The cost of speaking fell to nothing And nothing fell on everyone But some had channels, some had standing And nothing lands on those alone
For two years this newsletter has worried, in one register or another, about a scarce good. A skilled advocate was expensive. The people who most needed one to navigate a benefits tribunal, a planning objection, an unfair dismissal, a deportation hearing, were exactly the people who could least afford the hours of a person who knew how to make their case. Voice was rationed by wealth, and a great deal of what looks like injustice is really the distribution of access to competent argument.
So there is a version of this week that should be good news outright, and it gets full strength here before anything complicates it. If capable advocacy just became abundant and nearly free, then the thing that was rationed by wealth is about to be available to the people who were priced out. That is not a small claim. It might be the most important thing the technology does.
The trouble is that the same sentence, read from a different chair, describes a catastrophe. So today is a fork, both tines are sharp, and they do not resolve.
The good version
The evidence is real.
Conversational systems raised political knowledge about as effectively as a session of self-directed searching in the tested UK conditions. That is a citizen, not a campaign, using the abundant advocate to understand something they were previously going to guess at. And in a separate study, brief conversations with an AI standing in for a member of a political outgroup corrected people's exaggerated beliefs about that group and left them more willing to speak with a real one. Cheap conversation opened a door that people usually keep shut, the one that leads to talking to someone they are sure they disagree with.
Extend that from politics to the tribunal. An abundant advocate is one that a public defender's office, chronically overloaded, could put in front of every client instead of the fifteen minutes they currently get. It is the patient explainer a small charity could never staff, walking someone through a form at their own pace at two in the morning. It is the tireless drafting help behind a grassroots campaign that has passion and no communications budget. Everywhere that a good argument was rationed by the cost of the person who could make it, abundance is a redistribution, and it flows toward the people who had the least.
The good news comes with its footnotes. Both studies are preprints. The warmth toward the outgroup largely faded within a week. That tells you the effect is real and shallow, a nudge rather than a conversion. And the synthetic-contact result carries a question it cannot answer, which is who authorised the machine to speak for the group it was standing in for, a question that has come up here before and will again. But with all of that granted, the direction is real. Cheap advocacy can widen the room.
The same fact, from the other chair
Now read the same fact from the other chair, because the exact property that makes abundance redistributive also makes it the opposite.
Advocacy became cheap for everyone. Not for the underserved specifically. For everyone, which very much includes the party that already had all the advocacy it wanted and would now like to have a thousand times more. The study's own authors name this directly: the plausible concentration of persuasive capacity among well-resourced deployers, and among the model suppliers who decide what their systems will argue. A drop in the price of a good does not preferentially benefit the poor. It benefits whoever is best positioned to buy at scale, and at scale is precisely where a well-funded institution lives and a public defender's office does not.
So picture the tribunal again. Yes, the claimant now has a capable advocate they could never have afforded. And the department they are up against, the insurer, the landlord's managing agent, the employer's legal team, has the same abundance, deployed by people whose full-time job is to deploy it, integrated into their systems, tuned on every prior case, generating objections faster than the claimant's tireless advocate can generate responses, in a process where the institution also controls the timetable. The floor rose for the claimant. The ceiling rose much further for the other side, and they were already standing on it.
This is the tell that abundance does not choose a side. It lowers the cost of speech for whoever was already positioned to use it, and being positioned to use it, at scale, through owned channels, with staff and integration and control of the venue, is itself a form of wealth that a cheap advocate does not touch. The thing that was scarce, competent argument, got distributed. The things that were never scarce for the powerful, channel, attention, standing, and the staff to run all three, stayed exactly where they were.
What abundance cannot redistribute
Which points at the deeper reason the fork does not resolve, and it recurs all week.
Cheap advocacy redistributes argument. It does not redistribute the three things argument needs to matter, and those were the scarce goods all along. Attention: the abundant advocate the charity now has cannot make anyone read what it produces, and the department it is arguing with has ways of commanding attention that money still buys. Authentication: the campaign bot reaches you because it has your verified phone number through a channel with a right to be there, and the grassroots voice, however eloquent, is shouting into a feed that routes around it, exactly as the influence operation with plenty of output and no breakout discovered. And standing: the quality of being someone whose argument carries a consequence, which is not a function of how well the argument is made and never was.
The Thirteenth Floor asked who receives a new capability as usable infrastructure and who receives only its lagging, degraded form. The abundant advocate is that question again, and the answer is the same discouraging shape. It reaches the penthouse and the stairwell both. It arrives in the penthouse integrated, staffed, wired into the channels that carry it and the venues that must heed it. It arrives in the stairwell as a chat window and the invitation to make your own case, better than before, into an air that was never obliged to listen.
It does not balance out
Power for power, advocate for advocate, and may the best argument win. It sounds fair, it lets the reader put the question down, and the balance was never between the arguments in the first place. Two tireless advocates arguing at machine speed is a fight resolved by everything around the arguments: who owns the channel, who can be authenticated into it, who controls the clock, and who the venue is obliged to take seriously. Abundance made the arguments cheap and left every one of those untouched, which means it handed a real gift to the underserved and a larger one to the people they are underserved by, and calling that a wash is just declining to look at the size of the two gifts.
The fork is open, and its resolution is a choice rather than a property of the technology. Whether cheap advocacy widens the room or floods it depends on the channel, and the channel is owned.
Which is exactly the question The Positions It Will Argue takes up. We have been talking as if everyone gets the same advocate and only their circumstances differ. They do not even get the same advocate. The supplier decided, upstream of every user, which arguments the machine is willing to make and whose objections it will quietly decline to voice, and that turns out to be a sharper power than access ever was.
Companions
- The constructive case: conversational AI and political knowledge, and synthetic contact with a political outgroup.
- Production without reach: OpenAI on PRC-linked influence operations.
- The distributional question this inherits: The Thirteenth Floor.
- The scarcity that moved: The Side That Never Runs Out.
- In the wings: the Culture's Special Circumstances, a vastly capable advocate acting for emancipatory ends while the people it helps keep little standing over the strategy chosen for them.
These notes come out of Sociable Systems, a practice that reads AI-shaped documents the way a hostile reviewer will, before a lender or a court finds the gap. The argument has an operational form: the Interim Protocol sets out four rules for AI use in environmental and social deliverables, covering disclosure at touch-point grain, evidence custody, the phrases no automated screening may settle, and a hostile read before anything ships. Free, and written to be cited or retired once institutional guidance arrives.