Number 10 North and the Two-Hundred-Mile Perimeter

Number 10 Downing Street was jerry-built. Sir George Downing - a man Pepys thought a miser and history has judged more harshly - threw up a terrace of speculative townhouses in the 1680s on the edge of what had been the Westminster marsh, the medieval Thorney Island, mortaring for profit rather than permanence and setting them on shallow foundations that began to settle almost at once.¹ The house has been quietly subsiding into the London clay more or less ever since. By the 1950s the decayed buildings had reached a condition their own custodians called critical;² Harold Macmillan's Crawford Committee, appointed in 1957, recommended a full reconstruction, and between 1960 and 1963 Numbers 10 and 11 were effectively gutted and rebuilt on deep piling under the architect Raymond Erith.³

We govern, in other words, from a seventeenth-century speculative development that has spent three centuries being propped up. This is worth remembering before we grow too anxious about a fortnight-old office in Manchester, because it sets the correct baseline: the home of British government has never been secure by design, only secure by accretion, and everyone who has worked in it has managed the gap between the two.

I raise this because the arrival of Number 10 North invites exactly that comparison, and the comparison mostly flatters the new arrangement. Prime Minister Andy Burnham opened the Manchester base at GCHQ's Heron House, on Albert Square, on 24 July;⁴ he intends to work there weekly, with Cabinet colleagues attending regularly,⁵ while a permanent civil service campus is built a mile east at Ancoats.

Early commentary has fixed on two things - the politics, which is not my province, and the armed guarding at the door, which is only part of it. I want to look instead at the parts nobody photographs: the building's security culture, the behaviour of the people now moving through it, and the two hundred miles of railway that have quietly become part of the government estate.

Upfront, my opinion is that the building is the soundest element of the whole scheme and the anxiety aimed at it is largely misplaced - Heron House will hold. The real issue lies with the potential compromise that comes with numerous MPs, civil servants, and associated staffers traipsing twice a week, up and down the West Coast Main Line with the working papers (and personal electronic devices) of state in their hand luggage.

Some early commentary has fastened on armed protection: whether Greater Manchester Police can sustain the task, what a second protected residence-in-effect means for national arrangements - close protection of the Prime Minister ordinarily falls to the Metropolitan Police's Royalty and Specialist Protection command⁶ - and what it says that the force is currently covering the site from existing resources while longer-term funding is negotiated.⁷ A former head of royal protection has already christened the site "Number 10A", which tells you the protection community is paying attention.

I am not a protection officer, armed policing is a specialist discipline with its own doctrine and its own inspectorate, and I will not pretend to understand expertise I do not hold. I offer only one observation from the general protective security side: static armed guarding is the easy half of this problem. It is visible, it is politically legible, and it will therefore be resourced, eventually, because its absence would be intolerable on camera. The hard half is everything that moves - the routine, repeating, twice-weekly pattern of people and material in transit - I will touch on my concerns with the transiting workforce later.

On paper the choice makes itself. Heron House has been GCHQ's North West site since 2019;⁸ it is routinely described as the most secure building in Manchester (some say the only genuinely hardened one);⁹ the clearances, the accreditation, the technical infrastructure and the security-cleared workforce already exist inside it. Nothing else in the city could have been stood up at short notice to anything approaching the same standard, and standing up a new secure facility from a cold start is a programme measured in years and tens of millions. If the requirement was "a secure interim home, available now", Heron House was not the best answer; it was the only answer.

Two caveats deserve more attention than they have had. The first is that "interim" is carrying an unusual load. The Ancoats campus is not expected until around 2032, which is to say that this temporary arrangement is measured in parliaments rather than months, and should be planned, resourced and governed as a medium-term operating site rather than a stopgap. Temporary arrangements that last six years acquire permanent risks.

The second is that co-location cuts both ways. A working signals intelligence site now shares its front door with the most photographed political address in the north of England. That is an aggregation of target value - a serving intelligence facility, a Prime Minister on a weekly rhythm, visiting secretaries of state, and broadcast trucks on the square outside - concentrated in a single postcode, and every hostile reconnaissance effort in the country has just been handed its aiming mark.

The building can absorb that; it was designed to. What is subtler is the collision of cultures: an organisation that avoids attention professionally now cohabits with a trade that seeks it professionally, and the discreet regional hub of 2019 is now the backdrop to pieces to camera. (Whether the Prime Minister of the United Kingdom should emerge from a secure lobby more or less directly onto the pavement outside a Slug and Lettuce is, as one Manchester outlet dryly observed, a question all of its own.)

But I do not want to play the naysayer, because the interim arrangement contains the seed of something better than what it replaces, and it is worth saying so plainly. The permanent home - the government's new digital campus rising on the old Central Retail Park at Ancoats, some 890,000 square feet for around 7,000 civil servants, Treasury-approved in March and due around 2032¹⁰ - will be built. That single fact changes the analysis, because a building that is built, rather than inherited, can be built secure by design: protective security engineered into the concrete and the cable runs from the first drawing rather than retrofitted around a listed facade and a speculative Restoration terrace.

Everything I have catalogued as friction at Heron House - the device lines, the escorting choreography, the collision between an open political culture and a closed intelligence one - is, at Ancoats, a requirements document waiting to be written. Standoff, access control, screened delivery, secure but usable meeting space, a threshold designed for a mixed population rather than a cleared one: these are cheap on the drawing board and ruinous to add later. The lesson of Downing Street is precisely that you cannot easily bolt security onto a building that never anticipated it; the opportunity of Ancoats is that, this once, we get to anticipate it.

There is a second, subtler advantage, and it concerns calibration. Heron House is a GCHQ site, which means it is almost certainly secured to a standard well above what a devolution secretariat and a weekly prime ministerial visit actually require - a genuine mismatch, since security pitched too high is not a free surplus but a source of its own failures, driving the workarounds and the displaced work I come to below.

A purpose-built campus can be pitched at the right level: proportionate to the real threat and the real classification of the work, informed by risk assessment rather than inherited from the most secretive tenant in the building. Getting security right is not the same as getting the most of it, and the campus is a chance to demonstrate the difference.

The precedents for moving a government, encouragingly, run more to nuisance than catastrophe. When the Federal Republic of Germany relocated from Bonn to Berlin in 1999 it shifted some tens of thousands of posts and, tellingly, kept a substantial governmental footprint in Bonn - dual-sited by law, partly as political settlement and partly as deliberate resilience, a hedge that looks wiser with every passing decade of concentration risk.¹¹

Closer to home, the Palace of Westminster's long-postponed Restoration and Renewal programme, with its planned "full decant" of both Houses, is the standing British reminder that "temporary" relocations of the seat of power are neither quick nor cheap, and that the interim is where the risk actually lives.¹²

The cautionary examples cut the other way: the states that have built brand-new, security-by-design capitals - Egypt's New Administrative Capital, deliberately set forty-odd kilometres from the crowds of Tahrir; Myanmar's Naypyidaw, raised from scrubland in 2005 and pointedly distant from Yangon's protest geography - have tended to buy their security by putting physical distance between the governors and the governed.¹³ Burnham has chosen the opposite: a base in the middle of a living and vibrant city, steps from its central square, its bars, its markets, and its trams. That is the braver choice democratically, and by some way the harder one to secure. The whole of this essay is, in a sense, about the size of that bill.

It is worth pausing to run the site through the same lens I previously applied to the home base at large, because each pillar lands here in a specific way, and because the most uncomfortable finding concerns people who will never hold a pass.

Terrorism first. The symbolic value of a "second Downing Street" is precisely the currency in which the current threat trades, and the operating rhythm - the Prime Minister weekly, Cabinet regularly - is in effect published. But the harder point is displacement: hardening works, and one of the things hardening does is move the attack. A genuinely hardened building overlooking a civic square potentially transfers risk outward onto the softest thing in view, which is the crowd in front of it.

Albert Square is not a sterile forecourt; it is one of Manchester's principal gathering places - protest, celebration, the seasonal markets - and it has just acquired the most attractive target in the north of England as a neighbour. No city in Britain understands this trade better: this is the city of the 2017 Arena attack - where the device was detonated not inside the secured hall but in the City Room foyer, the crowd space beyond the ticket line, as the audience dispersed¹⁴ - and the birthplace of Martyn's Law,¹⁵ and the question the square now poses is not only "can they get in" but "who is standing next to the building when they cannot". The wider public realm - the square, the approaches, the event calendar - must be inside the protective boundary of the planning, because it is already inside the target description.

And the displacement logic travels. Once principals and their staff begin moving on the same routes at more or less the same times - the pattern the weekly operating rhythm all but guarantees - the target stops being an address and becomes a journey. A hardened building can turn an attack away; a protected person on a predictable schedule merely relocates it, to the concourse at Euston beneath the hated electronic information boards, to the subsequent scrum and Mario Kart dash when the platform is finally announced , to the carriage itself - every one of them a crowded public space in which the people standing nearest the target have no idea they are standing near anything at all. This is why pattern-setting is the cardinal sin of protective travel: it converts the routine of the protected few into the inherited risk of the travelling many, none of whom volunteered for it.

The fixated individual, the self-initiated attacker, the tasked proxy - none of them needs to reach Albert Square; each needs a ticket and a timetable, and both are on general sale. Whatever the eventual armed-protection arrangements deliver, the cheapest control remains the one protective doctrine has always prescribed: vary the pattern¹⁶ - the route, the hour, the service, the coach¹⁷ - because predictability is a gift no adversary should be handed twice a week, and because the bill for it, if it is ever presented, will be paid substantially by the public who happened to share the train.

Espionage extends the aiming-mark point from Part 3. Close-target reconnaissance now has a fixed address and a repeating pattern of life to build against; every regular attendee is a potential cultivation target; and, as the national authority itself observes, the harder the physical and technical defences, the more attractive the recruitment of an insider becomes.¹⁸ Hostile services canl therefore map the supporting cast first - drivers, cleaners, caterers, contractors, the hospitality staff of every venue within a five-minute walk - because the supporting cast is always softer than the principal, and because personnel-security screening reaches down the sub-contracting chain only as far as it is made to.¹⁹

Sabotage need never touch the building. The site's dependencies are the target set: power, telecommunications, and above all the rail corridor on which the entire operating model runs. Disrupt the West Coast Main Line on the right morning and you have disrupted the rhythm of government without approaching a fence - and the proxy model for exactly that kind of cheap, deniable interference is, as I set out in the Omnicrisis assessment, already proven on British soil,²⁰ whilst the November 2025 sabotage attempt against a passenger railway line between Lublin and Warsaw demonstrated that adversaries are not averse to attacking commuter infrastructure for their ends.²¹

Subversion, meanwhile, treats the site as a stage. A second Downing Street is a narrative object as much as a building: protest will likely concentrate on the square because concentrating on that square is what Manchester protest does; the stories about cost, secrecy and "the spies running the government" write themselves; and every security incident, however trivial, will be amplified as proof of institutional folly. The information environment around the project is part of its protective posture, whether or not anyone has planned it that way.

Organised crime supplies the market that makes Part 7's losses matter. This is where the claim has to be evidenced rather than asserted, because the lazy version of it - "stations are dangerous" - is both untrue and beneath the subject. The accurate version is narrower and better documented: theft of passenger property is consistently the single most common recorded offence on the railway, and the two termini of this particular corridor sit at the top of the national table by volume.

In the most recent official figures, theft of passenger property is the single largest category of recorded crime on the railway - very nearly a quarter of the total - and the busy London termini and interchanges, Euston and Manchester Piccadilly among them, concentrate both the passengers and the opportunity.²²

The essential caveat belongs in the same breath, in fairness to both cities: measured per passenger rather than by raw count these are not the country's riskiest stations at all - that distinction falls to tiny rural stops with a handful of incidents against a few thousand users - and serious crime on the network remains rare, well under one offence per hundred thousand journeys.²³ But volume is exactly what matters here. A high absolute number of property thefts at your two boarding points is not a comment on London, or on Manchester; it is a statement of the ambient probability that a distracted traveller in a crowd will be separated from a bag, and it applies to a cabinet minister's staffer exactly as it applies to anyone else. It now has a premium product line, because theft-to-order is a short step from theft-by-chance. The thief does not need to know what they have taken. They need only know that somebody will pay to find out.

None of these pillars requires the building to fail. That is the point of running the taxonomy: the building is the one element of this arrangement that is genuinely hard, and every pillar simply routes around it - to the crowd, the supporting cast, the infrastructure, the narrative, and the luggage rack.

Here is the paradox worth naming: the site's greatest strength - its uncompromising security posture - is also where the first breaches will likely come from. A building run to intelligence-community standards imposes disciplines that most of government has never lived under: restrictions on personal electronic devices at the threshold, escorting protocols, controlled areas, rules about what may be said, carried, photographed and connected. Into that environment we are now injecting a population of political staff, private office, visiting officials and guests who did not grow up inside it. Security culture is trained, not issued with the lanyard.

The early incidents will therefore not be hostile; they will almost certainly be habitual. The phone that stays in the pocket past the point it should not. The visitor who does not understand what escorted status actually requires of the escort, or even someone who fails to understand their escorting duties. The tailgate through a door held open out of ordinary courtesy. The photograph taken in the wrong atrium, because political staff photograph everything; it is the job. Each is individually trivial, and the aggregate is a steady erosion of exactly the controls the accreditation depends on, along with an induction and supervision burden that lands on the host. None of this is unmanageable, but it must be managed as a programme - briefed, drilled and enforced from day one - rather than absorbed as friction.

Friction has its own failure mode, which is the more serious of the two: friction displaces work. If the building is hard to work in - if the phone must be surrendered, the call cannot be taken, the document cannot leave - the work inevitably and lazily migrates to wherever it is easier: the hotel lobby, the coffee shop on the square, the train. This is not a hunch but one of the better-documented findings in security research: when controls impose more cost than users judge the task can bear, they do not stop working, they route around the control - the phenomenon Sasse and colleagues term 'shadow security', the workaround that gets the job done at the price of the protection.²⁴ A site can be secure enough that it pushes the sensitive conversation outside its own wire. That is the precise sense in which "too secure" stops being a compliment and becomes a finding.

It is at this point that the assessment has to leave buildings and timetables and drill into the thing that actually determines outcomes, which is human behaviour. The civil service is not untrained; every official at Number 10 North will have completed mandatory security training, most of it annually, much of it recently. But training and culture are different substances. Training is what an organisation says; culture is what it tolerates. Training tells you the rule; culture is what you do with it when you are tired, late, carrying two bags, and the platform has just been announced. The annual e-learning module was written for a Whitehall context - a pass-controlled department, a government laptop, a specific coloured lanyard - and it says very little about how to behave inside a working intelligence facility, and nothing at all about how to behave on the 18:15 from Euston.

The set-up raises a bespoke set of behavioural challenges that generic training was never designed to meet. The population is mixed in a way no single security regime anticipates: intelligence officers with a career's immersion in the disciplines; career officials with the standard baseline; political staff with high churn, high mission-urgency and low prior exposure; ministers and their outer offices; police, contractors, visitors. Each group arrives with a different internal model of what "secure" requires, and the controls will be experienced by some as second nature and by others as obstruction. And the behavioural failure mode in mixed populations is well understood: it is not the dramatic breach but the normalisation of deviance - the term Diane Vaughan drew from the Challenger inquiry for the process by which an unsafe practice, repeated without immediate consequence, quietly becomes the accepted standard.²⁵ The workaround goes unchallenged, is repeated, is copied, and within a month is simply how things are done here. Every uncorrected shortcut is a lesson taught to everyone who watched it.

The status asymmetry makes the correction mechanism itself fragile. A challenge culture is easy to prescribe and socially expensive to practise, and at this site the person most likely to breach a rule will frequently outrank, or out-influence, the person whose job it is to enforce it. Whether a relatively junior officer at a threshold feels able to stop a secretary of state's special adviser over a phone - and whether the system visibly backs them when they do - will tell you more about the site's real security posture than any accreditation document. Culture is set by what leaders do, not what inductions say: if the Prime Minister surrenders his device at the line without ceremony, that fact will circulate faster than any policy; if his team treats the controls as beneath the mission, that will circulate faster still.

Which is why the timing matters more than it appears to. Number 10 North was conceived, announced and opened within the Prime Minister's first week in office - an impressive feat of political logistics, and a guarantee that behaviour arrived before the behavioural framework did. Site-specific operating procedures, tailored inductions, escorting ratios, the quiet choreography of who challenges whom: these normally precede occupation, and here they are being retrofitted around a working routine that is already forming. That is assurance debt, and like technical debt it compounds. Habits set like concrete - workable for a fortnight, then permanent - and the first month of this arrangement is therefore not a settling-in period to be forgiven its roughness but the single most consequential security window the site will ever have. What is tolerated in week one is policy by week six, whatever the policy says.

Which brings me to the part of this arrangement that genuinely concerns me, because I have spent a career watching it go wrong and advising against it elsewhere. A weekly prime ministerial rhythm, plus Cabinet attendance, plus the officials who follow both, converts the West Coast Main Line into a commuting artery of government - and every commuter carries a load-out: laptop, official phone, security token or YubiKey, the notebook and affiliated Post-its (the most underrated compromise medium in government: unencrypted, unaudited and irreplaceable), and the unauthorised USB stick that should not exist but somehow always does.

Consider first the simple loss problem, and consider it at Euston. Anyone familiar with that station will understand: the platform is announced minutes before departure, several hundred people surge from beneath the boards in a single chaotic mass, and bags, laptops and coats change hands, shoulders and tables in the clamour for unreserved seats (or they are reserved, but not displayed as such yet).

It is, from an asset-control perspective, close to a designed loss event - and the crime statistics bear the intuition out rather than merely dramatising it, Euston sitting at the head of the national table for recorded theft of passenger property.²⁶ We have now scheduled our government to pass through it twice a week, indefinitely, frequently at the end of a long day. That items will go missing is not speculation but documented history: in 2008 a senior Cabinet Office intelligence official left Top Secret assessments of al-Qaeda and Iraqi forces on a London commuter train, where a passenger picked them up and passed them to the BBC,²⁷ and as recently as 2023 a Ministry of Defence official working on Afghan relocations left a laptop screen in view on a train, one of dozens of breaches from a single unit.²⁸ Government kit goes astray in quantity even without a train involved - hundreds of MOD laptops and devices lost or stolen in recent years alone.²⁹ The only open questions here are the rate, and whether the reporting culture surfaces the losses in hours or in weeks.

Then consider the working environment itself. A delayed Avanti West Coast service is a mobile open-plan office in which nobody at the next table has been vetted. Papers will be worked on without privacy screens, because privacy screens are issued as an option and options lose to habit. Calls will be conducted at carriage volume; "I'm on the train" remains Britain's favourite security breach, announced helpfully at the start of each one. Overheard and overlooked material is the cheapest collection there is: legal to gather, deniable, repeatable, and now available on a published timetable - an adversary does not need to follow anyone when the train from Euston follows itself. The tasking here is easier still than the reconnaissance I described in the Omnicrisis assessment, and the price of entry is a standard ticket. The aggregation logic I have argued elsewhere applies in full: no single glimpsed paper, seat reservation or lanyard is classified, and a season of them, assembled, is an operational picture of how this government moves.³⁰

The corridor must be treated as part of the estate, with a travel security policy that has teeth rather than a tone. In practice that means: protectively marked and sensitive material does not travel open on laps (and depending on said sensitivity, at all without the proper and rather stringent handling direction), and the default for the carriage is reading, not producing; privacy screens are issued, fitted and treated like the pass - their absence a reportable matter, not a style choice; sensitive calls are not taken in carriages, full stop, and principals' staff are briefed to end them rather than lower them; kit moves on the "one bag, one body" rule and the bag does not go overhead out of reach; asset inventories are maintained so that loss is visible, and loss reporting is drilled and destigmatised so the YubiKey dropped at Euston on Tuesday is reported on Tuesday rather than confessed on Thursday; travel patterns for staff must be varied and staggered where both the rail and work timetable allows; and every new pass-holder at Heron House receives an induction run like a pre-deployment brief, because that is what it is - with the first challenge at the threshold visibly backed from the very top the first time it is tested, since that single moment will set the site's real policy for years.

None of this requires new money. It requires the same thing as our critical national infrastructure, and every other soft edge I have written about: posture, applied where the risk actually sits.

So, to answer the question I set out with: is Heron House the right fit? For now, yes - it is the only building in Manchester that could have carried this at a week's notice, and it will hold. But "for now" is the operative phrase, and the more interesting answer lies ahead of us rather than behind. The building is the soundest part of the scheme; the vulnerabilities are the human and mobile parts, an unfamiliar population learning a demanding perimeter and a familiar routine forming outside it, and both of those are matters of culture and habit that we still have time to shape before they set.

That is ultimately the hopeful reading, and I will end on it rather than on alarm, because alarm is the lazier note. We are running the government of the North from a hardened borrowed vault while we build the real thing down the road - and the real thing can be built right, secure by design and pitched at the proportionate level, if the lessons of the interim are written down rather than merely survived. Downing Street teaches what it costs to secure a building that never expected to be secured; Ancoats is the rare chance to expect it from the foundation up.

The perimeter of Number 10 North, in the meantime, is not the wall of Heron House. It is the departure board at Euston, the table seat on the delayed 18:15, and the pocket of whoever is carrying the notebook - and no vault, however good, protects contents its couriers hand out on the train. Get the culture right in the first month and the building right by 2032, and this works. Security follows the people, or it does not follow at all.