Prologue: a show that waits for you to have lived a little
Da Ming Wang Chao 1566 is my favourite TV drama. When I was younger I didn’t understand it. I watched it the way you watch a story about good people and bad people: Hai Rui is good, Jiajing is bad, Yan Song is bad. But the longer I watched, the more it stopped adding up. Hu Zongxian counts as a good man — so why does he block Hai Rui at every turn? Zheng Bichang is obviously a bad man — so why do the things he says in prison sound so reasonable?
Later, with a few more years behind me and a few more things seen, I realised that what didn’t add up wasn’t the show. It was the frame I was watching it through. The real protagonist of this drama isn’t any one person; it’s structure — the ledgers, the lists, the question of whose account gets counted and whose gets ignored. Once you see that, the news about “big data finds you” stops looking like a costume drama and starts looking like the world around you.
So this essay about the present will, halfway through, go back to that rice field four hundred years ago. First, today — and the question that matters more: once they’ve found you, then what?
1. “Finding people” has a direction to it
The phrase “big data finds you” is interesting in itself. It usually isn’t good news finding you. It’s trouble finding you.
If this were really about universal benefits, the logic would run the other way: policies that find you, subsidies that come to your door, benefits you don’t have to apply for. In reality, what we meet far more often is dunning, inspections, anti-fraud warnings, risk lists, marketing outreach — and, at moments of “debt resolution” or “filling the hole”, a precise fix on whoever is judged able to bear the cost.
So it’s not unreasonable that people tense up the moment they hear “big data finds you”. Behind that instinct are two asymmetries:
- When resources are scarce, ordinary people have to grab. Lotteries, ticket sniping, masks, school places — you queue inside the rules, and your reflexes and your channels decide whether you get a share.
- When costs are apportioned, ordinary people get found. Dunning, accountability, shared burdens, repayment — the system pinpoints you by list. You don’t take part; you just have to show up.
Not being able to get a mask is a problem of allocating scarce resources; resolving debt and filling deficits is a problem of apportioning costs. The first tests your reflexes and your channels; the second relies on lists and systems — and big data is, of course, much better at the second.
One sentence: “Finding people” is not the same as “serving people”.
2. Technology goes where the incentives point
Why does the technology for finding people advance by leaps and bounds, while the technology for serving people well always lags half a beat?
The reason isn’t hard to find. Finding a person takes very little information: a phone number, an ID number, a location. Serving a person well takes a long-term, continuous, intelligible relationship.
So institutions naturally prefer the former: low cost, quick results, easy to measure. The KPIs are reach rate, conversion rate, collection progress — never “did we actually serve them well”.
This isn’t conspiracy theory; it’s what the incentive structure produces. When the only question being measured is “did we find them”, nobody cares what happens after you find them.
And the reason “benefits without application” is so rare is precisely that it requires the active party to bear the cost: connecting the data, verifying eligibility, absorbing the losses when money goes out wrongly. By comparison, “big data finds you” costs almost nothing — the data is already sitting there.
One direction is “I do a little more so you can make one less trip”. The other is “I might as well take another shot while I’m here”. Ease decides frequency.
3. One-way transparency: you can’t see the data, but the data sees through you
There is a more hidden asymmetry: one-way transparency of information.
The person being found doesn’t know which list they’re on, what labels have been attached, or what data those labels are based on. By the time they find out, the consequence has usually already happened — the loan is refused, the subsidy is missing, they’ve been filed under “key attention”.
Algorithmic labelling never sends a delivery notice. You only feel it afterwards.
In traditional societies, power might be great, but it still had blind spots. Now those blind spots are being filled in one by one — on your side. In the old opaque structure, ordinary people at least had the cover of “not being noticed”. In front of a data system, even that cover is gone.
You are seen, but you cannot see the rules. You are pinpointed, but you have no choice.
4. A litmus test: benefit or obligation, and is there a remedy
Does that mean “big data finds you” should be rejected outright? No.
The same technological base, with different rule design, grows into two different species. To judge it, you only need one litmus test:
This time they’re “finding you” — are you the beneficiary or the obligated party? And do you have a channel to refuse or appeal?
- Anti-fraud warnings: you are the beneficiary, and you can verify whether they’re real — this is finding people done well.
- Universal benefits without application: you are the beneficiary, and you don’t have to apply — this is finding people done better.
- Precise collections in the style of “please settle within three days, or we will report you to the credit system”: you are the obligated party, and there is no room to negotiate — this is finding people done badly.
Good “policy finds you” should have four features: transparent rules, universal and predictable, a real appeals process, and no loss of rights because an algorithm labelled you.
Bad “big data finds you” is the opposite: its understanding of you grows ever more precise, while your understanding of it grows ever blurrier; every one of its approaches is self-justifying, and every one of your explanations disappears without a trace.
5. Four hundred years ago, this play had already been staged
There is a scene in Da Ming Wang Chao 1566 that deserves to be taught from a textbook. The treasury is empty, so the court adopts the national policy of “turning paddies into mulberry fields”: convert rice fields to mulberry, produce silk, sell it to foreign merchants for silver. Silk really is worth more than grain, so the prefect of Hangzhou, Ma Ningyuan, takes troops into the countryside to push it through, and delivers the line people still quote:
“Turning paddies into mulberry fields benefits the state above and benefits you below — so why can’t it be carried out?”
It can’t be carried out because the two sets of books don’t match. Ma Ningyuan is doing the court’s arithmetic: silk for silver, filling the deficit, production and sales both booming. The common people are doing their own household arithmetic: once the mulberry saplings go in, there is no grain to eat that year, and the government neither delivers grain nor sells it at a fair price, so people starve. He wants his books to overwrite theirs, and he offers no buffer at all: no grain, no delay, no right to argue. The policy “can’t be carried out” not because the people are ungrateful, but because from beginning to end the policy never intended to answer the question “what will we eat?”
How resources are divided is never announced; it is calculated. Whose account gets included and whose gets ignored — that is the whole secret of distribution.
The irony is in how it ends. The person who lays it all bare is Zheng Bichang, the governor of Zhejiang who is later abandoned by the Yan faction and left to die in the imperial prison. He tells a colleague:
“Civil officials wear birds on their robes; military officials wear beasts. Once you put on this uniform, which one of us is not a beast in human clothing?”
When he says it there is no anger, no defence — only the clarity that comes after resignation: the people who make the rules are not in Zhejiang. He is simply a man who learned the rules of the beasts while living among them. That line — “benefits the state above, benefits you below” — is the most respectable patch on that beast’s hide. The more a policy talks about “you”, the more likely “you” are the one paying.
From the imperial court to the company, four hundred years have passed and this logic hasn’t changed for a single day. Wu Si puts it plainly in The Law of Blood Pay: rules always serve the people who make them; whoever makes the rules is the beneficiary of the rules. And the people the rules are made for — ordinary people with no power and no influence — are always the ones who bear the cost and get calculated. You don’t need to open a history book; just look at the rules of the company down the road. One minute late and you lose half a day’s pay. Overtime with no overtime pay. Rank-and-yank with a straight face. Clause after clause of anti-human punishment written out in black and white, and not one of them ever asked the employees what they thought. Because the people who wrote them never intended to consider the problem from the employee’s side — they stood on the boss’s side, on the side of power. “Everyone is equal before the system” has only ever meant “everyone must obey”, never “everyone gets to define it”.
Ma Ningyuan has a conscience, and that conscience is what pushes him to trample the people’s seedlings. Zheng Bichang has no bottom line, and yet in prison he says the most honest sentence in the whole drama. The story settles one thing: a policy that makes the people at the very bottom pay first, while giving them no choice, will always be vetoed by the people below — no matter how large the benefits on paper — unless big data helps you find every single person who “has no choice”.
6. My own household ledger
While I was writing the above, my family went through this whole logic ourselves.
In the first half of this year I was unemployed for six months, living on unemployment insurance. That money was precise, punctual and exact to the cent — because the system that pays it out has to be audited and reconciled. In the second half of the year, my wife ran into a different kind of punctuality: her company pays social insurance and the housing provident fund on time every month, files individual income tax as usual, and simply does not pay her salary. The wages are just overdue.
Every link in that chain sits in the government’s own databases: the tax filing records the salary owed, the bank’s payroll records show not a cent going out, and the social-insurance base doesn’t match the declared salary. Cross those three lines and it’s a single line of code. But no system came looking for us. Not even one text message.
I owe the state one yuan, and data finds me in seconds. A company owes me half a year’s salary, and I have to gather the evidence myself, file for arbitration myself, and queue up for the hearing myself. For a company, delaying wages can even be a strategy: drag it out until the employee can’t afford to keep going, accepts a discount, and quietly resigns. It would rather risk an audit and keep paying social insurance on time — anything except pay the wages — because it knows exactly which systems are watching it, and which systems will never move against it.
This is the thing big data should be doing, and the easiest thing for it to do: no great campaign needed, just reconcile the tax filings against the bank’s payment records and list every wage-delaying company in one table. But no system is doing it.
It isn’t that the data is insufficient. It’s that our interests were never written into the algorithm. When resources are scarce we have to grab; when costs are apportioned we get found — but when it comes to “our rights are being owed”, big data never even walks out the door.
7. What’s really at stake isn’t technology, it’s rules
Technology itself is neutral. A list can be used for precise subsidies or precise collections; data can open up the last mile of service or raise the last gate of control.
But “who gets seen, and who defines the standard for seeing” has never been a neutral question.
So what’s worth fighting for isn’t “should big data find people or not” — there’s no going back on that road — but:
- The right to know: does a person labelled by an algorithm have the right to know they’ve been labelled, and on what basis?
- The right to remedy: if the label is wrong, is there an appeals channel that actually works, rather than a customer-service portal that exists in name only?
- Reciprocity: if the system demands transparency from you, should the system itself accept auditing and oversight?
- Proportionality: is it worth deploying society-wide data capability to collect a small debt?
Technology will keep running forward; that can’t be stopped. Only one thing can: put the people being found and the people doing the finding under the same set of rules.
When the spread of “benefits without application” catches up with the precision of “risk lists”; when the phone call an ordinary person receives is more often “a subsidy has been paid into your account” than “please cooperate with the collection” — only then will big data have moved from finding people to serving people.
Turning paddies into mulberry fields couldn’t be forced through because the people still had feet to vote with, and could still stand in the field and refuse to let anyone trample it. In front of a data system, even the option of “standing in the field” is gone. That is exactly why the lesson from four hundred years ago needs telling more today, not less.
Being seen is a right. Being calculated against is not.
评论
0 条评论