他们不回你,是因为你值六美元They Don't Text Back Because You're Worth $6
725条。今年一年,一个纽约人收到725条政治募捐短信,几乎全部不是发给他的。他回复了49次STOP,其中一次是2025年11月对Jon Ossoff的竞选团队说的——之后又收到13条为Ossoff募捐的短信,来自9个不同的号码。他对DCCC回了“eat shit”。也没用。
这不是骚扰,这是定价。联邦选举委员会记录显示,被误认的那位亲戚John自2017年以来捐了大约160次,总共约1700美元,中位数6美元。今年捐了6次,共51美元。DSCC从他身上拿到过一笔5美元,2020年。此后,这个号码收到了DSCC的60条短信,其中22条集中在10月的前20天,来自28个不同的号码。5美元买60条短信,这笔账算得过来——因为短信便宜,撒出去,总有人咬钩。
真正的问题不在“谁在发”,在“谁在卖”。几乎每条短信的链接都经过几次跳转落到ActBlue,链接里带着追踪码。作者从这些字符串里反复拼出一个词:Sterling。Sterling Data Company,一家用AI帮民主党找潜在捐款人的小公司。2022年《福布斯》报道它建了约3000万政治捐款人的数据库,有些人的档案细到开什么车、在Netflix上看什么。Sterling的CEO Martin Kurucz对作者说,高频 solicit 是“a necessary evil”,并且确认公司自己运营短信业务,同一条记录可以租给多个竞选团队。他说没有数据库是完美的,中等置信度的匹配通常也够用,反正打错的号码会因为收件人不理而自然淘汰。
这就是机制。你的手机号不是你的身份,是数据库里的一行,匹配错了也不影响它继续被租出去。Kurucz给的建议是:别回复。作者自己找到的另一条路是起诉——至少有两个人靠和解拿到了“再也不联系”的承诺,但两起诉讼都没弄清错名字是怎么贴到号码上的。
用书里的框架看,这是第三章的认知入口被彻底工业化之后的样子。表达武器化的前提是“投放”,而投放需要地址。Sterling卖的就是地址。竞选团队不需要认识你,不需要你说服,甚至不需要你说的是对的话——它只需要一个置信度足够高的猜测,把“可能给钱的人”和“一串数字”连起来,然后按次计费。政治募捐在这里已经和scam共用同一套基础设施:群发、押概率、不指望多数人回应。区别只在于scam的收款方没有FEC备案。
更值得盯的是那个“800%匹配”“900%匹配”的话术。House Majority PAC从未从John那里收到过一分钱,照样不断提出把捐款放大到900%。这不是承诺,是定价技术——给你一个紧迫感,让你以为自己的6美元被乘了九倍。第三章里讲浪漫爱叙事怎么把钻石和婚姻绑定,这里是同一个手法的小型版:把一笔小额转账包装成“民主存亡”的参与凭证。
而这一切发生在你孩子的睡前。作者写得很清楚:孩子刚要睡着,叮一声,Nancy Pelosi来告诉你,不立刻捐5美元,美国民主基本就完了。几分钟后是Cory Booker。再几分钟是James Talarico。
这里没有受害者流血,但有一层东西被削掉了:注意力。725条短信、812个号码、1735条自2022年3月以来的记录,占用的是一个具体的人在下班后、开车时、做饭时、哄睡孩子时的时间。这些时间本来可以属于别的东西。差额就是暴力,哪怕施暴者是一串自动生成的字符串。
Kurucz说错号会自己掉出流通,因为收件人不理。他没说的是:不理的人被当成噪音过滤掉,而系统照常运转,因为总有人会点。作者的手机号还在名单上。DSCC现在给他发短信比他妈还勤。
他最后一句写的是:再回一百次“eat shit”也改变不了。对。因为这条短信的收件人从来就不是他。
725. That's how many political fundraising texts one New Yorker received this year, and almost none of them were meant for him. He replied STOP 49 times. In November 2025 he told Jon Ossoff's campaign some version of stop. Since then he has received 13 more texts boosting Ossoff, from nine different numbers. He told the DCCC to eat shit. That didn't work either.
This isn't harassment. It's pricing. FEC records show the relative he's confused with—call him John—made roughly 160 contributions since 2017, totaling about $1,700. The median gift was $6. This year John donated six times, for $51. The DSCC got $5 from him once, in 2020. Since then the number has received 60 DSCC texts, 22 of them in the first 20 days of October, from 28 different numbers. Five dollars buys sixty texts. The math works because texts are cheap and you only need a few people to bite.
The real question isn't who sends. It's who sells. Almost every text carried a link that, after a few redirects, landed on ActBlue, and the links carried tracking codes. Out of those strings the author kept pulling one word: Sterling. Sterling Data Company, a small firm using AI to help Democrats find likely donors. In 2022 Forbes reported it had assembled a database of roughly 30 million political donors, with hundreds of details on some of them, including what cars they drove and what they watched on Netflix. Sterling's CEO, Martin Kurucz, told the author that high-frequency solicitation is "a necessary evil," and confirmed the company runs the texting itself and can rent the same record to multiple campaigns. No contact database is perfect, he said; even moderate-confidence matches are generally accurate enough to be worth using, and wrong numbers drop out because recipients don't engage.
There's the mechanism. Your phone number is not your identity. It's a row in a database, and a bad match doesn't stop it from being rented again. Kurucz's advice: don't respond at all. The author found another exit—lawsuits. At least two people have extracted settlements from Democratic organizations over apparently misdirected texts, presumably including never being contacted again. Neither suit established how the wrong names got attached to their numbers.
In the book's terms, this is what the cognitive entrance looks like after industrialization. Weaponized expression needs a delivery address. Sterling sells the address. The campaign doesn't need to know you, doesn't need to persuade you, doesn't even need to be right about you—it needs a confidence score high enough to connect "likely donor" to "string of digits," and then it bills per send. Political fundraising now shares infrastructure with scams: spray, pray, don't expect most people to answer. The only difference is that the scam's收款方 doesn't file with the FEC.
Watch the "800 percent match," the "900 percent match." House Majority PAC never received a dime from John that the author could find, and kept offering to multiply contributions by as much as 900 percent. That's not a promise. It's a pricing technique—manufacture urgency, let the mark believe his $6 has been multiplied ninefold. Chapter 3 describes how the romance narrative bound diamonds to marriage. This is the miniature version: a small transfer packaged as a participation certificate in "democracy's survival."
And it happens at bedtime. The author is precise: his kid is nodding off, ding, Nancy Pelosi says donate $5 now or American democracy basically collapses. Minutes later, Cory Booker. Then James Talarico.
No one bleeds here, but something is being shaved off: attention. 725 texts, 812 numbers, 1,735 messages since March 2022—all of it consuming one specific person's time after work, while driving, while cooking, while putting a child to sleep. That time could have belonged to something else. The gap is the violence, even when the perpetrator is an autogenerated string.
Kurucz says wrong numbers fall out of circulation because recipients don't engage. What he leaves out: the ones who don't engage get filtered as noise, and the system keeps running, because someone always clicks. The author's number is still on the list. The DSCC now texts him more often than his mother.
His last line: replying "eat shit" another hundred times won't change that. Correct. The recipient of these texts was never him.