Nine problems, nine answers — what tokenization actually fixes, mapped one by one
The opening article of this series diagnosed nine dysfunctions in how real-world assets move, and collapsed them into one cause. The previous article imagined the world where that cause is removed. This article does the unglamorous work in between: each of the nine, one at a time, matched to the specific property of a shared, enforceable asset record that removes it — with a public precedent showing that property already working, and an honest note on what each answer does not fix. Not "tokenization fixes it." Nine problems, nine answers, each one checkable.
Every dysfunction in this series comes from institutions keeping their own copies of the same asset and paying, daily, to agree with each other. Give each asset one live record that enforces its own rules and the agreeing doesn't get faster — it stops being produced. That single property, read nine ways, answers all nine problems. Two get faster. Six become impossible to express. The last dies as the bill for the others.
- Every answer in this article takes the same form: name the property of the record that removes the dysfunction, point to that property running somewhere today, and say plainly what remains unfixed. An answer that cannot survive all three steps is a slogan, and there are no slogans in the table below.
- The answers split into the two kinds the previous article distinguished. Two of the nine are velocity answers — a much faster version of the old system could, in principle, deliver them, and incumbents are already partly capturing them. Six are structural answers — they require different institutions to read and enforce the same asset state, which no speed-up of separate ledgers can produce. The eighth (leakage) inherits from the others.
- The nine answers are not nine technologies. They are one property applied nine times — a shared, live, verifiable record whose rules are enforced by the record itself — which is exactly what the diagnosis predicted: one problem wearing nine costumes should need only one cure, worn nine ways.
- Every answer is also a revenue event for somebody in the current chain. The map of who loses which line is the honest map of where adoption will be fast, slow, and fought.
- The answers do not abolish risk, institutions, or judgment — and encoding rules into assets creates one genuinely new problem, hard enough that the next article is entirely about it.
The discipline, restated
"Tokenization solves this" is the least useful sentence in the industry. It names a technology instead of a mechanism, so it cannot be checked, and claims that cannot be checked accumulate until serious readers stop listening. So this article holds itself to the rule set by the previous one. For each dysfunction we ask: which property of a shared, enforceable record removes this — and could a very fast version of the old system deliver the same result? Where the answer to the second question is yes, we say so, and call it a velocity gain. Where it is no, the answer is structural, and those are the six that matter most — because they are the ones an incumbent modernisation programme, however well funded, cannot reach.
| # | Dysfunction | The property that answers it | Which prize |
|---|---|---|---|
| 1 | Markets that sleep | Transfer rules enforced by the record itself — a change of hands no longer needs a venue to be open | Velocity |
| 2 | Settlement that lags reality | Atomic delivery-versus-payment — both legs are state changes in one transaction; either both happen or neither | Velocity |
| 3 | State that doesn't propagate | Nothing to propagate — downstream holds a reference to the object, not a copy of it | Structural |
| 4 | A long chain of thin roles | Roles that manufacture agreement lose their product; roles that carry responsibility remain | Structural |
| 5 | No common system of record | The record is the answer — this was the root; the cure is definitional | Structural |
| 6 | Double-pledging | Encumbrance as enforced state — a second pledge cannot be created on the record, rather than detected after | Structural |
| 7 | Forged provenance | Attestation bound to the record at issuance — the certificate and the asset cannot come apart | Structural |
| 8 | Leakage | Boundary taxes die with the boundaries — fees for work survive, tolls for agreement do not | Inherits the others |
| 9 | Frozen wealth | Verification that travels — verify once at the record, and the product stack above the asset becomes economical | Structural |
Now each one, in turn — problem, property, precedent, and what remains.
1 · Markets that sleep → transfer that doesn't need the market open
The problem, from Article 1: assets trade in the office hours of their home venue while capital and risk are global and continuous. Alice could not act on the news that repriced her fund; Lena met a margin call in cash because two custodians' books were never open at once.
The property that answers it: a venue's opening hours exist to guarantee that transfers happen under supervision — that the seller owns what it sells, the buyer may hold what it buys, and the exchange happens in an orderly way. When those checks are rules the record itself enforces, a transfer no longer needs the venue's staff, batch cycle, or bell. It needs only two permitted parties and the object, at any hour. Tokenized money-market funds already transfer between eligible holders around the clock; in a widely reported industry demonstration, collateral was mobilised across financial hubs in real time, outside any settlement window — the mechanism of Lena's re-run Thursday.
What remains unfixed: 24/7 transfer is not 24/7 liquidity. Market-makers, pricing desks, and buyers still sleep, and a thin market at 03:00 is thin whatever ledger it runs on. What disappears is the hard wall — the hours in which a transfer is impossible rather than merely quiet. Turning always-open transfer into always-deep markets is a commercial project, not a technical one. And because a fast version of the old world can extend its hours by ordinary means, this is a velocity answer: real, valuable, and not the deep prize.
2 · Settlement that lags reality → atomic delivery-versus-payment
The problem: ownership and payment move separately, so risk lives in the gap and cash in transit earns yield for whoever holds it. The industry's proudest recent reform — compressing the standard cycle from two days to one — took years of industry-wide coordination, the final step of a three-decade crawl from T+3.
The property that answers it: when the security and the money are both records on rails a single transaction can touch, delivery-versus-payment stops being a promise between institutions and becomes a constraint of the transaction — either both legs move or neither does. There is no gap for settlement risk to live in and no window for float to accrue in. This is among the most mature properties in the field: the operator of the US market's central settlement infrastructure processed its first production tokenized-settlement trades in 2026, and tokenized repo between major institutions already settles same-day, at cumulative volumes in the trillions of dollars.
What remains unfixed: atomic gross settlement is not free. Netting — the old system's genuine invention — lets a day's trades cancel against each other so that only balances move; settle every trade instantly and gross, and liquidity must be present at the moment of each trade. Real treasury design work lives there, and serious implementations blend atomic settlement with netting windows by choice rather than by limitation. This too is a velocity answer in principle — T+0 was always possible by heroic conventional means — but note what the diagnosis observed: for part of the chain, delay is the product. The gap is a revenue line, and closing it is a decision somebody has to make against their own income statement.
Run down the table and price each answer from the incumbents' side: answer 2 deletes float income and funding spreads; answer 4 deletes re-verification fees; answer 5 deletes the reconciliation cost line that justifies entire operations divisions; answer 8 is the sum of the others. Every answer is a revenue event for somebody currently in the chain. That map — not technical readiness — is the honest predictor of adoption speed. It is why the most mature deployments so far are led either by infrastructure operators who sit above the fee pool (depositories, settlement utilities) or by principals who sit outside it (asset managers, treasurers, and the banks whose own capital is what the gap traps). The resistance is rational. So is the route around it.
3 · State that doesn't propagate → nothing to propagate
The problem: a lien, a default, a corporate action, a regulatory flag changes the underlying — and downstream records update late or never, so instruments trade on stale truth. Kwame's cargo was unreadable to every lender after the first.
The property that answers it: propagation is only a problem because downstream participants hold copies. When they hold a reference — when the derivative, the loan, and the custody statement all read the same live object — there is nothing to propagate. A lien written at the record is, in the same instant, the truth for every permitted reader. Digital bonds already service coupons and enforce transfer restrictions in the instrument itself; the state is not reported to holders, it is read by them.
What remains unfixed — and what opens up: this property cannot be reached by speeding up the old system, because no message speed makes separate books into the same book; it is the first structural answer, and it is the hinge to composability. The loan that reprices as its collateral's risk verifiably falls — the previous article's second Lego — is exactly this property used forward: a contract that reads state instead of waiting to be told. The honest limit is that the record's state and the world's state are not the same thing. Who attests that the metal is really in the warehouse, and what happens when attestation is wrong or stale, is a real question — large enough that the eighth article in this series is devoted to it.
4 · A long chain of thin roles → roles collapse to the ones that carry responsibility
The problem: broker to exchange to clearing house to depository to custodian to sub-custodian — six institutions, five reconciliation boundaries, each hop a fee, a delay, and a failure point, most of it manufacturing agreement rather than value.
The property that answers it: the chain is long because no two links share a record, so each exists partly to vouch for the others. Give every permitted participant the same object to read and the vouching work loses its reason to exist. The chain does not shorten because a regulator decrees it; it shortens because each link is now priced against the alternative of reading the object directly — and re-verification cannot survive that comparison. Production digital-bond issuances already run with markedly fewer parties than their conventional twins, with registrar and paying-agent workflows automated in the instrument — the speed of issuance is the velocity part of that story; the shortened chain is the structural part.
What remains unfixed: the roles that carry responsibility rather than agreement — legal finality, custody of the physical thing, standing behind the money, answering to a supervisor — do not disappear, and should not. The deepest version of this answer is not disintermediation but transformation: somebody must still operate, govern, and be accountable for the shared record itself. What that institution looks like — a depository whose product is state rather than reconciliation — is where this series is ultimately heading.
5 · No common system of record → the record is the answer
This one needs the least argument, because it is the root the other eight grow from, and the cure is definitional: one live record per asset, read directly by every permitted participant, in place of many private copies kept in agreement by an industry built for the purpose. Shared does not mean public — the live systems are permissioned, with eligibility and visibility rules enforced in the object and supervisors among the readers. Reconciliation does not get faster in this world. It stops being produced, because disagreement between copies requires copies.
The nine answers are not nine technologies to be procured nine ways. They are one property — shared, enforceable state — read nine times from nine angles. Which is exactly what a correct diagnosis predicted the cure would look like.
6 · Double-pledging → encumbrance the object enforces
The problem: the same collateral pledged to several lenders because no one sees the whole picture — Qingdao's duplicated warehouse receipts, billions raised against metal counted several times over, and the defensive discount that honest paper has carried ever since, because honest and duplicated paper look alike.
The property that answers it: a pledge stops being a document about the asset and becomes a state of the asset. A record whose rules say "these tonnes may carry one first-ranking pledge" will not accept a second — not because an auditor would catch it, but because the state transition does not exist to be made. This moves the defence from detection to construction: the fraud is not found faster, it is not expressible on the record. Collateral tokens moving between major institutions already carry encumbrance status in the object, and live trade-finance networks exist precisely to authenticate documents against duplication.
What remains unfixed: the record enforces uniqueness only over what it covers. An asset pledged once inside the system and once outside it — against a paper receipt in a jurisdiction that still honours paper — recreates the seam. The answer's strength is therefore proportional to its coverage, which is why the interesting battles in commodity finance are about which records become the record, not about the cryptography. Partial coverage still pays: every tonne verifiably inside the system is a tonne the discount no longer applies to.
7 · Forged provenance → the certificate that cannot come apart from the asset
The problem: forged receipts, fake certificates, title fraud — paper trusted to stand in for the asset, and nothing in the system forcing the two to stay true to each other. Kwame's data pack, re-verified by every counterparty; Fatima's gold, unverifiable by anyone.
The property that answers it: in the old world, a certificate is a claim about an asset, issued once and then free to drift from the truth. In the new one, the attestation — the assay, the survey, the audit — is performed by an accountable party and bound into the asset's record at issuance, and every subsequent state change appends to the same object. The certificate cannot come apart from the asset because the certificate is the asset's record. Provenance stops being a stack of documents to re-examine and becomes a history to read.
What remains unfixed: the record is only as true as its first entry. A false assay creates a well-formed lie that then travels with all the fluency of the truth — arguably more dangerously, because readers have been taught to trust the record. Garbage-in is not a footnote to this answer; it means the verification effort that used to be spread across every counterparty concentrates at the issuance boundary, and the attestor's accountability becomes the security perimeter of the entire system. Getting that boundary right — who may attest, how attestation decays, what physical state even means as a recorded fact — is the subject of the eighth article.
8 · Leakage → the tax dies with the boundary
The problem: value bleeding to fees, spreads, float, fraud, and manual error at every ledger boundary — none of it dramatic, all of it constant, a standing tax on moving an asset at all.
The property that answers it: leakage was never a tenth dysfunction; it was the other eight, invoiced. Float dies with the settlement gap (answer 2). Fraud loss dies with enforced encumbrance and bound attestation (answers 6 and 7). Manual keying error dies with the reconciliation it existed to serve (answer 5). The verification premium inside spreads dies as reading the object replaces re-proving it (answers 3 and 9). What survives — deliberately — is every fee that pays for real work: custody of real things, credit judgment, market-making, supervision, the attestor's liability. As the previous article put it about Priya's payment: the fees do not go to zero; the four institutions' worth of re-verification does.
What remains unfixed: precision about the boundary between toll and fee is commercially explosive, because every incumbent will describe its tolls as fees. A useful test: would the charge survive if every participant could read the object directly? If yes, it is work. If no, it is agreement-manufacture, and its days are numbered in proportion to coverage.
9 · Frozen wealth → verification that travels
The problem: the quietest dysfunction and the largest. In deep markets a real asset is the base of a stack — property carries mortgages, securitised income, portfolio credit. In most of the world the stack is missing: wealth exists, and the financial system that would let it work does not. Fatima's thirty years of savings, doing nothing, because proving what they are costs more than any single use of them justifies.
The property that answers it: the stack is missing because every layer of it must re-verify the base, and for most assets in most places that cost sinks the product before it launches. Verification that is performed once, at the record, and then inherited by every subsequent participant changes the economics of the entire stack at a stroke: the lender, the insurer, the fund assembler, and the fractional buyer all read the same verified object, and none of them pays the proving cost again. This is the answer the velocity lens cannot see at all — no faster version of re-verification-per-participant reaches it — and it is where the previous article's composability argument lands on the ground: the warehouse fund, the gold that becomes the base of a financial life rather than the end of it. Regulated gold-loan markets already prove verified household collateral is bankable at national scale; a tokenized property programme in production has shown fractional real-asset demand clearing in minutes to hours.
What remains unfixed: a readable asset still needs a legal system that recognises what the record says, buyers who want it, and prices that tell the truth. Records do not conjure demand, and some wealth is frozen for reasons no ledger can touch — disputed title, absent law, assets that should not be financialised at all. The claim is narrower and stronger: where demand exists and law permits, the proving cost stops being the reason the product doesn't.
Tally the table: two velocity answers, six structural, one inherited. This ratio is the strategic content of the whole mapping. An incumbent modernisation programme — faster messaging, T+1, better custody portals — captures most of answers 1 and 2 and touches nothing else, because answers 3 through 9 all require different institutions to read and enforce the same object, and modernising separate ledgers, however brilliantly, produces faster separate ledgers. This is why "we already settle T+1" is not an argument against tokenization. It is a concession that the two velocity answers are being pursued by other means — while the six structural ones, where the durable value sits, are not being pursued at all.
What the nine answers do not answer
Three limits, carried forward from the previous article because they bear repeating at the moment of maximum tidiness. The answers do not abolish risk: a readable cargo can still sink, a verified warehouse can still stand empty, and readability tells you what you own, not whether you should own it. They do not abolish institutions: every answer above quietly relies on someone accountable — an attestor, a custodian, an operator of the record, a supervisor reading it. And they do not run themselves: rules enforced by the record are rules somebody wrote, and rules that enforce themselves perfectly also enforce their mistakes perfectly.
That last point is not a caveat. It is the door to the series' second half. An asset that carries its rules inside it faces a question paper certificates never had to answer: when the world's rules change — a sanction, a court order, a new regulation — should the asset change with them? Who may reach into a live instrument and rewrite what it permits? The next article takes that question seriously, because the honest case for enforceable state has to include the day the state itself is what needs to change.
- How real-world assets move today — the ecosystem problem
- Five people, five assets, one broken system — the problem made real
- Imagine the other world — financial Legos and the positions nobody can take today
- Nine problems, nine answers — what tokenization actually fixes (you are here)
- Rules change. Should the asset? — the new problem tokenization creates
- The Universal Asset State Machine — Decibel Labs' answer
- One machine, five assets — the state machine tested against real asset classes
- Three planes — compliance, physical, economic
- The Universal Asset Token stack — making state enforceable