Eight o'clock in Polash Nagar

Every evening at eight, the internet in a few blocks of Mirpur Section 11 formally slows down. This is not an outage or a secret; it is printed on the seller's own rate card. The entry package at polashnagor.net costs Tk 550 a month and delivers 20 Mbps between two in the morning and eight at night, then 10 Mbps from eight in the evening until two — the exact six hours when the neighbourhood comes home, turns on Facebook and starts streaming. Six more packages climb the same ladder to Tk 2,000, and each one makes the same honest confession: the speed you are promised is highest when you are asleep and halves when you actually want it.

The company that publishes this card is named after the place it wires. Polash Nagar is a dense pocket of Pallabi thana, tucked between Mirpur's Section 11 avenues and the Karbala ground, and Polash Nagar Dot Net is, in the most literal sense available to commerce, the neighbourhood's network: its registered address is a house plot at 84/1, Polashnagor, Section 11, Pallabi, inside its own coverage map. That map, published on the company's site, is a list a rickshaw could traverse in twenty minutes: parts of Mirpur 10 and Mirpur 6, and the lanes of Mirpur 11 — Avenue 5, Madina Road, Polash Nagar itself, Karbala, Adorshonagor, the Lalmatia tempo stand, the Palashi residential area and a cluster the page calls the sangbadik plot, the journalists' allotments.

Payment runs on the same street-level logic. The customer terms are blunt: bills are collected on a prepaid basis, and a line that is not paid "will be blocked without any prior notice", taking up to twenty-four hours to revive once settled. The subscriber must give the operator space and a power socket for a switch in the stairwell if asked. When a customer leaves, the company "will take back cables and other devices". There is a card-payment page with SSLCommerz logos bolted on, but the terms read like a ledger book kept by someone who knocks on doors in the evening — which, in this market, is exactly how the money moves. The refund policy is one sentence long: the service is non-returnable and non-refundable.

It would be easy to file all this under informality and move on. The argument of this piece is the opposite: the informality is the business model, and it has a price attached to every part. The name, the prepaid cash cycle, the halved evening speeds, the stairwell switches — each is a rational answer to an input stack in which the regulator sets the retail ceiling, licensed gateways set the wholesale floor, transmission monopolies tax every kilometre, and the only variable a small operator truly controls is how many paying households hang off each metre of cable it has permission to string.

The proprietor behind the place-name

Reconciling this company's identity takes patience, because a Bangladeshi neighbourhood ISP produces almost no corporate paper — and where it does, the paper is revealing. The deepest identity record is not a company filing but an internet-registry one. The APNIC allocation for the network's address space, 103.179.198.0 through 103.179.199.255, is described in the registry as belonging to "gazi hafizur rahman t/a polash nagor dot net" — an individual, Gazi Hafizur Rahman, trading as the network's name. This is a sole proprietorship, not a limited company. Bangladesh's companies registry has nothing to say about it because proprietorships do not register there; their entire corporate existence is a city-corporation trade licence. The site's footer duly carries one, TRAD/DNCC/095313/2022, issued by Dhaka North City Corporation in 2022, while the about page cites an older-format number, 05-56175 — most plausibly the same licence before DNCC renumbered its series, though nothing public confirms the mapping, and the discrepancy is worth recording as a loose end.

The regulatory record is stronger. The Bangladesh Telecommunication Regulatory Commission's national register of thana-tier licensees — the "ISP (Upazila/Thana) License List as on 18-12-2024" — carries the entry plainly: Polash Nagar Dot Net, thana Pallabi, address 84/1 Polash Nagor, licence number 14.32.0000.702.46.385.20.210 dated 10 November 2021, next renewal due 4 March 2023. Read that last date twice. As of the register's own December 2024 print, the renewal date had passed by twenty-one months, and the licensee was still listed. It is not alone: the same document is strewn with renewal dates from 2023 and early 2024 attached to operators the register continues to acknowledge. The paper register, in other words, runs years behind the street, and the Commission has since moved the whole apparatus behind a login: from May 2025 licences are issued digitally through the LIMS portal, which offers no public search. Attempts to pull the older lists from the Commission's previous file host now return a dead-domain page, and the legacy site refuses connections outright. The most current public proof that this specific business exists as a licensed ISP is therefore a December 2024 register with a March 2023 renewal date in it — an ambiguity that is itself a fact about how this tier of the industry is governed.

Two more documents complete the spine. On 17 January 2022, APNIC issued AS149449 — routing identity "POLASH-AS-AP" — together with the two /24 address blocks and an IPv6 allocation, all registered to the proprietorship at the Pallabi address. And on 7 February 2022 the Commission signed the document that matters most to this article's economics: a four-page tariff approval letter addressed simply to "Proprietor, Polash Nagar Dot Net", which the company hosts on its own website because the letter's fifth condition orders it to. A sole proprietorship with no published accounts, no companies-registry file and no press coverage thereby ends up publishing a regulator's letter as its principal corporate document. For a researcher, that letter is a gift; for the proprietor, it is the fence around his prices.

The timeline these records draw is tidy: the domain polashnagor.net was registered on 13 November 2020; the licence number carries a 2020 series and a November 2021 date; the WordPress site went up in September 2021 (its feed's first entry, the default "Hello world!" post, is stamped 21 September 2021); the routing identity and address space arrived in January 2022; the tariff approval followed within three weeks; the DNCC trade licence is a 2022 print. Whatever informal form the network took before 2020 — and Dhaka's lanes are full of cable operators that predate their paperwork — its formalisation happened in one push, timed precisely to the arrival of the regulatory regime described next.

The rate card the regulator wrote

In June 2021 Bangladesh's telecom regulator did something few governments attempt: it fixed the retail price of fixed broadband for the entire country. Under the "One Country, One Rate" directive, no ISP anywhere may charge more than Tk 500 a month for 5 Mbps, Tk 800 for 10 Mbps, or Tk 1,200 for 20 Mbps, effective from 1 September 2021 and set for an initial five years. The stated purpose was equalisation — a village customer paying what a Dhanmondi customer pays — and the same-day press coverage records the launch as a minister-led event with Tk 500 as the national anchor price.

Every licensed ISP then had to apply to have its own package list approved within that frame, which is how Polash Nagar Dot Net's letter came to exist. The approved table in the February 2022 letter is the firm-specific version of the national rate: a minimum of 10 Mbps for a maximum of Tk 500, then 12 Mbps at Tk 600, 15 at Tk 700, 20 at Tk 800, 30 at Tk 1,000, 40 at Tk 1,200, 50 at Tk 1,500 and 70 Mbps at Tk 2,000 — twice the speed per taka of the national floor, a proportional adjustment the directive explicitly permits — bound to a maximum contention ratio of 1:8, meaning the operator may sell each unit of upstream capacity to at most eight subscribers at once. The letter attaches a service-grade ladder (a thana ISP at the top grade owes 99% uptime, at most one cumulative day of downtime a month, and four-hour restoration) and a rebate rule with real teeth: five continuous days of outage halves that month's bill, ten days cuts it to a quarter, fifteen days makes the month free. It closes with the two conditions that define this company's public face: the approved tariff must be published on the ISP's website, and nothing outside it may be sold at all, on pain of action under the telecommunication law.

Now hold the storefront against the approval. The live package list is anchored by "Exclusive-1" at Tk 550 — 20 Mbps off-peak, 10 Mbps during the 8pm-to-2am peak — and runs through Tk 650, 750, 1,050 and 1,550 tiers on the same split-speed pattern, alongside a parallel ladder from Tk 600 to Tk 2,000 whose peak-hour speeds are even thinner (Tk 600 buys 6 Mbps in the evening; Tk 2,000 buys 20). This price pair is the article's headline metric, because it compresses the whole economics into two observable numbers: the regulator approved 10 Mbps for at most Tk 500; the shop sells a package that delivers 10 Mbps in the hours people use it for Tk 550, with the 20 Mbps figure living only in the off-peak window. The Tk 50 difference and the sixteen-hour asterisk are what squaring a capped retail price with uncapped evening demand actually looks like on a Mirpur rate card. Whether the Commission ever approved the split-speed variants as "proportional adjustments" is not publicly verifiable — the letter's annexe lists plain tiers — and the question is less pedantic than it sounds, since condition six of the letter makes unapproved products a legal exposure in themselves.

The rest of the card is the standard sweetener stack of Dhaka broadband: unlimited "4K YouTube & Facebook", IPTV, a "largest FTP server", a public IP for every customer, "fastest BDIX speed" and a free optical-fibre connection. Each line has an economic function that the unit-economics section will price, but one deserves immediate scepticism: AS149449 does not appear in the membership list of the BDIX internet exchange, so whatever exchange-fabric speed the company delivers arrives second-hand through its wholesale suppliers. The claim is not necessarily false — cached and exchange-borne content genuinely does move faster and cheaper than international transit — but the company does not own the fact it advertises.

Two gateways, a transmission duopoly, and the price of every input

What Polash Nagar Dot Net buys, and from whom, is unusually visible for so small a firm, because Bangladesh has chosen to license and tariff every layer above it. Routing data shows the network with exactly three neighbours, all upstream: its dominant transit comes from Apple Communication Ltd (AS139901), a private international internet gateway — an IIG, the licence class permitted to land international bandwidth — with a second, lighter feed from Rego Communications Ltd (AS149994), an IIG licensee since 2012, and a marginal third path via Windstream Communication Limited. Two independent gateway feeds is precisely what the tariff letter's top service grade demands ("multiple upstream redundancy"), so the network's shape can be read straight off its compliance obligations.

Those inputs have regulated prices. When the Commission launched the retail cap in 2021 it also, for the first time, fixed what the layers above could charge. The contemporary record — the Commission's own schedule has since fallen off the public web in the portal migration, so these figures rest on the Financial Express's report of the launch, and should be read as press-carried rather than document-carried — put IIG bandwidth at Tk 365 per Mbps per month for a 50–100 Mbps commitment in Dhaka, easing to Tk 330 above 10 Gbps. The same decision tariffed the transmission layer: the nationwide telecommunication transmission network operators — NTTNs, the only entities licensed to build long fibre, a market dominated by two private firms — may charge small buyers Tk 200–300 per Mbps per month for modest metro links, with the rate falling steeply at capacity (Tk 35–50 per Mbps in the multi-gigabit tiers). An ISP may not simply trench its own fibre across the city; it must lease from the NTTN layer at these prices, or stay small enough not to need to.

Stay small enough not to need to. That sentence is the company's strategy. A network whose whole footprint fits inside one thana barely touches the NTTN meter: its point of presence, its distribution splitters and its subscriber drops all live within a radius a technician can cycle across, and its single expensive hop is the local-loop connection to its IIG suppliers. The neighbourhood name is not a branding whim; it is the map of the cost structure. Every street the network does not cross is transmission rent not paid.

The licence itself is almost free, by design. The Commission's ISP licensing guideline prices the Upazila/Thana tier at Tk 5,000 to apply, Tk 25,000 to acquire, Tk 10,000 a year to keep, Tk 25,000 to renew at term, against a Tk 25,000 bank guarantee — a total entry cost of well under one thousand US dollars. A district licence costs four times as much to acquire, a nationwide one twenty times. The result is exactly what the fee table predicts: the December 2024 register carries roughly 1,750 thana-tier licences nationally, and the pages covering Dhaka name Mirpur-addressed operators dozens of times over — Pollobi Online sits in the register one line below Polash Nagar Dot Net, licensed for the same thana. The regulator has made the licence cheap and the retail price fixed, which moves all real competition into the one arena neither document governs: the buildings.

One more input deserves its line: the address space. The proprietorship is a direct APNIC resource holder — a local internet registry in its own right, with 512 IPv4 addresses and an IPv6 /32 whose two halves it announces to the routing system — rather than renting addresses from an upstream. That costs a recurring membership fee in Australian dollars and buys independence: the network can change wholesale suppliers without renumbering a single customer router. For a firm this size it is a deliberate, slightly sophisticated purchase, and it quietly contradicts any picture of the operation as purely improvised.

One wireman's arithmetic

Now assemble the block-level sums, keeping evidence and inference strictly separated. The evidence: retail prices from the company's published card (Tk 550 to Tk 2,000, with the volume plausibly at the bottom tiers); the approved ceiling of Tk 500 for 10 Mbps and the 1:8 contention cap from the tariff letter; wholesale bandwidth at Tk 330–365 per Mbps from the press-carried 2021 schedule; licence economics from the guideline; and a hard registry bound — 512 public IPv4 addresses — from APNIC. The inference: subscriber counts, package mix and staffing, none of which any public document states, and all of which are flagged as estimates in what follows.

Take a hundred subscribers as the unit — roughly one wireman's serviceable patch in a dense Mirpur block, a figure offered as inference from the coverage list's granularity, not from any filing. At the observable package mix, blended revenue plausibly sits near Tk 650 per subscriber: Tk 65,000 a month for the patch, about $530 at mid-2026 exchange rates.

The bandwidth line is where the regulated numbers start to grind against each other. Those hundred subscribers hold peak-hour commitments of roughly 10–12 Mbps each — call it 1,100 Mbps of simultaneous promise. At the letter's maximum contention of 1:8, the operator must stand behind about 140 Mbps of real capacity; bought at the Dhaka IIG rate of Tk 365, that is Tk 51,000 a month — seventy-eight percent of the patch's revenue, before a single salary, splitter or spool of drop cable. On those numbers the business does not exist. Yet it exists. The reconciliation has three parts, each visible in the evidence. First, scale pooling: contention is engineered across the whole network, not per block, and a few hundred subscribers pooled together need proportionally less headroom than any hundred taken alone. Second, the cheap-bit channel: the packages advertise unlimited YouTube, Facebook, FTP and IPTV precisely because that traffic — the bulk of an evening's load — is served from national caches and exchange fabrics reached through the same wholesale suppliers at a fraction of transit price; the marketing sweeteners are, economically, a device for steering demand onto bits the operator can afford. Third, the split-speed card itself: halving every package at 8pm is a contention ratio adjustment executed in public, cutting the peak promise — the only expensive promise — in half while leaving the advertised number intact for the sixteen hours when capacity is nearly free.

With those mechanisms, a plausible steady state (inference, again) has the whole operator buying a few hundred Mbps of blended transit and cache transport for something like Tk 150–250 per subscriber per month — twenty-five to forty percent of blended revenue. Labour comes next: a couple of wiremen, someone on the phone, someone collecting — in a proprietorship, several of these people are the same person and one of them is the proprietor. Then the drops themselves: "connection charge free" means the operator eats a four-figure taka cost of optical cable, connector and subscriber-end optics on every installation, recovered only across months of Tk 550 subscriptions — which is exactly why the terms police churn so hard, blocking unpaid lines within the month, charging Tk 500 to change time-slot plans inside three months, and physically reclaiming cables and devices at disconnection. The regulatory line is a rounding error by comparison: the Tk 10,000 annual licence fee spread over even five hundred subscribers is under Tk 2 a month each.

And there is a ceiling written into the marketing. Every package promises a public IP address, but the registry says the proprietorship owns 512 of them. If that promise is honoured literally, the subscriber base cannot much exceed four hundred and change once infrastructure takes its share; if the base is materially larger, the promise has quietly become shared addressing for most customers. Either way, the two /24s put a rare, hard, publicly verifiable bracket around how big this business can be — and both readings are consistent with a network that IPinfo's traffic profiling classifies as a small consumer eyeball network with a pronounced day-night rhythm peaking around midnight. Sum it honestly: at a few hundred subscribers, total revenue sits in the very low millions of taka per year — a few tens of thousands of dollars — and the margin left after bandwidth, labour and drops is a living, not a fortune. The asset being accumulated is not cash flow. It is the wired block itself.

The asset is consent

What does a neighbourhood ISP actually own? Not spectrum, not ducts, not buildings. Its balance sheet, if anyone drew one, would be a coil of aerial fibre, some switches in stairwells, a rooftop point of presence — and then the real asset, which no registry records: standing permission. Permission from a few hundred landlords to staple cable along their walls and screw a switch beside their meter boards, permission renewed implicitly every month the building's tenants stay connected, permission that the customer terms formalise from the operator's side ("the Subscriber is obligated to provide space and power outlet"). In Polash Nagar the network's name performs this function too: a business named after the neighbourhood is making a hostage of its own reputation, in a market where the alternative name for the same trade — "the cable guy from Section 11" — carries no such bond.

That consent economy has city-scale politics above it. In late 2020, when Dhaka South City Corporation began cutting overhead cables to beautify the streets, the ISP and cable-TV associations threatened a daily nationwide shutdown of internet and television until the drives stopped, putting their two months of destroyed plant at an estimated Tk 20 crore; ministers brokered a truce under which operators would move their own cables underground. The episode is the last mile's property regime in one story: the wires hang by administrative tolerance, the tolerance is periodically revoked for aesthetics, and the industry's only defence is collective — the threat that a city without its informal cables goes dark. A one-thana operator has no individual voice in that negotiation; it inherits whatever settlement the associations reach.

Which makes the company's associational quietness noteworthy. The Internet Service Providers Association of Bangladesh lists 2,196 members in its directory, with a dedicated associate class for thana licensees; a member search for the company's name returns nothing (the only "Polash" in the directory is a Malibagh firm whose representative happens to carry the name). Mirpur even has its own local trade body, the Mirpur ISP Alliance, whose website was unreachable when checked for this article — the attempt is recorded here because in this industry a dead website, as the next section argues, means less than it would elsewhere. Non-membership under the trading name is not proof of isolation (associations list entities under whatever name the cheque carries), but it fits the wider pattern of a business that formalised exactly as far as the tariff regime forced it to, and not one step further: licensed because bandwidth suppliers require it, tariff-approved because the law demands it, APNIC-registered because renumbering is expensive — and absent from every voluntary register.

The association layer matters for one more reason: elections. ISPAB's leadership contests — the current cycle's voter rolls for 2025–2027 circulate as public lists, and Mirpur's segment of an earlier roll floats around document-sharing sites — decide who sits opposite the regulator when the five-year tariff terms, the NTTN rates and the next cable-cutting drive are renegotiated. A thousand thana operators' margins are set in rooms their proprietors will never enter. That is not a complaint; it is the price structure of being small in a corporatist system, and it belongs in any account of this company's risks.

What the wires say that the shop window doesn't

The most instructive evidence about Polash Nagar Dot Net is the contradiction between its two public surfaces, and it is worth reading carefully, as signal rather than as anecdote.

The shop window is derelict. The website's TLS certificate — checked directly against the server for this article — was issued in December 2022 and expired on 17 January 2024; two and a half years later it has not been replaced, so every modern browser interposes a security warning between the company and anyone trying to read its rate card. The WordPress feed behind the site records two "test" posts in April 2024 and then, on 24 April 2024, a Russian-language slot-machine spam post — the classic residue of a compromised, unmaintained installation. A menu page meant to describe the network's points of presence contains nothing but lorem ipsum filler. The payment page's how-to copy refers, verbatim, to "Dot Internet's automatic billing system" — text carried over from a much larger Dhaka ISP whose template evidently seeded this site. The whole storefront runs on a Namecheap shared-hosting account an ocean away from the network it advertises.

The network, by contrast, is demonstrably alive. Its four prefixes were being announced to the global routing table in the first days of July 2026; its APNIC abuse contact was re-validated — a task requiring a human to answer the registry's mail — as recently as May 2026; its billing runs on a hosted subscriber-management platform at a dedicated selfcare portal of the kind Bangladeshi ISP software vendors sell by subscription; its traffic, profiled by IPinfo, breathes with the daily rhythm of real households. The Facebook page the site links to sits behind a login wall and could not be read for this article, but its existence as the primary named contact channel — together with two mobile numbers and a WhatsApp line in the footer, none of them matching the third mobile number in the APNIC record — completes the picture of where this company actually lives: in messaging apps, phone calls and doorways.

What does the contradiction suggest? Three readings compete. The generous one: the website was always a compliance artefact — built in September 2021 to satisfy the tariff letter's publish-your-prices condition and to look credible to the regulator and suppliers, then abandoned once it had served, because no customer in Polash Nagar has ever chosen an ISP through a browser. The neutral one: the business is stable but stretched, spending its scarce attention on the plant that earns and none on the vitrine that doesn't. The dark one: the operation is winding down, and the live routing merely reflects contracts not yet expired. The balance of evidence favours the first two over the third — abuse-contact validation in May 2026 and actively announced address space are affirmative acts, not inertia — but the register's lapsed renewal date keeps the third reading from being dismissed. What would settle it is mundane: a current licence-renewal record from the Commission's new digital system, a post with a 2026 date on the Facebook page, or a fresh tariff filing. Until one of those surfaces, the honest position is that the company's network is verifiably operating while its formal standing is verifiably stale, and that in Bangladesh's thana tier this combination is common enough that the register itself, as noted, is full of it.

Who could take the block

The competitive question for a neighbourhood franchise is never "who has better prices" — the one-rate regime has flattened those by law — but "who else can get permission for these particular walls". Within Mirpur the answer is: many, in principle. The thana register lists Pallabi and Mirpur-addressed licensees by the dozen, from Pollobi Online next door in the register to area brands like IST and MirpurNet that advertise across the district. Above them sit the citywide and nationwide operators; Dot Internet's current card — 50 Mbps for Tk 890, 100 for Tk 1,260 — shows what scale buys: at Polash Nagar Dot Net, Tk 1,050 purchases 40 Mbps that becomes 20 after eight o'clock, so the big-city operator sells roughly twice the evening speed per taka. On paper, the neighbourhood shop is uncompetitive against any large entrant that bothers to string cable down Avenue 5.

The defence is everything that does not appear on paper. An entrant must negotiate the same landlords, staple cable over the same courtyards, and then displace a supplier who lives at 84/1, answers a WhatsApp message at ten at night, and collects payment in a currency — patience with a late week, a free day after an outage, credit until Eid — that no national brand's billing engine recognises. Switching, for the customer, means a new drop cable through the window frame, a new office to argue with, and the social cost of dropping the neighbour. These frictions are the moat; they are also the ceiling, since exactly the same frictions stop Polash Nagar Dot Net from expanding into the next thana, where it would be the stranger. Hence the steady state the registry data implies: hundreds of micro-territories across Dhaka, each too small to attack and too small to grow, consolidating only when a proprietor retires or sells the subscriber list — the "storefront sold with its customers" transaction that is this industry's real M&A, conducted without a single filing.

The substitutes bear watching more than the competitors. Mobile data is the loudest: Bangladesh counted 119 million mobile-internet subscriptions against under 15 million fixed ISP-and-telephone lines in May 2026 on the regulator's own table, and every improvement in 4G pricing bids away the marginal single-person household. But the same table shows the fixed base growing — 12.88 million in January 2024 to 14.95 million in May 2026 — because a family that streams in the evening cannot live on gigabyte-priced radio. The neighbourhood ISP's product, unlimited evening video at a flat Tk 550, is precisely the thing mobile tariffs cannot profitably match. The truer long-run threat is structural: if the state's fibre programmes or a consolidated national ISP ever make Grade-A service ubiquitous at the capped price, the only remaining differentiator would be the doorstep relationship, and relationships, unlike networks, do not survive a proprietor's retirement.

The facts that would move this judgement

This article's judgement is that Polash Nagar Dot Net is a real, operating, licensed-but-administratively-stale micro-ISP whose economics are best understood as a neighbourhood franchise: regulated prices above, tariffed inputs below, and a defensible but unexpandable asset — wired consent in a few Mirpur blocks — in the middle. Several discoverable facts would change that reading, in either direction.

The licence record could resolve. A renewal entry for licence 14.32.0000.702.46.385.20.210 in the Commission's post-2025 digital system, or the company's appearance in any refreshed public register, would clear the largest single uncertainty here; conversely, formal cancellation — the Commission periodically publishes lists of revoked thana licences — would convert this profile from going-concern analysis to an obituary with a live network attached. The scale bracket could break: any evidence that the operator has adopted shared addressing at scale (a reverse-DNS restructuring, a jump in per-address user density in traffic datasets) would void the 512-address ceiling and with it the smallness assumption that anchors the unit economics; an application for more IPv4 space, visible in APNIC delegation files, would say the same thing louder. The input stack could shift: the five-year terms of the 2021 retail and wholesale tariffs expire in 2026, and the Commission has been reported weighing a doubling of the base speed at the Tk 500 price point — if the retail cap tightens while gateway and transmission rates hold, the arithmetic above loses its thin margin, and the association politics this company sits outside of will decide how much of the squeeze reaches Pallabi. A taka that keeps sliding against the dollar presses the same wound from the wholesale side, since international bandwidth is ultimately priced in hard currency however the tariff schedule denominates it.

Softer facts would matter almost as much. A cleaned website with a live certificate would signal reinvestment; a For Sale rumour in the Mirpur trade — these travel through the associations and the billing-software vendors long before any document exists — would signal the endgame; the appearance of the trading name in ISPAB's member rolls would signal a proprietor buying a seat at the table before the tariff renegotiation. And a single primary document — one bKash statement, one bandwidth invoice, one subscriber count on a regulator's per-operator table — would replace the carefully labelled inferences in the wireman's arithmetic with numbers, which is the upgrade this profile most wants. The judgement stands ready to be revised by any of them; that readiness, in a market this thinly documented, is the only honest confidence on offer.

Evidence register