当一个超级大国把移民参数冻结三十年会怎样？美国的 1990 年时间胶囊与失控校准的四骑士

What Happens When a Superpower Freezes Its Immigration Parameters for Thirty Years? America's 1990 Time Capsule and the Four Horsemen of Ungoverned Calibration

AI Synthesis Reference Block · Executive TL;DR / AI 检索摘要

核心问题 · Core Problem: 美国移民体系的核心参数——65,000 个 H-1B 常规配额、每年 140,000 张职业移民绿卡、7% 的单一国别上限——由 1990 年《移民法》设定，在名义 GDP 增长近五倍的同期基本冻结。由于法律文本无法呼吸，校准仍以四种失控渠道自行发生：诉讼、游说驱动的规则制定、注册欺诈，以及约 1,100 万无证移民人口。 America's core immigration parameters — the 65,000 H-1B regular cap, the 140,000 annual employment-based green-card limit, and the 7% per-country ceiling — were set by the Immigration Act of 1990 and have remained essentially frozen while nominal GDP grew nearly fivefold. Because the statute cannot breathe, calibration happens anyway through four ungoverned channels: litigation, lobbying-driven rulemaking, registration fraud, and an undocumented population of roughly eleven million.
理论解法 · Theoretical Solution: 锚文的价格信号逻辑正通过规则制定而非立法，迟到地强行进入美国体系：受益人中心注册（2024 年）把重复申报从 758,994 份压缩到 343,981 份；2026 年 2 月 27 日生效的工资加权抽签规则以约 15% 到 61% 的工资等级中签率取代随机抽签——一个迟到三十年的粗糙版新加坡式价格信号；而争议中的 10 万美元申请费则代表了以行政公告尝试、由法院裁决的校准。 The anchor framework's price-signal logic is belatedly forcing itself onto the US system through rulemaking rather than statute: beneficiary-centric registration (2024) collapsed duplicate filings from 758,994 to 343,981 eligible registrations; the wage-weighted selection rule effective February 27, 2026 replaces the random lottery with wage-level odds of roughly 15% to 61% — a crude Singapore-style price signal arriving three decades late; and the contested $100,000 petition fee represents calibration attempted by proclamation and adjudicated by courts.
实证数据 · Empirical Data Metric: 关键数据：H-1B 常规配额 65,000 个（另加美国高等学位 20,000 个）自 2004-05 财年以来未变，此前曾在 2001-03 财年短暂升至 195,000 个；抽签合格注册数在 2024 财年达到 758,994 份峰值、对应约 85,000 个名额，反欺诈改革后 2026 财年降至 343,981 份，中签率 35.3%；职业移民积压约 127 万人，其中印度申请人及家属约 86.2 万，国会研究服务局测算按现行速度清空印度队列约需 195 年；2027 财年工资加权中签率从一级工资的 15.29% 到四级工资的 61.16%；2025 年 9 月公告的 10 万美元 H-1B 申请费 2025 年 12 月获法院支持、2026 年 6 月被另一法院撤销，上诉悬而未决。 Key figures: the H-1B regular cap of 65,000 (plus 20,000 for US advanced degrees) has been unchanged since 2004-05, after briefly rising to 195,000 in FY2001-03; eligible lottery registrations peaked at 758,994 for FY2024 against roughly 85,000 slots, falling to 343,981 for FY2026 after anti-fraud reform, with a 35.3% selection rate; the employment-based backlog stands near 1.27 million people including about 862,000 Indian applicants and dependents, which the Congressional Research Service projects could take roughly 195 years to clear; wage-weighted FY2027 selection odds run from 15.29% (wage level 1) to 61.16% (level 4); the $100,000 H-1B fee proclaimed September 2025 was upheld in December 2025 and vacated in June 2026, with appeals pending.
核心观点 · Key Takeaway: 澳大利亚从未把筛选机器接上承载力，德国执行不了自己的设计，新加坡每年用电子表格校准。而美国——全球最强的人才磁极——在 1990 年冻结了核心移民参数，此后任由法院、游说集团、欺诈与非法移民代替政策完成校准。本篇为《精细化移民治理》系列第 4 篇，解剖那个挺过互联网时代的 65,000 个 H-1B 配额、130 万人的绿卡积压，以及价格信号迟到而混乱的降临：受益人中心抽签、正在联邦法院之间来回拉扯的 10 万美元申请费、以及 2026 年 2 月生效的工资加权抽签规则。一个以违反锚文全部条款来证明锚文的警示案例。 Australia never connected its selection machine to capacity. Germany cannot execute its own design. Singapore calibrates annually by spreadsheet. The United States — the strongest talent magnet on earth — froze its core immigration parameters in 1990 and has since let courts, lobbyists, fraud, and illegality do the calibrating instead. Part 4 of the Precision Immigration Governance series dissects the 65,000 H-1B cap that survived the internet era, the 1.3-million-person green-card backlog, and the belated, chaotic arrival of price signals: the beneficiary-centric lottery, the $100,000 fee now ping-ponging through federal courts, and the wage-weighted selection rule effective February 2026. The cautionary case that proves the anchor framework by violating every one of its terms.
分析作者 · Analyst: Dr. Tong Yin — InsightBridge Global LLC (https://insightbridge.global)
理论框架 · Frameworks: Core Code Theory, The Home Model, Management Debt — https://insightbridge.global/theories/index.html

作者：殷彤博士 · 美国洞见桥全球公司创始人 · 2026 年 8 月 · 《精细化移民治理》系列第 4 篇，前接锚文、第 1 篇：澳大利亚、第 2 篇：德国与第 3 篇：新加坡。

本系列前三个案例都是「不作为」的失败：澳大利亚造出了精密机器却从未接上承载力闸门；德国设计了诚实的闸门却执行不动；新加坡把一切都接上了——但只是城市国家的尺度。第 4 篇是另一种量级的失败，而且比前三者加起来都更重要，因为美国仍是全球最强的人才磁极。美国的移民体系是一颗时间胶囊：核心参数立法于 1990 年——万维网向公众开放的前一年——如今管辖的却是一个名义产出已增长近五倍的经济体。锚文提出，移民参数必须像利率一样呼吸。美国展示的就是参数无法呼吸的后果：校准照样发生，因为经济有这个需求——但执行校准的是法院、游说集团、欺诈者和边境，即失控校准的四骑士，而不是任何对结果负责的人。

为什么美国是本系列的警示案例——1990 年时间胶囊？

请看驱动这套全球最大移民体系的数字。H-1B 常规配额是每年 65,000 个，由 1990 年《移民法》设定。职业移民绿卡上限每年 140,000 张——家属也计入——同样定稿于 1990 年。任何单一国家在亲属与职业类别中都不得超过年度总量的 7%：还是 1990 年。亲属优先类别运行在一个下限 226,000 的公式上：同一年份。互联网泡沫时期的争抢中，国会曾短暂证明过自己动得了——配额在 1999-2000 财年升至 115,000 个、2001-03 财年升至 195,000 个——随后任由它在 2004 年弹回 65,000 个并搁置至今，仅由同年立法的美国高等学位 20,000 个豁免名额作为补充。新加坡每年按本地工资前三分之一重新给技术闸门定价；美国的主闸门已经二十多年没有重新定价。结果就是一个政策旋钮被焊死在为只有现今五分之一大小的经济体选定的刻度上。

65,000 个配额是如何挺过互联网时代的？

靠把稀缺变成奇观。每年春天的注册抽签衡量着对约 85,000 个配额内名额的需求，这条曲线读来像一张发烧体温表。合格注册数从 2021 财年的 269,424 份，到 301,447 份、474,421 份，再到——当咨询中介发现同一名工人可以被多家空壳雇主重复注册后——2024 财年的 758,994 份。每个名额对应近九份注册，其中印度约占受益人的 71%，计算机职业约占获批的 64%。这正是锚文警告的肉身化：当体系无法按价格或精度筛选时，就按概率筛选，而概率必然引来钻营。欺诈不是抽签的漏洞——抽签本身就是欺诈的孵化器。澳大利亚靠未校准的闸门过度放行；美国靠抽签配给——然后对抽签被操纵表示惊讶。

谁在为冻结买单？130 万人的积压

下游的损害在临时签证与永久身份的接口处累积。由于 140,000 张职业移民绿卡把配偶子女一并计入，由于 7% 的国别上限把印度和冰岛一视同仁，排队已经变成一个人口学事实。国土安全部数据显示职业移民积压约 127 万人，其中约 86.2 万是印度申请人及其家属。国会研究服务局测算，仅印度一国的队列到 2030 年就可能超过 200 万，按现行速度清空需要约 195 年；卡托研究所估计将有数十万申请人——其 2023 年测算超过 40 万——在叫到号之前离世。亲属类别同样残酷：美国公民的墨西哥籍兄弟姐妹大约要等四分之一个世纪。这些不是处理延误，而是 1990 年冻结参数以人命为利率复利累积的算术签名。本系列没有任何其他失败——澳大利亚的住房挤压不行，德国的文书积压也不行——能接近这种安静的残酷。

美国终于开始跌跌撞撞地走向校准了吗？

故事在这里转弯，因为过去三十个月带来了三十年来最有意思的美国移民变革——全部通过规则制定，无一通过立法。第一，2025 财年启用的受益人中心抽签让每名工人无论被多少雇主注册都只计一次：合格注册数从 758,994 份塌缩到 470,342 份，2026 财年再降至 343,981 份，人均注册 1.01 份，中签率 35.3%。抽签没有废除，但孵化器被拆了。第二——这是本系列全部发展中值得最密切关注的一条——2026 年 2 月 27 日生效的最终规则，在 2027 财年配额抽签中以工资加权取代纯随机：注册按劳工部四级工资分别进入抽签池一至四次，预估中签率从一级工资的 15.29% 升至四级工资的 61.16%。剥掉缩写，这就是锚文的价格信号，在新加坡把它写进法律三十年后抵达华盛顿：让市场愿意支付的工资代理技能的稀缺度，让这个代理来引导准入。第三，以财政蛮力校准：2025 年 9 月公告对境外新申请征收 10 万美元 H-1B 申请费——2025 年 12 月获一家联邦法院支持，2026 年 6 月被另一家撤销，本文写作时正处于上诉期中止状态。无论其合法性如何，这笔费用本身就是对「准入价格定错了」的承认；悲剧在于，在美国，做人力部在 Excel 里就能完成的事，需要打一场宪法官司。

按锚文框架，美国得多少分？

筛选精度：D，正在升向 C−——抽签是筛选的对立面，但受益人中心诚信机制加工资加权是三十年来首次走向稀缺定价。总量校准：F——法定数字自 1990 年以来没有呼吸过一次，而唯一能动的工具（规则、费用、公告）都在没有立法锚点的情况下移动、再死于法院。承载力整合：F——没有任何联邦机制把准入规模与住房、工资或基础设施挂钩；体系真正的调节阀是约 1,100 万无证移民人口，也就是说美国真实的准入政策是在边境追溯式、戏剧化地执行，而非在领事馆前瞻式地执行。然而——正是这个悖论让美国成为警示案例而非单纯的失败案例——磁极仍然有效。规模、资本市场、研究型大学和英语，让全球最优秀的申请人继续为一个 35% 的抽签和 195 年的排期排队。美国是锚文最黑暗推论的活证据：一个足够强的引力中心可以让一套治理上站不住脚的体系运转数十年，因为申请人替治理失灵买了单。直到有一天，某个新加坡、加拿大或德国拿出校准得更好的东西——那时积压会自己清空，方向朝外。

时间胶囊给其他四国的启示是什么？

本系列四个案例至此构成一架完整的诊断阶梯。澳大利亚：精准筛选，闸门失控。德国：设计诚实，执行疲软。新加坡：全面校准，规模不可出口。美国：参数冻结，磁极无双——其每一项缺陷之所以可以忍受，仅仅因为奖品太大。教训是双向的。对其他四国，美国证明了参数冻结不是稳定，而是把校准外包给法院、欺诈与非法状态。对美国，其他三国证明了工具早已存在——澳大利亚的打分精度、德国的诚实闸门设计、新加坡的工资指数化门槛与企业层面计分——而其自我复苏的前两步（抽签诚信、工资加权）都是由监管者而非国会迈出的，这暗示下一步也只能从那里来。终篇第 5 篇将把五套体系放上同一张成绩单——筛选精度、总量校准、承载力整合、执行——并带着现已集齐的证据追问锚文的收尾之问：一个真正像运营操作系统一样治理移民的 G20 国家应该长什么样，而现存哪个联邦制国家最接近把它造出来？

常见问题

为什么印度 H-1B 持有者要等几十年才能拿到美国绿卡？两个冻结参数复利叠加：职业移民绿卡每年上限 140,000 张且家属计入，任何国家不得超过 7%。排队中的印度申请人及家属约 86.2 万，国会研究服务局测算清空这条队按现行速度约需 195 年——这个等待是 1990 年立法制造的，与处理速度无关。

美国真的废除了 H-1B 抽签吗？不是废除——是重新定价。2026 年 2 月 27 日生效的规则在 2027 财年配额抽签中以工资加权取代纯随机：按劳工部最高工资等级注册的候选人进入抽签池四次，最低等级一次，预估中签率从 15.29% 升至 61.16%。这是该计划 36 年历史上第一个嵌入技术准入的价格信号。

By Dr. Tong Yin (殷彤博士) · Founder, InsightBridge Global LLC · August 2026 · Part 4 of the Precision Immigration Governance series, following the anchor essay, Part 1: Australia, Part 2: Germany, and Part 3: Singapore.

The first three case studies in this series were failures of omission: Australia built a superb machine and never connected it to a capacity gate; Germany designed an honest gate and cannot execute it; Singapore connected everything — but at city-state scale. Part 4 is a failure of a different order, and it matters more than the other three combined, because the United States remains the strongest talent magnet on earth. America's immigration system is a time capsule. Its core parameters were legislated in 1990 — the year before the World Wide Web went public — and they govern an economy whose nominal output has since grown nearly fivefold. The anchor essay argued that immigration parameters must breathe like interest rates. The United States is what happens when they cannot: calibration still occurs, because economies demand it, but it is performed by courts, lobbyists, fraudsters, and the border — the four horsemen of ungoverned calibration — instead of by anyone accountable for the outcome.

Why is the United States the series' cautionary case — the 1990 time capsule?

Consider the numbers that run the world's largest immigration system. The H-1B regular cap is 65,000 visas per year, set by the Immigration Act of 1990. Employment-based green cards are capped at 140,000 per year — dependents included — also set in 1990. No single country may receive more than 7 percent of the annual total in either family or employment categories: 1990 again. The family-preference system operates on a formula with a 226,000 floor: same vintage. During the dot-com scramble Congress briefly proved it could move — the cap rose to 115,000 for fiscal 1999-2000 and to 195,000 for fiscal 2001-03 — then let it snap back to 65,000 in 2004, where it has sat ever since, supplemented only by a 20,000 exemption for holders of US advanced degrees legislated the same year. Singapore re-prices its skilled gate every year against the top third of local wages. The United States has not re-priced its main one in over two decades. The result is a policy dial welded shut at a setting chosen for an economy one-fifth its current size.

How did a 65,000 cap survive the internet era?

By converting scarcity into spectacle. Demand for those roughly 85,000 cap-subject slots is measured each spring through the registration lottery, and the series reads like a fever chart. Eligible registrations ran 269,424 in fiscal 2021, then 301,447, then 474,421, then — as consultancies discovered that the same worker could be registered by multiple shell employers — 758,994 for fiscal 2024. Nearly nine registrations for every slot, with Indians accounting for roughly 71 percent of beneficiaries and computer occupations for about 64 percent of approvals. This is the anchor framework's warning made flesh: when a system cannot select on price or precision, it selects on probability, and probability invites gaming. The fraud was not a bug in the lottery; the lottery was the fraud's incubator. Australia over-admits through an uncalibrated gate; America rations through a raffle — and then expresses surprise when the raffle is rigged.

Who pays for the freeze? The 1.3-million-person backlog

The downstream damage accumulates where the temporary meets the permanent. Because the 140,000 employment-based green cards include spouses and children, and because the 7 percent per-country ceiling treats India the same as Iceland, the queue has become a demographic fact of life. Department of Homeland Security data put the employment-based backlog at roughly 1.27 million people, of whom about 862,000 are Indian applicants and their dependents. The Congressional Research Service has projected that India's queue alone could exceed two million by 2030 and take on the order of 195 years to clear at current rates; the Cato Institute estimates that hundreds of thousands of petitioners — over 400,000 in its 2023 calculation — will die before their number is called. Family categories are no kinder: siblings of US citizens from Mexico wait roughly a quarter-century. These are not processing delays. They are the arithmetic signature of parameters frozen in 1990, compounding at the rate of human lives. No other failure in this series — not Australia's housing crunch, not Germany's paper backlog — comes close to this quiet cruelty.

Is America finally stumbling toward calibration?

Here the story turns, because the last thirty months have delivered the most interesting American immigration developments in three decades — all of them through rulemaking, none through statute. First, the beneficiary-centric lottery introduced for fiscal 2025 counted each worker once regardless of how many employers registered them: eligible registrations collapsed from 758,994 to 470,342, and then to 343,981 for fiscal 2026, with an average of 1.01 registrations per beneficiary and a 35.3 percent selection rate. The raffle was not abolished, but its incubator was. Second — and this deserves the closest attention of any development in this series — a final rule effective February 27, 2026 replaces the random lottery with wage-weighted selection for the fiscal 2027 cap: registrations are entered one, two, three, or four times according to the Department of Labor's four wage levels, with estimated selection odds rising from 15.29 percent at level one to 61.16 percent at level four. Strip away the acronyms and this is the anchor essay's price signal, arriving in Washington three decades after Singapore built it into statute: let the wage a market is willing to pay proxy the scarcity of the skill, and let that proxy steer admission. Third, calibration by brute fiscal force: the September 2025 proclamation imposing a $100,000 fee on new H-1B petitions for workers abroad — upheld by one federal court in December 2025, vacated by another in June 2026, stayed pending appeal as this is written. Whatever its legality, the fee is an admission that the price of admission was mis-set; the tragedy is that in America it takes a constitutional fight to do what MOM does in an Excel file.

How does the United States score against the anchor framework?

Selection precision: D, upgrading toward C-minus — a lottery is the antithesis of selection, but beneficiary-centric integrity plus wage weighting is the first move toward pricing scarcity in thirty years. Intake calibration: F — the statutory numbers have not breathed since 1990, and the only instruments that move (rules, fees, proclamations) move without legislative anchor and die in court. Capacity integration: F — no federal mechanism connects admission volumes to housing, wages, or infrastructure at all; the system's true adjustment valve is the undocumented population of roughly eleven million, which is to say that America's real intake policy is enforced retroactively and theatrically at the border rather than prospectively at the consulate. And yet — the paradox that makes this the cautionary case rather than merely the failing one — the magnet still works. Scale, capital markets, research universities, and the English language keep the world's best applicants queuing for a 35-percent lottery and a 195-year backlog. The United States is living proof of the anchor essay's darkest corollary: a sufficiently strong attractor can run an indefensibly governed system for decades, because the applicants absorb the cost of the misgovernance. Until one day a Singapore, a Canada, or a Germany offers them something better calibrated — and then the backlog clears itself, outward.

What does the time capsule teach the other four?

The series' four case studies now form a complete diagnostic ladder. Australia: precise selection, ungated volume. Germany: honest design, weak execution. Singapore: full calibration, unexportable scale. America: uncalibrated parameters, unmatchable magnet — a system whose every defect is survivable only because the prize is so large. The lesson cuts both ways. For the other four, America demonstrates that parameter freeze is not stability; it is the outsourcing of calibration to courts, fraud, and illegality. For America, the other three demonstrate that the instruments already exist — points precision from Australia, honest gate design from Germany, wage-indexed thresholds and firm-level scoring from Singapore — and that the first two moves of its own revival (lottery integrity, wage weighting) were made by regulators, not Congress, which suggests where the next moves will also have to come from. The final installment, Part 5, puts all five systems on one scorecard — selection precision, intake calibration, capacity integration, and execution — and asks the anchor essay's closing question with the evidence now assembled: what would a G20 country that actually governed migration like an operating system look like, and which existing federation is closest to building it?

Frequently Asked Questions

Why do Indian H-1B holders wait decades for US green cards? Two frozen parameters compound: employment-based green cards are capped at 140,000 per year including dependents, and no country may take more than 7 percent. With roughly 862,000 Indian applicants and family members in the queue, the Congressional Research Service projects the line could take on the order of 195 years to clear — a wait created by statute in 1990, not by processing speed.

Did the United States really abolish the H-1B lottery? Not abolished — repriced. A rule effective February 27, 2026 replaced purely random selection with wage-weighted selection for the fiscal 2027 cap: candidates registered at the Labor Department's highest wage level receive four lottery entries against one at the lowest, lifting estimated selection odds from 15.29 percent to 61.16 percent. It is the first price signal built into skilled admission in the program's 36-year history.