮这个小忙更贵。这个盲点是系统性的，方向永远是低估，所以「反正人家不会答应」这个直觉本身就错。同一个不对称也解释了渠道：邮件给了对方一个零成本的拒绝方式。三个旋钮做的是同一件事：把答应的成本压到低于拒绝的成本。
边界：这条说的是小忙。请求越大（写推荐信、担保、投钱），低估越不成立，那时该先把账户存够，别靠统计学壮胆。另外「当面更有效」不是让你去堵人：利用对方当场无法拒绝的处境施压，已越过「影响力六原理」那期讲的说服与操纵的界线——检验是你敢不敢在请求里明写「拒绝我完全没关系」，并且说到做到。
"Twenty minutes, three specific questions — here they are." —— 具体到他一行字就能回。
"A no costs you nothing with me." —— 台阶要明写，暗示没人接得到。
"Thanks for saying so directly." —— 被拒后的这一句，买的是下一次。
六十分钟，四步。今天只做存入，不做提款——除了最后一步。