A stable system is not necessarily a visible one.

Public markets expose changing expectations continuously. Bonds trade, prices move and losses become visible even before a borrower misses a payment. Private credit operates differently. Loans are negotiated directly, held within funds and valued periodically rather than continuously.

That difference can be useful.

It gives lenders time to work with borrowers, reduces pressure to sell into weak markets and allows capital to remain committed through short-term disruption.

It also changes what observers can see.

A private loan may continue to carry a stable reported value while the borrower’s revenues weaken, refinancing becomes more expensive or the probability of full repayment declines. The risk has not disappeared. Its recognition has been delayed, modelled or transferred into a negotiation that occurs away from public markets.

Private credit risk is therefore not defined only by whether loans default.

It also depends on when deterioration is recognised, where leverage is held and how quickly investors, lenders and regulators can see the complete system.

The central structural question is:

Does private credit appear resilient because its financing is genuinely patient—or because deterioration becomes measurable later?

Erths assessment: Private credit can absorb volatility without eliminating risk. Its resilience depends not only on borrower performance, but on how valuations, leverage, liquidity and interconnections behave when losses can no longer remain private.

1. What private credit actually is

Private credit generally refers to loans made by non-bank investment vehicles directly to companies. The loans are usually negotiated between a small group of lenders and a borrower rather than issued as publicly traded bonds.

The market historically served small and medium-sized companies that could not obtain suitable financing from banks or public markets. It now extends into larger corporate transactions, infrastructure, real estate, asset-backed lending and companies owned by private-equity sponsors.

Investors commonly include pension funds, insurers, endowments, sovereign institutions and wealth-management clients. Asset managers assemble their capital into funds, originate or purchase loans and manage the relationship with the borrower.

Most traditional private-credit funds are closed-ended. Investors commit their money for several years and cannot ordinarily demand it back at short notice. Fund lifecycles commonly align more closely with the maturity of their loans than the funding structures of deposit-taking banks or daily-dealing investment funds.

This is an important source of resilience.

A lender holding long-term loans with long-term capital is less exposed to the immediate withdrawal pressure that can destabilise a deposit-funded bank or an open-ended fund.

Private credit is therefore not simply bank lending conducted with less regulation.

Its funding structure, contractual flexibility and investor base are different. Those differences can make it better suited to some forms of patient and specialised finance.

2. A market whose size is difficult to define

The Financial Stability Board estimated the private-credit market at between $1.5 trillion and $2 trillion at the end of 2024.

Other official estimates are higher. The Bank for International Settlements has placed global private-credit assets under management above $2.5 trillion. The apparent difference reflects varying definitions, reporting boundaries and market coverage rather than necessarily showing that either estimate is wrong.

Private credit does not have one universally applied boundary.

Some estimates include only direct-lending funds. Others include distressed debt, mezzanine finance, business development companies, asset-backed lending or related forms of non-bank corporate credit.

Funds, borrowers and financing structures may also span several jurisdictions.

The difficulty of measuring the market is itself part of the case.

Public markets produce centralised prices, issuance records and trading data. Private credit is distributed across contracts, funds, service providers and regulatory regimes.

Authorities may see particular institutions without seeing every layer of exposure connecting them.

The market can therefore grow faster than the system used to observe it.

That does not make private credit unregulated or unknowable.

It means that the complete picture must be assembled from multiple partial views.

3. Why the market grew

Private credit expanded because it solved genuine financing problems.

A direct lender can negotiate a loan around the specific needs of a company. It can move more quickly than a public bond issue, preserve confidentiality and offer terms that would be difficult to standardise for a widely distributed security.

Borrowers may receive one financing package rather than coordinating several banks or public investors. Private-equity sponsors can obtain financing aligned with an acquisition timetable.

Companies with complex assets, irregular cash flows or limited collateral can receive capital that a traditional lender may be unwilling to provide.

Investors receive access to floating-rate loans, contractual protections and an illiquidity premium. Fund managers may also develop specialised knowledge of particular industries and monitor borrowers more closely than a dispersed group of bondholders.

The market’s expansion was also encouraged by structural changes elsewhere.

Institutional investors searched for higher returns during the long period of low interest rates. Bank regulation and capital requirements changed the economics of lending to more leveraged companies.

Research from the BIS indicates that private credit expanded partly because its relative funding position improved and because it could serve borrowers inadequately supplied by existing banking systems. It also found that individual funds often remained concentrated in a relatively narrow range of industries.

These forces did not merely push risk out of the banking system.

They created a new form of intermediation with its own strengths and vulnerabilities.

4. Stability without continuous prices

The defining visual feature of private credit is the absence of a continuously traded market price.

A public bond can fall sharply in value because investors reassess the borrower, the sector or the economy.

That movement may be uncomfortable, but it provides information. It shows that the market’s assessment has changed.

A private loan is generally valued through periodic appraisal.

The process may use:

  • the borrower’s financial performance;
  • comparable public securities;
  • expected future cash flows;
  • recent transactions;
  • the judgement of the fund manager;
  • an external valuation provider.

This can produce smoother reported returns.

Smoother returns are not automatically artificial.

A lender that does not need to sell an asset today may reasonably value it according to the payments expected over its remaining life rather than the price available in a distressed transaction.

But lower reported volatility and lower economic risk are not the same thing.

Several different values can coexist:

  • the amount the borrower is contractually required to repay;
  • the value produced by the fund’s valuation model;
  • the price another investor would pay today;
  • the amount ultimately recovered if the borrower defaults.

In calm conditions, those values may remain close enough for the distinction to appear unimportant.

Under stress, the gap can become the central issue.

The Bank of England’s private-markets system-wide exploratory scenario identifies infrequent valuation as a potential source of vulnerability. It notes that infrequent valuations can reduce visible return volatility while making risks harder to assess and increasing the possibility of sharp, unexpected repricing.

The system may therefore look stable partly because it updates more slowly.

5. Flexibility or delayed recognition?

Private lenders can respond to a struggling borrower in ways that public markets often cannot.

They may:

  • extend a maturity;
  • amend a covenant;
  • alter an interest schedule;
  • provide additional capital;
  • allow interest to be added to the loan balance rather than paid immediately in cash.

These actions can preserve real economic value.

A viable business may be experiencing a temporary fall in cash flow. Forcing it into insolvency could destroy jobs, customer relationships and productive assets.

A lender with committed capital and detailed knowledge of the company may be better positioned to wait for recovery.

The same flexibility can also postpone the recognition of impairment.

A maturity extension does not by itself show whether the borrower needs more time or cannot repay.

Payment-in-kind interest can preserve cash today while increasing the amount owed tomorrow.

A covenant amendment can remove an unnecessary restriction—or acknowledge that the original protection no longer reflects the borrower’s condition.

The distinction between patient restructuring and deferred loss is rarely visible from outside the transaction.

The Bank of England reported in July 2026 that some borrowers were managing refinancing and cash-flow pressure through amended and extended loans, payment-in-kind structures and other forms of forbearance. It cautioned that these measures might not be sustainable for every borrower over the longer term.

This creates a measurement problem.

The relevant question is not simply how many borrowers have formally defaulted.

It is how much economic deterioration has been absorbed through amendments, extensions, additional leverage and revised valuations before a default becomes necessary.

A low reported default rate can coexist with rising pressure.

The metric remains accurate within its definition.

Its interpretation becomes less reliable.

6. The network behind the loan

Private credit is often described as finance outside the banking system.

That description is incomplete.

Banks may lend directly to the same private-equity-sponsored companies. They may provide subscription lines and net-asset-value facilities to funds, finance business development companies, arrange leveraged loans or supply liquidity to investment vehicles.

Insurers and pension funds provide capital to private-credit funds and may hold private loans directly.

Asset managers can operate private-credit, private-equity and collateralised-loan-obligation businesses within the same group.

A borrower may have obligations across:

  • private loans;
  • public bonds;
  • bank facilities;
  • leases;
  • sponsor financing;
  • structured credit markets.

The risk has not necessarily left the traditional financial system.

Its path has become more complex.

The Financial Stability Board has identified deepening connections between private-credit funds, banks, insurers and private-equity firms, alongside potential vulnerabilities involving leverage, liquidity, concentration and cross-border exposure.

The Bank of England’s current system-wide exploratory scenario includes 46 participating institutions, including banks, pension funds, insurers, endowments, liquid-credit managers and alternative asset managers.

The broad participant group is necessary because no single institution represents or observes the complete system.

The exercise is designed to examine not only losses produced by a severe downturn, but the actions participants might take in response.

A bank may reduce financing to funds.

An institutional investor may slow new commitments.

A private lender may preserve cash for existing borrowers.

A sponsor may inject equity into one company while withholding it from another.

Each action may be rational in isolation.

Taken together, they may reduce the supply of credit more sharply than any one participant intended.

7. Where liquidity enters an illiquid market

Traditional closed-ended private-credit funds do not promise immediate access to investor capital.

This reduces the risk of a classic run.

The market, however, is changing.

Private credit is increasingly being offered through evergreen funds, interval funds and non-traded business development companies that permit periodic withdrawals.

These structures broaden access and can provide useful flexibility to investors.

They also introduce a potential mismatch.

The investor may be able to request cash every quarter, while the fund holds loans that cannot be sold quickly without accepting a discount.

Redemption limits, gates and notice periods are designed to manage that mismatch, but they do not make the underlying loans liquid.

The Bank of England reported elevated redemption requests at several retail-oriented private-credit funds in 2026, with some limiting withdrawals. These funds generally offered periodic redemptions subject to defined caps.

Although the mechanisms largely operated according to their terms, the episode illustrated how concerns about valuations and future access to capital can reinforce one another.

An investor who fears that a future redemption window may be restricted has an incentive to submit a request earlier.

A structure designed to provide limited liquidity can therefore produce defensive behaviour before the assets themselves experience a realised loss.

This does not describe the whole private-credit market.

Most funds remain closed-ended.

It does show how efforts to make private assets more accessible can import vulnerabilities associated with more liquid financial products.

8. What happens in a downturn?

Private credit at its current scale has not been tested through a broad economic downturn in a higher-rate environment.

The FSB has identified this absence of a full stress test as one of the principal uncertainties surrounding the market’s current size and complexity.

A severe stress would not begin with every loan failing simultaneously.

It would develop through several connected pressures:

  1. Corporate revenues weaken.
  2. Floating-rate debt remains expensive or refinancing costs rise.
  3. Highly leveraged borrowers seek amendments or additional capital.
  4. Lenders update valuations at different speeds and with different assumptions.
  5. Investors reduce new commitments or request available withdrawals.
  6. Banks reassess financing provided to funds, sponsors and borrowers.
  7. Fund managers preserve liquidity and become more selective.
  8. Viable companies find that credit is less available even though they have not defaulted.

The financial-stability risk is not only the loss on an individual loan.

It is the possibility that uncertainty about valuations and exposures causes multiple institutions to act defensively at the same time.

When participants cannot distinguish clearly between resilient and impaired assets, they may price for the weaker case.

Financing conditions can tighten beyond the companies that originally generated the concern.

Private credit could dampen that process if long-term capital continues to support viable borrowers.

It could amplify it if leverage, liquidity pressure and uncertainty cause lenders and investors to retreat together.

The answer depends on behaviour, not simply balance-sheet totals.

The Bank of England’s stress exercise is specifically examining whether the combined responses of banks and non-bank institutions could amplify financial stress, disrupt related credit markets and reduce finance available to companies.

9. The role of reporting

Better reporting can improve visibility.

Authorities need information about:

  • borrower leverage;
  • loan amendments;
  • fund-level borrowing;
  • investor redemption rights;
  • sector concentration;
  • valuation practices;
  • connections with banks and insurers.

More consistent data can help identify where risks are accumulating and how stress may be transmitted.

The Bank of England is combining regulatory information with a system-wide scenario because ordinary institution-by-institution supervision cannot fully show what happens when participants respond to one another.

But reporting has limits.

A standardised form can record the terms of a loan.

It cannot determine with certainty whether an extension will preserve value or delay a loss.

A valuation methodology can become more comparable without becoming a continuously tradable price.

A map of interconnections can reveal where exposure sits without predicting how every participant will behave under pressure.

Private credit is built around bespoke agreements and negotiated responses.

Removing all discretion would remove part of what makes the market useful.

The regulatory objective is therefore not to make private credit identical to public credit.

It is to make the complete system visible enough that flexibility does not become indistinguishable from concealment.

This is the central reform constraint:

Greater transparency can improve the measurement of risk without eliminating the uncertainty inherent in illiquid, customised loans.

The Bank of England has indicated that expanded reporting and system-wide analysis should improve its view of private-market risks, while continuing targeted supervisory work on banks and insurers exposed to private credit.

10. The Erths framework

Hidden Instability

Private loans can retain stable reported values while borrower quality, refinancing conditions or expected recoveries deteriorate.

The absence of a daily price can prevent temporary market sentiment from forcing unnecessary adjustment.

It can also delay the point at which accumulated weakness becomes visible.

Measurement Breakdown

Default rates, fund returns and modelled valuations may each be correct according to their definitions while providing an incomplete picture of economic stress.

Amendments, extensions and payment-in-kind interest can change the meaning of continued payment performance.

A loan that has not defaulted is not necessarily a loan whose risk has remained unchanged.

Structural Drift

Private credit developed partly to provide patient, specialised capital to borrowers poorly served by standardised markets.

As the sector grows, attracts retail capital and becomes integrated with larger financial groups, incentives may shift towards:

  • preserving valuations;
  • maintaining fee-earning assets;
  • retaining investor capital;
  • extending fund lifecycles.

The market can drift from using flexibility to preserve economic value towards using flexibility to preserve reported stability.

Reform Constraints

Improved disclosures, stress exercises and supervisory data can expose leverage and interconnection.

They cannot create a reliable market price for every bespoke loan, remove uncertainty from future cash flows or determine in advance whether lender forbearance is economically justified.

Sudden Collapse

Private credit is not necessarily more likely to collapse suddenly than public credit.

Its structure may absorb shocks more gradually because capital is locked in and lenders can negotiate directly.

The sudden element arises when delayed recognition is forced:

  • a refinancing fails;
  • a valuation is marked down;
  • withdrawals reach their limit;
  • several institutions revise their assumptions at once.

The visible adjustment may be abrupt even when the deterioration was gradual.

Signals to watch

The private-credit market should be assessed through more than headline assets under management or reported defaults.

1. Payment-in-kind interest

Is a growing share of interest being added to loan balances rather than paid in cash?

2. Amendments and maturity extensions

Are changes resolving temporary problems, or repeatedly moving repayment obligations into the future?

3. Valuation dispersion

Do different lenders assign materially different values to comparable borrowers or to different parts of the same capital structure?

4. Non-accruals and realised recoveries

How many loans stop producing recognised interest, and how much is eventually recovered after impairment?

5. Refinancing requirements

What volume of debt must be replaced, and at what cost relative to the conditions under which it was originally issued?

6. Fund-level leverage

How much borrowing exists above the underlying corporate loan, including subscription facilities, net-asset-value finance and leverage within investment vehicles?

7. Redemption pressure

Are investors requesting withdrawals faster than funds can generate cash from repayments and ordinary portfolio activity?

8. Bank and insurer exposure

Are regulated institutions increasing direct loans, fund financing, insurance allocations or indirect exposure to the same borrowers and sponsors?

9. Sector and sponsor concentration

Are individual funds dependent on a narrow group of industries, private-equity sponsors or recurring financing structures?

10. Credit availability to companies

During stress, does private credit continue to finance viable businesses—or does defensive behaviour restrict investment and employment across the wider economy?

What would change the assessment?

The Erths assessment would improve if:

  • valuations adjusted consistently as borrower conditions changed;
  • amendments and payment-in-kind structures declined as refinancing conditions normalised;
  • losses and recoveries remained manageable through a sustained downturn;
  • redemption structures operated without persistent restrictions or forced selling;
  • banks, insurers and funds demonstrated that their exposures were not excessively concentrated;
  • private lenders continued supplying capital to viable companies during stress;
  • improved reporting reduced major gaps between institutional and system-wide views.

The assessment would weaken if:

  • reported values remained stable while comparable public credit deteriorated materially;
  • payment-in-kind interest and repeated extensions became widespread;
  • investor withdrawals rose across several semi-liquid funds;
  • banks reduced fund finance and corporate credit simultaneously;
  • different institutions discovered overlapping exposure to the same borrowers only after losses emerged;
  • uncertainty about private valuations caused credit conditions to tighten beyond the impaired assets themselves.

The decisive evidence will not be whether private credit avoids visible volatility.

It will be whether the market can recognise losses, allocate them and continue financing viable companies without requiring stability to be preserved through obscurity.

Conclusion

Private credit has become an important part of modern finance because it can do things that banks and public markets cannot always do efficiently.

It can negotiate around complex borrowers, commit capital for longer periods and restructure loans without forcing an immediate sale.

Those features can make the financial system more diverse and, under some conditions, more resilient.

But private credit changes the timing and location of information.

Risk that would appear through a traded price may instead appear through:

  • a valuation committee;
  • a covenant amendment;
  • a refinancing negotiation;
  • a restricted redemption window.

The system can remain calm while important assumptions change beneath the surface.

That calm should not automatically be treated as deception.

Nor should it automatically be treated as resilience.

The central question is whether flexibility is preserving economic value or postponing the recognition of loss.

Answering it requires visibility across borrowers, funds, banks, insurers and investors—not merely confidence in any one part of the system.

Private credit does not remove risk from view because it has removed the risk.

It changes where the risk can be seen, who is responsible for measuring it and when the wider system is required to acknowledge it.

The market above the surface may remain still.

The structure beneath it determines whether that stillness can last.

Picture of James Callard

James Callard

Structural Analyst
James Callard writes on structural risk, institutional change, and the dynamics of complex systems. His analysis focuses on the patterns that shape outcomes before they become visible in markets or policy.

A queue is supposed to establish order.

It takes a group of competing claims and arranges them into a sequence. Those at the front move first. Those behind them wait. The mechanism works only while position retains a meaningful relationship with progress.

Great Britain’s electricity connection queue gradually lost that relationship.

By mid-2025, more than 738 gigawatts of generation and storage capacity was waiting to connect to the electricity network. The queue had expanded tenfold in five years and contained roughly four times the generation capacity considered necessary for the government’s 2030 clean-power objectives. Some projects capable of proceeding were being offered waits of up to ten years.

The problem was not simply that the queue had become long.

It was that a project’s position no longer reliably indicated whether it possessed land, planning consent, finance, technical maturity or a realistic probability of being built. Projects that were ready could sit behind projects that were not. Capacity could be reserved for developments that might never progress, while more viable projects waited for network space to be released.

The queue remained administratively ordered.

Operationally, it had become misleading.

Erths assessment: The central failure was not excess demand alone. It was the collapse of the relationship between queue position, project readiness and system need.

1. What was the queue supposed to do?

New power stations, solar farms, wind farms, battery installations, interconnectors and large electricity users cannot simply attach themselves to the network.

A developer must apply for a connection. Network operators then assess where and when the project can connect, what infrastructure must be constructed or reinforced, and how the project will affect the wider electricity system.

Under the previous model, projects were generally prioritised according to when they accepted their connection offer. In principle, this created a neutral first-come, first-served sequence.

That method was understandable when the number of applications was manageable and most applicants had a reasonable prospect of proceeding.

It became less effective as the energy system changed.

The expansion of renewable generation, battery storage, electrification and speculative project development produced far more proposed capacity than the network could accommodate or the country was likely to require within the stated timescales.

The rules still treated acceptance of a connection offer as the organising event. But accepting an offer did not necessarily mean that a developer had secured planning permission, committed finance, obtained land rights or reached the stage at which construction could begin.

Chronological priority became a substitute for evidence of delivery.

2. The queue began measuring applications rather than projects

Every proposed development represented a claim on future network capacity.

Some claims were attached to mature projects. Others were exploratory, commercially uncertain or dependent on approvals that had not yet been secured. Yet once entered into the process, they could still affect the dates and infrastructure assumptions applied to projects behind them.

This produced two different queues occupying the same administrative structure:

  1. A queue of recorded connection agreements.
  2. A much smaller, less visible queue of projects with a realistic probability of being built.

The system had extensive information about the first and insufficient confidence about the second.

That distinction is important. A 738GW queue did not mean that Great Britain was preparing to construct 738GW of additional generating capacity. It meant that applications representing that volume had secured positions within the connection process.

The headline number therefore combined several different things:

  • credible projects approaching delivery;
  • early-stage projects still seeking permission or finance;
  • projects competing for the same commercial opportunity;
  • technologies concentrated in places where network capacity was limited;
  • projects unlikely to proceed within their contracted timetable;
  • capacity that exceeded strategic expectations for particular technologies.

The number was accurate as a count of contractual claims.

It was increasingly inaccurate as a forecast of likely infrastructure.

This is the defining characteristic of Measurement Breakdown: the metric continues to be calculated correctly after its relationship with the underlying reality has weakened.

3. Administrative order concealed physical impossibility

Ofgem’s assessment exposed the scale of the mismatch.

It found that 213GW of projects had connection offers dated before the end of 2030. Delivering all of them on schedule would have required a connection rate more than five times the historical average of approximately 8GW per year. Ofgem did not consider that feasible.

The dates therefore created an appearance of scheduled delivery that the historical rate of connection could not support.

Connection offers for both generation and demand were extending into the 2040s. The queued mix was also misaligned with the technology ranges considered necessary for a secure and operable 2030 electricity system. Batteries and solar projects were particularly oversubscribed in some areas, while other technologies risked being undersupplied.

None of this meant that every project was fictitious or deliberately speculative.

It meant that the system could not reliably distinguish, early enough, between several categories of claimant. A project could be commercially serious and still fail. It could possess land but not finance, planning consent but not equipment, or technical readiness without a viable route through the network.

A first-come system treated these uncertainties as matters to be resolved later.

As the queue expanded, “later” became the source of the blockage.

4. Why the distortion mattered

A congested connection queue does more than delay individual developers.

It changes how the wider system is planned.

Network companies must decide where substations, transmission lines and reinforcements will be required. If the project pipeline contains large volumes of uncertain or duplicative capacity, planners face a choice between two risks:

  • building infrastructure for projects that never materialise; or
  • withholding investment until the project pipeline becomes clearer.

The first risks unnecessary cost. The second risks delaying viable generation, storage and industrial development.

The queue also influences investment decisions. A developer receiving a distant or uncertain connection date may delay expenditure, seek another location or abandon the project. Investors cannot treat a connection offer as a dependable schedule if projects ahead of it may disappear, move or fail to satisfy their milestones.

This allows a self-reinforcing pattern to form:

Uncertain projects enter the queue → the queue creates uncertain network assumptions → connection dates become less credible → viable projects face greater uncertainty → developers submit more options to preserve future access.

A mechanism intended to coordinate investment can therefore encourage additional defensive applications.

The queue does not merely record uncertainty. It begins to produce it.

5. The structural drift

The original objective of the connection process was to allocate access fairly and coordinate the infrastructure required to connect customers.

Over time, its practical function changed.

It increasingly became a system for reserving future optionality.

For an individual developer, entering early could be rational even if the project was not ready. Waiting until every commercial and planning condition had been resolved might mean losing years in the connection sequence.

The rational action for each participant—apply early and preserve a position—produced an irrational aggregate outcome.

This is Structural Drift.

The formal purpose of the system remained unchanged. Its incentives gradually pushed behaviour away from that purpose.

The distinction matters because the failure cannot be attributed only to irresponsible applicants. Developers were responding to the rules presented to them. When early entry is rewarded and uncertainty carries insufficient cost, speculative participation becomes a predictable feature of the system.

The queue did not become distorted despite its incentives.

It became distorted through them.

6. From “first come” to “ready and needed”

In April 2025, Ofgem approved the TMO4+ reform package.

The new process replaced simple chronological priority with criteria intended to establish whether a project is both:

  • sufficiently ready to proceed; and
  • aligned with identified strategic need.

Projects meeting the requirements can receive Gate 2 status, which provides a confirmed connection point, date and queue position. Projects that do not meet the Gate 2 criteria are placed at Gate 1. Gate 1 does not reserve capacity, provide a firm queue position or guarantee a connection, although projects may reapply through future application windows.

Readiness is assessed through evidence such as project progress, land rights and planning status. Strategic alignment is determined through national capacity pathways, protections for certain advanced projects and limited powers to designate projects serving wider system requirements.

The reform therefore changes what the connection process measures.

Under the old system, the organising fact was when an offer had been accepted.

Under the new system, the organising facts include readiness, technology, location, delivery phase and alignment with strategic energy plans.

This should produce a pipeline that is more useful for network planning than a purely chronological queue.

It also represents a significant transfer of discretion.

A first-come system may generate poor outcomes, but its ordering rule is comparatively simple. A ready-and-needed system requires institutions to decide:

  • what counts as sufficient readiness;
  • how evidence should be weighted;
  • which technologies are required;
  • how much capacity is permitted in each region;
  • which projects deserve protection;
  • when strategic designation should override ordinary criteria.

Ofgem acknowledged stakeholder concerns that NESO’s discretion could reduce transparency, while concluding that some bounded discretion was necessary.

The reform therefore exchanges one form of risk for another.

It reduces the risk that chronological position protects undeliverable projects. It increases the importance of institutional judgement, data quality and strategic forecasting.

7. What the reordering achieved

In December 2025, NESO announced the first major results of the reformed process.

Approximately 283GW of generation and storage projects were identified to move forward:

The pipeline also included almost 100GW of new or expanded transmission-connected demand. More than 300GW of projects from the previous queue did not move forward at that stage, either because they were not ready, were not aligned with the relevant capacity pathways, did not apply, or selected a Gate 1 outcome.

The detailed results show the scale of active selection.

Phase 1—the projects associated with 2030 requirements—contained approximately 143.4GW when transmission-connected demand was included. Batteries represented 34.5GW, offshore wind 32.1GW and solar 29.9GW. Phase 2, covering needs through 2035, contained approximately 238GW, including 86.7GW of transmission-connected demand.

By April 2026, government and Ofgem reported that 221GW of projects holding firm connection agreements had been moved out of the main queue because they were not required by 2035 or were no longer progressing. Additional projects had voluntarily moved to Gate 1, meaning the total filtered capacity was higher.

The queue became smaller and more selective.

That is meaningful progress.

But it is not yet evidence that connections have become materially faster.

Reclassification is an administrative outcome. Delivery is a physical one.

8. Reform met the constraints the queue had concealed

Once the pipeline had been reordered, NESO and the network companies still had to convert queue outcomes into technically credible connection offers.

That process encountered delays.

Historic connection agreements contained data errors. Network assumptions had to be updated. Engineering studies had to be rerun in areas where the volume or type of Gate 2 projects differed from previous expectations. Government and Ofgem publicly stated in April 2026 that further slippage was unacceptable.

The revised timetable extended the issue of some offers well beyond the original queue-reordering announcement:

  • Phase 1 transmission and large embedded offers: through mid-September 2026;
  • Phase 1 distribution offers: through mid-November 2026;
  • Phase 2 transmission and large embedded offers: through mid-January 2027;
  • Phase 2 distribution offers: through mid-March 2027.

This does not prove that the reform failed.

It demonstrates that the queue was only one layer of the constraint.

Removing weak projects does not automatically:

  • construct new transmission lines;
  • increase engineering capacity;
  • accelerate planning decisions;
  • procure transformers, cables and switchgear;
  • resolve local network limitations;
  • eliminate errors inherited from previous agreements;
  • guarantee that selected projects reach construction.

The queue could be reorganised through policy and contract changes. The electricity network must still be expanded through physical work, regulated investment and planning consent.

This is Reform Constraints: intervention can correct the allocation mechanism without immediately correcting every dependent system.

9. The distortion moved to the demand side

The most revealing development came after the generation and storage queue had been reordered.

Demand projects—such as data centres and other large industrial users—had not originally been restricted by the same strategic capacity limits. Policymakers wanted to support economic growth and electrification rather than cap legitimate new demand.

That created a new opening for speculative claims.

On 29 July 2026, Ofgem reported that demand connection applications had surged from 41GW to 125GW in less than a year, with data centres accounting for at least 80GW. It opened consultation on commitment fees and new progress milestones intended to encourage less viable projects to leave the queue.

This is not a separate problem.

It is the same structural incentive appearing in another part of the system.

Where access to future network capacity is scarce and commercially valuable, participants have an incentive to reserve it before they can prove they will use it. Tightening the rules for one class of applicant can shift speculative behaviour into another class if the underlying value of early reservation remains.

The system therefore illustrates a broader principle:

Filtering the queue does not remove the incentive to manufacture optionality. It changes where that incentive appears.

Generation reform may have produced a more credible supply pipeline while leaving demand-side applications exposed to similar inflation.

This is why the success of the reform cannot be judged by one headline reduction in queued capacity.

The relevant question is whether the entire connection system now distinguishes credible demand from strategic reservation more effectively than before.

The Erths framework

Measurement Breakdown

Queue position ceased to function as a dependable indicator of readiness.

The queue’s total capacity remained an accurate measure of applications and agreements, but became an unreliable measure of projects likely to connect. Contracted dates also exceeded historically achievable connection rates.

The measurement was not false.

Its implied meaning had failed.

Structural Drift

The connection process moved from coordinating deliverable projects to preserving developer optionality.

Early application was individually rational because future access was scarce. The accumulation of those decisions weakened the queue for everyone.

The system’s behaviour drifted away from its stated purpose without a formal change in that purpose.

Hidden Instability

For a period, the queue looked orderly because every project possessed a position and nominal date.

The instability was concealed inside assumptions about which projects would progress, which infrastructure would be built and whether contracted dates were technically achievable.

Administrative precision masked delivery uncertainty.

Reform Constraints

Gate 1 and Gate 2 can improve selection and release capacity held by weaker projects.

They cannot independently accelerate planning, construction, equipment supply, engineering studies or every network reinforcement required to deliver the selected pipeline.

A better queue does not itself create a bigger grid.

Sudden Collapse

This is not primarily a Sudden Collapse case.

The connection process did not fail through a single abrupt breakdown. It deteriorated through accumulation. However, the system risked abrupt investment consequences if developers ceased to believe their connection dates or if strategically important projects became unable to proceed within policy timescales.

The visible crisis arrived late. The structural deterioration occurred much earlier.

Signals to watch

The reform should be judged through delivery rather than announcement.

1. Actual connection dates

Do Gate 2 projects connect materially earlier than comparable projects under the previous system, or do dates continue to move after offers are issued?

2. Milestone enforcement

How many Gate 2 projects fail readiness or progression milestones, and how quickly is their capacity released to other projects?

3. Network construction

Are transmission and distribution reinforcements completed quickly enough to support the prioritised pipeline?

4. Pipeline replacement

When projects leave Gate 2, can ready replacements enter without recreating years of uncertainty?

5. Demand-queue growth

Do commitment fees and progression requirements reduce speculative data-centre and industrial applications, or does demand continue to expand faster than credible delivery?

6. Technology concentration

Does the prioritised pipeline retain excessive concentrations of batteries, solar or particular demand types despite strategic capacity ranges?

7. Institutional discretion

Are designation, protection and strategic-alignment decisions published clearly enough for developers and investors to understand why one project advanced and another did not?

8. Repeated timetable revisions

Do the revised offer dates hold, or does administrative reordering continue to expose unresolved data and engineering problems?

What would change the assessment?

The Erths assessment would improve if the reformed process produced:

  • shorter realised connection times;
  • fewer projects holding capacity without progressing;
  • more reliable offer dates;
  • faster replacement of failed projects;
  • clearer network-planning assumptions;
  • transparent and predictable strategic decisions;
  • a demand queue tied more closely to evidence of financial and operational commitment.

The assessment would weaken if the queue became smaller on paper while:

  • physical connection rates remained broadly unchanged;
  • offer dates continued to slip;
  • selected projects failed at similar rates;
  • speculative capacity migrated between generation, storage and demand;
  • strategic discretion became difficult to scrutinise;
  • network construction remained the dominant unresolved bottleneck.

The decisive evidence will not be the number of gigawatts removed.

It will be the number of viable projects connected.

Conclusion

Great Britain’s electricity connection queue did not fail because it lacked rules.

It failed because the rules continued to produce orderly outputs after those outputs had lost much of their practical meaning.

A project could possess a place, a capacity allocation and a date without possessing a credible path to delivery. The queue measured contractual sequence more effectively than physical readiness.

The TMO4+ reforms directly confront that failure. They replace chronological priority with an evidence-based and strategically managed pipeline. The removal or downgrading of hundreds of gigawatts has produced a more plausible picture of what the system may actually need and build.

But the reform also reveals the limits of administrative correction.

The selected projects still depend on network construction, planning, engineering, equipment, finance and institutional execution. Data problems have already delayed parts of the offer programme. Meanwhile, speculative pressure has begun to reappear in the rapidly expanding demand queue.

The old queue asked:

Who arrived first?

The new process asks:

Who is ready, and which projects does the system need?

That is a better question.

Whether the institutions responsible can answer it accurately—and then deliver the infrastructure implied by their answer—remains unresolved.

Structural analysis for decision-makers. Published when there’s something precise to say — not on a schedule.  Subscribe →

Picture of James Callard

James Callard

Structural Analyst
James Callard writes on structural risk, institutional change, and the dynamics of complex systems. His analysis focuses on the patterns that shape outcomes before they become visible in markets or policy.

Erths Briefings — Issue #8

How new constraints emerge after adjustment

Adjustment is often treated as an endpoint.

Pressure is released.

Conditions stabilise.

The system resumes normal function.

But adjustment rarely returns a system to its previous state.

Instead, adjustment reshapes the environment in which future decisions are made.

The system resolves one condition while creating another.


1. What changed

Once systems move through periods of forced adjustment, several patterns typically emerge:

  • Stabilising mechanisms become embedded
  • New dependencies form around temporary solutions
  • Behaviour begins adapting to the adjusted environment

Initially these changes appear corrective.

Over time they become structural.

The system no longer operates under previous conditions.

It begins operating under newly established constraints.


2. What this means

Adjustment changes more than conditions.

This extends the distinction outlined in Why adjustment rarely resolves the underlying problem.

It changes available options.

As systems adapt to previous disruptions:

  • certain behaviours become reinforced
  • some responses become easier
  • others become more difficult

This creates a second-order effect:

Adjustment reduces one set of pressures while introducing another.

The immediate objective may have been stability.

The longer-term consequence is a modified system structure.


3. Where this leads

As new constraints accumulate, systems tend to develop:

  • Increasing path dependence
    Future responses become shaped by previous adjustment choices
  • Reduced flexibility
    Available responses narrow over time
  • Greater dependence on established mechanisms
    Stability becomes linked to systems that were initially temporary

This alters future behaviour.

Adjustment is no longer simply a response.

Earlier shifts toward threshold-based adjustment continue shaping behaviour long after immediate conditions stabilise, as outlined in What forces adjustment.

It becomes part of the system itself.


4. What to watch

The emergence of new constraints is often visible through:

  • Temporary measures becoming permanent structures
    Responses designed for short-term use persist over time
  • Increasing dependence on specific mechanisms
    Stability becomes linked to fewer supporting systems
  • Reduced diversity of response
    Different conditions increasingly produce similar actions
  • Changes in behavioural incentives
    Participants adapt to the adjusted environment rather than previous conditions

These signals indicate that adaptation is occurring.

But not without consequence.


5. Implication

When adjustment creates new constraints, system behaviour changes again.

This alters how recovery should be interpreted:

  • Stabilisation does not imply restoration
  • Adaptation does not imply flexibility
  • Resolution of previous pressure may create future limitations

At this stage, the system is functioning.

But it is functioning within a different structure than before.


Closing

This extends the progression from earlier briefings.

Stress was contained.

Corrective capacity weakened.

Adjustment became conditional.

Thresholds formed.

Systems adjusted.

Now adjustment itself is reshaping future behaviour.

Future briefings will track how these constraints accumulate — and how they influence the next cycle of system adaptation.


Erths Briefing
System-level analysis of how complex systems are shifting

Erths Briefings — Issue #7

Why adjustment rarely resolves the underlying problem


Adjustment is often interpreted as resolution.

Conditions change.

Pressure is released.

Stability returns.

But in constrained systems, adjustment frequently alters surface conditions without resolving the deeper structure that produced instability in the first place.

The system changes.

Its underlying dependencies often do not.


1. What changed

As systems move through periods of forced adjustment, a recurring pattern becomes visible:

  • Immediate pressure is reduced
  • Stability partially returns
  • Operational continuity is restored

Yet over time:

  • similar distortions re-emerge
  • dependence on intervention persists
  • adaptive capacity remains limited

This creates the appearance of resolution without full structural correction.


2. What this means

Adjustment and resolution are not equivalent.

Adjustment:

  • changes conditions
  • redistributes pressure
  • restores temporary stability

This follows the earlier transition toward threshold-based adjustment outlined in What forces adjustment.

Resolution requires something more difficult:

Structural realignment.

In constrained systems, this is often limited because:

  • existing dependencies remain intact
  • correction mechanisms are still weakened
  • stability continues to be prioritised over adaptation

As a result, adjustment tends to manage pressure rather than remove its underlying source.


3. Where this leads

When systems repeatedly adjust without resolving underlying constraints, several effects tend to emerge:

  • Recurring instability
    Similar pressures return in altered form
  • Increasing dependence on stabilisation
    Ongoing intervention becomes structurally necessary
  • Reduced flexibility over time
    Each adjustment narrows the range of future responses

These effects rarely emerge uniformly, particularly once thresholds begin forming unevenly across the system.

This produces a cycle where:

  • instability is managed
  • conditions temporarily improve
  • deeper constraints remain embedded

Over time, the system becomes increasingly shaped by unresolved adaptation.


4. What to watch

The distinction between adjustment and resolution is typically visible through:

  • Rapid return of similar pressures after stabilisation
    Conditions improve temporarily, then revert toward previous patterns
  • Persistent dependence on intervention
    Stability requires continued support rather than self-correction
  • Limited restoration of adaptive capacity
    Systems stabilise operationally but remain structurally constrained
  • Repeated use of similar responses across cycles
    Adjustment mechanisms become predictable and increasingly narrow

These signals indicate that the system has adjusted.

But not fundamentally realigned.


5. Implication

When adjustment occurs without resolution, systems remain vulnerable to renewed pressure.

This changes how periods of recovery should be interpreted:

  • Stabilisation does not necessarily indicate restored resilience
  • Improvement in conditions may reflect redistribution rather than correction
  • Recurrent instability may indicate unresolved structural dependence

At this stage, the system is capable of functioning.

But it remains shaped by the same underlying constraints that forced adjustment originally.


Closing

This extends the progression from earlier briefings.

Stress was contained.

Corrective capacity weakened.

Adjustment became conditional.

Thresholds began forming unevenly.

Now, adjustment is occurring without full structural resolution.

Future briefings will track how unresolved constraints shape the next phase of system behaviour.


Erths Briefing
System-level analysis of how complex systems are shifting

Erths Briefings — Issue #6

Where thresholds are forming

Systems operating under constraint do not move uniformly.

Pressure does not accumulate evenly.

Adjustment does not occur simultaneously.

Instead, stress concentrates in specific areas, where the capacity to absorb it is already limited.

This produces an uneven landscape.

Not all systems are equally close to forced adjustment.


1. What changed

As constraint deepens, differences between systems become more visible:

  • Some continue to absorb pressure with limited disruption
  • Others require increasing intervention to maintain stability
  • In certain areas, small changes produce disproportionate effects

Individually, these variations appear local.

In aggregate, they indicate divergence.

The system is no longer behaving as a single, continuous structure.

It is fragmenting into areas of relative stability and areas of emerging constraint.


2. What this means

Threshold-based adjustment does not occur everywhere at once.

It emerges where:

  • adaptive capacity is already reduced
  • intervention is most heavily relied upon
  • correction mechanisms have weakened the most

This creates a condition where:

Adjustment becomes location-specific rather than system-wide.

Some areas remain stable.

Others move closer to points where stability can no longer be maintained.

This is not random.

It reflects how pressure is distributed.


3. Where this leads

As divergence increases, systems tend to develop:

  • Early adjustment zones
    Areas where thresholds are reached first
  • Stability-dependent zones
    Areas where conditions hold, but require ongoing support
  • Transitional zones
    Areas where pressure is building but not yet forcing adjustment

This segmentation alters system behaviour.

Adjustment no longer propagates smoothly.

It emerges in stages.


4. What to watch

As thresholds begin to form, several indicators become more relevant:

  • Localised instability within otherwise stable conditions
    Disruption appears in specific areas without immediate system-wide impact
  • Increasing sensitivity in certain segments
    Small changes produce larger responses in particular areas
  • Uneven effectiveness of intervention
    Stabilisation works in some areas, but weakens in others
  • Delayed transmission of stress
    Pressure does not spread immediately across the system

These signals indicate that thresholds are not uniform.

They are forming unevenly.


5. Implication

When thresholds form unevenly, risk becomes concentrated rather than distributed.

This changes how conditions should be interpreted:

  • Stability in one area does not imply stability elsewhere
  • Localised disruption may indicate broader structural pressure
  • Adjustment is more likely to begin in constrained segments

At this stage, the system is still functioning.

But its behaviour is no longer uniform.

It is becoming segmented.


Closing

This extends the progression from earlier briefings.

Stress has been contained.

Systems have prioritised stability.

Corrective capacity has weakened.

Adjustment has become conditional.

Now, thresholds are forming unevenly.

Future briefings will track how adjustment emerges across these segments — and how it propagates through the system.


Erths Briefing
System-level analysis of how complex systems are shifting

Erths Briefings — Issue #5

What forces adjustment

Systems operating under constraint do not adjust continuously.

They maintain stability for as long as possible.

Correction is delayed, redirected, or suppressed.

This raises a critical question:

If systems are no longer correcting effectively,
what causes adjustment to occur?


1. The absence of continuous correction

In earlier conditions, systems adjust incrementally.

Small imbalances produce small responses.

This keeps the system aligned over time.

Under constraint, this process weakens:

  • Correction is delayed
  • Signals are absorbed
  • Adjustment becomes conditional

The system continues to function.

But it no longer adapts continuously.


2. What replaces it

When continuous correction breaks down, adjustment does not disappear.

It changes form.

Instead of gradual alignment, systems move toward:

threshold-based adjustment

Change occurs only when conditions exceed the system’s ability to contain them.

Below that threshold:

  • imbalance persists
  • intervention stabilises
  • adjustment is deferred

Above it:

  • correction occurs rapidly
  • often disproportionately
  • and with limited control

3. What triggers the threshold

Adjustment is typically forced by one of three conditions:

  • Accumulated internal imbalance
    Misalignment reaches a level that can no longer be stabilised
  • External shock
    An event introduces stress faster than it can be absorbed
  • Failure of stabilising mechanisms
    Interventions lose effectiveness or cannot be sustained

These triggers differ in form.

But they produce the same outcome:

The system is forced to adjust.


4. How adjustment occurs

When triggered, adjustment tends to be:

  • Discontinuous
    Change happens in steps, not gradually
  • Compressed
    What would normally occur over time happens quickly
  • Less controlled
    The system responds with fewer available options

This reflects the prior loss of adaptive capacity.

The system is not choosing optimal adjustment.

It is responding within constraint.


5. What to watch

In systems approaching threshold-based adjustment, several signals become more relevant:

  • Increasing stability despite visible imbalance
    Conditions hold longer than expected
  • Repeated intervention with narrowing effect
    Stabilisation continues, but with reduced impact
  • Growing dependence on specific mechanisms
    Stability relies on a smaller set of supports
  • Sensitivity to smaller disturbances
    Minor changes produce disproportionate reactions

These indicate proximity to a point where adjustment becomes forced.


6. Implication

When systems rely on threshold-based adjustment, risk changes in character.

It is no longer defined by:

  • gradual deterioration
  • visible instability

It is defined by:

the conditions under which adjustment becomes unavoidable

This changes how current conditions should be interpreted:

  • Stability may persist longer than expected
  • Adjustment may occur more abruptly than anticipated
  • Triggers may appear unrelated to underlying causes

At this stage, the system is not unstable.

But it is increasingly dependent on thresholds holding.


Closing

This extends the progression from earlier briefings.

Stress has been contained.

Systems have protected their current state.

Corrective capacity has weakened.

Adjustment is now conditional.

Future briefings will track where thresholds are forming — and how they are likely to be tested.


Erths Briefing
System-level analysis of how complex systems are shifting

Erths Briefings — Issue #4

Where these dynamics are now visible

The conditions outlined in recent briefings are not isolated.

They are beginning to appear across multiple systems simultaneously.

Not in identical form.

But with similar underlying structure.

The common pattern is not instability.

It is constrained stability.


1. What is emerging

Across several domains, the same configuration is becoming more apparent:

  • Surface conditions appear stable or controlled
  • Adjustment mechanisms are active but less effective
  • Intervention is required to maintain normal functioning

This is not a breakdown.

It is a system operating with reduced adaptive capacity.


2. How it presents

These conditions do not present as crisis.

They present as persistence.

Specifically:

  • Conditions that remain stable despite underlying imbalance
  • Repeated interventions that prevent deterioration but do not resolve it
  • Signals of stress that do not produce proportional adjustment

This creates a consistent but misleading impression:

That the system is holding.


3. Where it appears

This pattern is most visible in systems that:

  • Depend on continuous coordination
  • Require ongoing adjustment to remain aligned
  • Cannot easily absorb short-term disruption

In these systems:

  • Stability becomes an output that must be maintained
  • Correction becomes conditional rather than continuous

This is not sector-specific.

It reflects a broader shift in system behaviour.


4. What to watch

As this configuration becomes more common, several indicators become more relevant:

  • Stability that depends on sustained intervention
    Conditions remain stable only with continued input
  • Adjustment that produces diminishing effects
    Responses occur, but their impact weakens over time
  • Delayed recognition of misalignment
    Signals are present, but response lags increase
  • Localised disruptions that do not propagate immediately
    Stress appears in isolated areas without broader adjustment

These signals indicate that systems are still functioning.

But not correcting effectively.


5. Implication

The significance of this phase is not immediate risk.

It is structural change.

Systems in this condition:

  • Appear stable
  • Operate continuously
  • But adapt less effectively over time

This changes how current conditions should be interpreted:

  • Stability may reflect ongoing intervention rather than underlying strength
  • Lack of disruption may reflect delayed adjustment
  • Persistence may indicate constraint, not resolution

At this stage, the system is not failing.

But it is becoming increasingly dependent on conditions holding.


Closing

This extends the progression from earlier briefings.

Stress is being contained.

Systems are protecting their current state.

Corrective capacity is weakening.

These conditions are now observable.

Future briefings will track where this configuration begins to break — and how adjustment re-emerges.


 

Erths Briefing
System-level analysis of how complex systems are shifting

Erths Briefings — Issue #3

When systems lose the ability to correct

As systems begin protecting stability over function, their ability to adapt weakens.

Not immediately.

But progressively, through the repeated suppression of correction.

Over time, this produces a more significant shift.

The system does not simply carry misalignment.

It loses the capacity to resolve it.


1. What changed

Across systems already under sustained pressure, several signals are becoming clearer:

  • Corrective actions are delayed even when misalignment is visible
  • Interventions are repeated despite diminishing effectiveness
  • Adjustments produce smaller and shorter-lived effects

Individually, these can be interpreted as temporary friction.

In combination, they indicate a change in system capability.

Not just in what the system is doing.

But in what it is able to do.


2. What this means

All adaptive systems rely on correction mechanisms.

These mechanisms:

  • detect imbalance
  • trigger adjustment
  • restore alignment

When systems begin protecting their current state, these mechanisms are used less effectively.

Over time, this leads to a second-order effect:

The system retains the appearance of control, while losing the ability to correct.

Correction is no longer absent.

It is ineffective.

This distinction matters.

Because ineffective correction:

  • delays adjustment
  • reinforces existing conditions
  • reduces responsiveness further

3. Where this leads

When correction mechanisms weaken, systems tend to move toward:

  • Accumulation of unresolved imbalances
    Distortions persist and compound over time
  • Increasing sensitivity to external triggers
    Smaller shocks produce larger effects
  • Reduced range of viable responses
    Fewer effective options remain available

At this stage, the system is still functioning.

But its behaviour becomes more constrained.

Adjustment is no longer gradual.

It becomes conditional on disruption.


4. What to watch

The loss of corrective capacity is typically visible through:

  • Repetition of interventions with declining impact
    The same actions are taken, with weaker outcomes
  • Longer lag between signal and response
    Systems recognise imbalance but respond slowly
  • Narrowing response patterns
    Different conditions produce similar actions
  • Increasing reliance on external stabilisation
    Stability depends on sustained input

These signals indicate a system that is no longer adapting effectively.


5. Implication

Once a system begins to lose the ability to correct, its trajectory changes again.

The primary risk is no longer misalignment.

It is unresolved misalignment.

This alters how current conditions should be interpreted:

  • Stability may reflect suppressed adjustment
  • Intervention may indicate reduced effectiveness
  • Lack of visible change may indicate loss of response capacity

At this stage, the system is not yet failing.

But it is approaching a point where adjustment becomes dependent on disruption.


Closing

This extends the progression from Issues #1 and #2.

Stress has been contained.

The system has begun protecting its current state.

Now, its ability to correct is weakening.

Future briefings will track where systems retain adaptive capacity — and where adjustment becomes dependent on external shock.


Erths Briefing
System-level analysis of how complex systems are shifting

Erths Briefings — Issue #2-

When systems start protecting the wrong thing

As stress becomes contained within a system, its behaviour begins to change.

Not by failing immediately.

But by gradually shifting what it prioritises.

Systems are designed to adapt and correct.

Under sustained pressure, they begin to prioritise stability instead.


1. What changed

Across several large systems, a similar pattern is emerging:

  • Increasing effort is directed toward maintaining existing conditions
  • Interventions are designed to preserve outcomes rather than enable adjustment
  • Signals that would normally trigger correction are being absorbed or delayed

Individually, these are often interpreted as effective management.

In combination, they indicate a shift in system priorities.

From:
– adapting to changing conditions

To:
– maintaining continuity


2. What this means

All systems operate with implicit priorities.

Under normal conditions, these priorities support long-term function:

  • responsiveness
  • adaptability
  • correction of imbalance

Under sustained pressure, that alignment weakens.

The system begins to prioritise:

  • short-term stability
  • continuity of output
  • avoidance of visible disruption

This creates a structural inversion:

The system begins protecting its current state, rather than its underlying function.

Correction mechanisms still exist.

But they are no longer being applied effectively.


3. Where this leads

When stability is prioritised over function, several effects tend to follow:

  • Distortions persist longer than they should
    Imbalances are carried forward instead of resolved
  • Adjustment becomes progressively more difficult
    Each intervention reinforces the existing configuration
  • The cost of correction increases over time
    Because misalignment compounds beneath the surface

This does not produce immediate failure.

It produces a period where:

  • the system appears stable
  • but becomes progressively less adaptable

Over time, the range of possible outcomes narrows.


4. What to watch

The transition toward protective behaviour is typically visible through:

  • Interventions that prioritise outcomes over process
    Focus shifts to maintaining results rather than restoring underlying function
  • Suppression or delay of corrective signals
    Indicators that would normally trigger adjustment are muted or deferred
  • Reduced tolerance for short-term disruption
    Even minor instability is actively managed
  • Increasing uniformity in system responses
    Different conditions begin to produce similar interventions

These are not isolated decisions.

They indicate a change in how the system is operating.


5. Implication

Once a system begins protecting the wrong thing, its trajectory changes.

The immediate effect is stability.

The longer-term effect is reduced resilience.

This changes how current conditions should be interpreted:

  • Stability may indicate constraint rather than strength
  • Intervention may signal fragility rather than control
  • Lack of visible correction may indicate deferred adjustment

At this stage, the system is not failing.

But it is becoming less capable of adapting when it needs to.


Erths Briefing
System-level analysis of how complex systems are shifting

When Systems Start Protecting the Wrong Things

 

How complex systems drift from their original purpose and begin defending metrics, processes, and internal stability instead.


Introduction

Most systems do not fail because they stop functioning.

They fail because they continue functioning while gradually serving the wrong purpose.

What begins as a system designed to achieve a clear outcome — to educate, to govern, to allocate capital, to deliver reliable infrastructure — does not suddenly abandon that goal. Instead, it adapts to pressure. It introduces measurements, processes, and controls.

Over time, those adaptations begin to reshape the system itself.

Targets become central.
Processes become binding.
Stability becomes a priority.

The system still operates. In many cases, it appears more organised than before.

But what it is protecting has quietly changed.


From purpose to proxy

No complex system can operate directly on its underlying purpose alone.

It requires simplification.

A university cannot measure “education” directly, so it relies on graduation rates, rankings, and research output.
An institution cannot measure “effective governance” directly, so it relies on compliance, reporting, and procedural adherence.
A system cannot measure “resilience” directly, so it tracks efficiency, utilisation, and cost.

These are not the goal. They are proxies — imperfect representations of something more complex.

At first, they serve the system well.

They allow coordination, comparison, and accountability.

But over time, something shifts.

Instead of asking whether the system is achieving its purpose, the system begins to ask whether it is meeting its proxies.

And eventually, those proxies become what is defended.


How the shift occurs

This transition does not require failure, mismanagement, or bad intent.

It emerges from the normal pressures of operating a complex system.

Measurement

What can be measured becomes what is managed.

But what can be measured is often only a narrow slice of what matters.

So the system gradually optimises around what is visible, even if it is incomplete.


Standardisation

As systems scale, they rely on standard processes to maintain consistency.

These processes reduce variability and improve coordination.

But they also reduce discretion.

Over time, following the process becomes more important than questioning whether the process still serves the purpose.


Accountability

Systems require mechanisms to evaluate performance.

Clear indicators are necessary.

But those indicators simplify reality.

So performance becomes defined by what can be reported, rather than what is actually achieved.


Stability

Under pressure, systems prioritise continuity.

Avoiding disruption becomes a goal in itself.

Preserving the system begins to take precedence over improving it.


Each of these forces is reasonable on its own.

Together, they gradually reorient the system.


When protection replaces purpose

At a certain point, the system crosses a threshold.

It no longer primarily protects its original function.

Instead, it protects:

  • its metrics
  • its processes
  • its internal coherence
  • its appearance of success

This shift is rarely explicit.

Participants within the system continue to act rationally within the incentives they face.

But the outcome changes.

The system becomes more effective at preserving itself than at fulfilling its purpose.


Why this is difficult to detect

This form of drift is hard to recognise from within.

Because the system still works.

Operations continue.
Outputs are produced.
Targets are met.

In some cases, performance may even appear to improve — because the system has become more efficient at satisfying its own internal criteria.

But those criteria may no longer reflect reality.

This is where metrics begin to lose their meaning, and where feedback becomes weaker.

Signals that should trigger correction are filtered, delayed, or reinterpreted.

The system becomes less responsive without appearing unstable.


The consequences

Once a system begins protecting the wrong things, several patterns tend to follow.

Reduced adaptability

Change becomes difficult because it threatens the structures the system now depends on.


Increasing complexity

New layers are added to manage emerging problems, but these layers often reinforce existing structures rather than correct them.


Declining effectiveness

The system continues to operate, but outcomes drift further from its original purpose.


Delayed recognition

Because internal indicators still signal stability, problems are not fully acknowledged until they become difficult to ignore.


This is how hidden instability builds beneath visible order.


From misalignment to breakdown

Systems can operate in a misaligned state for extended periods.

But the longer the gap between purpose and behaviour persists:

  • the harder it becomes to correct
  • the more capacity is quietly eroded
  • the more dependent the system becomes on its own internal logic

Eventually, the difference between appearance and reality becomes too large to sustain.

When adjustment finally occurs, it is often more abrupt and more disruptive than it would have been earlier.

What appears to be a sudden failure is often the result of a long period of unnoticed misalignment.


Conclusion

Systems do not only fail because of external shocks or poor decisions.

They also fail because:

they gradually redefine what success means — and then optimise for the wrong definition.

When that happens, the system may become highly effective at achieving outcomes that no longer matter.

And by the time that misalignment becomes visible, it is often deeply embedded.

Structural analysis for decision-makers. Published when there’s something precise to say — not on a schedule.  Subscribe →

Picture of James Callard

James Callard

Structural Analyst
James Callard writes on structural risk, institutional change, and the dynamics of complex systems. His analysis focuses on the patterns that shape outcomes before they become visible in markets or policy.

Scroll to Top