Washington Broadband has to prove rural reach after the tower or fibre path fails

Summary

• Washington Broadband sold all of its cable and fibre infrastructure to Charter Communications in 2025 but retained the company, its fixed-wireless network and the separate Washington Telco phone business. The surviving access operation is therefore wireless, while former wired customers and plant moved to Spectrum.
• The remaining operator is visibly active. Washington Broadband markets current residential and business plans, says its wireless system has 35 towers and thousands of local customers, and originates 19 IPv4 and IPv6 routes through AS19116. Public routing observations show three upstream networks and a Seattle internet-exchange presence.
• Those facts do not establish physical resilience. No public record reviewed here identifies tower coordinates, site ownership, backhaul paths, common ducts, backup-power duration, sector utilisation, spare radios, field-crew capacity or restoration performance. Three visible upstreams can still share one route out of the valley.
• The practical test is address and failure specific: can an order receive the advertised tier at the busiest hour, and can the same location remain connected when its usual tower, utility feed, backhaul or upstream path fails? Until Washington Broadband publishes that evidence, its operating network merits a medium evidence grade and its redundancy claims should remain unproven.

The cable sale made the wireless network the whole access story

The most important fact about Washington Broadband is not hidden in an obscure registration. The company told customers directly. On 16 June 2025, president Forbes Mercy announced an agreement to sell its cable and fibre internet assets to Charter Communications, the owner of Spectrum. The sale notice says the transaction covered the entire wired infrastructure. It explicitly excluded Washington Broadband, Inc., the wireless network and Washington Telco LLC.

That boundary changed what the company has to prove. Before the sale, a customer who could not be served well by radio might eventually have been moved to Washington Broadband's cable or fibre system. The company's follow-up, "Now What for our Wireless Network?", says that option disappeared with the transfer. It also says management would return its attention to wireless upgrades while continuing to seek a buyer for the remaining business.

By October 2025, Washington Broadband described the transaction as complete. Its "Life After Fiber" update says the post-sale business was back to wireless, operated a 35-tower network from Tampico through the East Valley and from Wenas to Zillah, and served thousands of local customers. The same post says the company remained on the market because its owner planned to retire. These are consequential operating claims, but they come from the seller and are not accompanied by a tower list, customer count, sale filing or performance series.

The distinction between sold and retained assets must be carried through every assessment. Charter now controls the former cable and fibre access plant. Washington Broadband says it controls the wireless system and continues to provide wireless internet. Washington Telco is a separate phone company under common ownership. AS19116 remains registered to Washington Broadband and remains visible in global routing. None of those facts means that every transport strand feeding a retained tower is owned by Washington Broadband. A wireless access operator normally depends on leased fibre, microwave backhaul or both, and the reviewed public material does not identify that ownership line.

That is why the title asks what happens after either the tower path or the fibre path fails. Fixed wireless removes the cable between the final distribution point and the customer, not every cable in the network. Traffic still has to travel from tower radios to aggregation equipment, then across one or more transport links to an internet edge. Even a tower-to-tower microwave chain ultimately needs power and a route into wider networks. Selling the wired access estate can simplify the business, but it does not make the remaining wireless service independent of fibre.

Washington Broadband is therefore not a speculative entrant. Its own long history, current offers and live routing all support an operating regional ISP. It is also not the same infrastructure business that it was before June 2025. The evidence needed now is a post-sale account of the retained system: which facilities and routes still sit inside the operator's control, what has to be leased from others, and how customers are recovered when any one of those dependencies breaks.

A 35-tower claim is a footprint, not a serviceability map

Washington Broadband's public geography is unusually descriptive. The current about page says its network has grown to 35 towers covering hundreds of square miles in upper Yakima County. The October 2025 update names Tampico, the East Valley, Wenas and Zillah. An older about page describes service from Zillah to Naches and says 20 towers surrounded Yakima. The wireless FAQ is older still: one paragraph refers to seven towers, while the installation answer refers to 12 local sites.

Those numbers need not be contradictory descriptions of one moment. They may preserve different stages of a network that the company says it rebuilt end to end four times. The problem is that the pages do not date each inventory or reconcile the old site counts with the current 35-tower statement. A customer cannot tell whether all 35 are distribution sites, whether some are backhaul relays, whether all are active, or whether every site can sell the newest tiers.

The same FAQ makes the physical constraint explicit. It says a prospective customer's home or office must be within sight of a tower, historically within seven miles, and must permit an outdoor antenna installation. Installers test the link before mounting the antenna, run a cable indoors and connect it through a small power unit to the customer's router or computer. If the initial connection cannot be made, the FAQ says the customer is not charged.

That process is more informative than a broad coverage colour. A tower can be close in straight-line distance and unusable because a ridge, orchard, building or tree belt blocks the radio path. A distant site may work from an elevated roof. New construction or foliage growth can turn a formerly clean view into a marginal one. Different frequencies, antenna heights and radio generations alter the result. The only defensible unit of fixed-wireless reach is the individual path under a stated installation design.

The FCC applies a similar discipline to reported availability. Its fixed-broadband guidance says a provider should report places where infrastructure has been built and where it either already has a customer or can complete a standard installation within ten business days without delay or charges caused by extending the network. For terrestrial fixed wireless, coverage areas must be based on specified engineering assumptions, including cell-edge probability, loading and receiver height.

More detailed FCC supporting-data rules require fixed-wireless filers using coverage areas to submit base-station locations and heights, carrier information, link budgets and clutter data. Much of that material is regulatory support rather than a simple public tower map. It shows what has to be known internally before a broad claim can be made responsibly: where the transmitting sites are, what frequencies and power levels they use, how terrain and obstructions are treated, and how much traffic the cell is assumed to carry.

Washington Broadband does not need to publish sensitive coordinates for every cabinet to make its footprint more testable. It could provide an address checker tied to current network data, state which speed family each location can order, distinguish a confirmed installation from a preliminary view, and publish aggregate serviceability by community. It could also explain whether "Fiber Wireless" on the current residential page describes a fibre-fed radio service, a product family, or another architecture. As presented, the term is not a technical diagram.

The 35-tower figure is valuable evidence of scale only as an attributed company claim. It does not establish 35 independent failure domains, 35 fibre-fed sites or 35 sites with protected power. Nor does "hundreds of square miles" equal universal service inside the boundary. Rural reach is made of many narrow paths, and each path has to pass its own test.

Installed radio capacity is not the speed on a rate card

Washington Broadband's current residential plans expose the transition in product form. The page offers four "Fiber Wireless" tiers, from 25Mbps down and 5Mbps up for $39 a month to 150Mbps down and 30Mbps up for $99. It also keeps legacy tiers from 10/2 to 50/20 and separate in-city offers of 30/10 and 40/15. The company says the newer plans can be limited by distance.

Its business page displays a different and apparently older set of wireless tiers: in-city service up to 30/10 and rural service up to 40/20, with installation charges beginning at $149. The mismatch may reflect a page awaiting revision, different business provisioning, or limits on where upgrades are available. Without an address-level order result and a dated disclosure, it should not be flattened into one network-wide maximum.

A plan speed is a ceiling sold to one account. It is not the capacity of the tower sector shared by many accounts, the capacity of the tower's backhaul, or the throughput of the route leaving the valley. A 150Mbps customer can receive the full tier during a quiet test while a busy evening exposes contention among dozens of homes. Conversely, a sector with ample capacity may deliver a poor result to one house because its antenna is misaligned or its indoor Wi-Fi is congested.

Washington Broadband itself identifies several of these layers. The FAQ lists severe weather, blocked sight lines, local power outages, antenna alignment and customer-device problems as possible causes of slow service. Its 2024 discussion of speed-test limitations argues that in-home use, server selection and upload saturation can affect a reading. Those are valid cautions in principle. They do not remove the operator's responsibility to show what its own access and backhaul deliver under load.

The FCC now uses 100Mbps down and 20Mbps up as the benchmark for advanced fixed service. Its 2024 inquiry explains that the previous 25/3 threshold no longer supported contemporary usage and records a longer-term goal of 1,000/500. Washington Broadband's newest two residential tiers meet or exceed 100/20 on paper; the 25/5 and 50/10 tiers do not. The distinction is not a verdict on whether a household can use a lower tier productively. It matters for mapping, public investment and the company's stated aim to prevent subsidised overbuilding by reaching 100/20.

The company said in June 2025 that it had applied for a Priority Access License and wanted protected spectrum to support upgrades. FCC rules give a Priority Access License protection from lower-priority general users within its authorised geographic, frequency and time limits. The reviewed public material did not establish whether Washington Broadband obtained the licence, what channels it can use, which sites have compatible radios or how much customer capacity the change produced. An application is evidence of intent, not installed capacity.

Licensed or prioritised spectrum can reduce one interference risk. It cannot enlarge a congested backhaul link, power an unprotected tower, clear a new tree obstruction, replace a failed radio or create another physical path to Seattle. Nor does one fast sector upgrade every legacy sector. The current plan page is encouraging evidence that 100Mbps and 150Mbps products are being offered somewhere; it does not reveal how many addresses qualify or what proportion of customers remain on older equipment.

The cleanest performance disclosure would show the distribution of busy-hour download speed, upload speed and latency by plan family, together with sector capacity thresholds and the percentage of attempted installations that fail for sight-line reasons. It should separate wired customers transferred to Spectrum from retained wireless customers so that old hybrid-network measurements do not flatter the surviving access system. Installed capacity becomes usable capacity only when the radio path, backhaul, upstream edge and customer equipment can carry it at the same time.

AS19116 proves an active internet edge

The public routing evidence is substantially stronger than the physical-site evidence. The American Registry for Internet Numbers registered AS19116, named WASHINGTON-BROADBAND, to Washington Broadband, Inc. in October 2014. The associated organisation record gives the company's Yakima address. Registry records establish control of internet-number resources; by themselves they do not establish that the numbers are in use.

Here they are in use. On the research date, the RIPEstat overview reported AS19116 as announced. Its announced-prefix inventory returned 19 visible routes, including both IPv4 and IPv6. The observed list included the 104.245.128.0/21 aggregate, more-specific routes within it, several 71.93 and 71.94 blocks, 38.119.167.0/24 and IPv6 space beginning 2606:4680::.

An independent BGP inventory likewise describes AS19116 as an active eyeball network and identifies three current upstream paths: Cogent Communications, Ziply Fiber and Lumen. It records 17 originated IPv4 routes and two IPv6 routes, consistent with the 19-route total observed through RIPEstat. Counts can differ across collectors and moments because aggregates and more-specific announcements change. The important point is not a perfect count; it is that Washington Broadband is actively originating a meaningful address portfolio.

The route-security picture is also positive. The routing inventory marks most of the listed routes as covered by valid route-origin authorisations, and IPinfo's view of 104.245.132.0/22 identifies a valid authorisation for Washington Broadband. That reduces the risk that participating networks will reject those routes under origin-validation policies. It does not protect against a fibre cut, router failure, incorrect configuration or an outage at the authorised origin.

PeeringDB's AS19116 entry adds a view of the interconnection design. It lists a 10Gbps operational port at the Seattle Internet Exchange, with IPv4 and IPv6 addresses, and facilities in Seattle and Portland. It characterises the network as heavy inbound with a 1-5Gbps traffic range and an open peering policy. The network record was last updated in 2022, its facility entries are older, and public profiles can lag real changes. These values are useful leads, not audited utilisation or contracts.

Taken together, the routing records establish far more than a dormant ASN. Washington Broadband has its own internet identity, originates public routes, supports IPv6 and has several visible ways to exchange traffic. That matters for customers because an operator controlling its own routing can shift announcements among upstreams and peers rather than depending entirely on addresses borrowed from one carrier.

This is the strongest part of the company's infrastructure case. It also creates the next question. Logical routing diversity is only as resilient as the physical paths carrying it.

Three upstream names do not prove three roads out of Yakima

Cogent, Ziply and Lumen are distinct networks. A Seattle exchange port and facility references in Seattle and Portland suggest more than one interconnection opportunity. Yet none of the reviewed records shows how AS19116 reaches those places from its Yakima aggregation points. The three routes could leave the valley on separate cables, enter different buildings and terminate on separate routers. They could also share a pole line, conduit, leased transport provider, central office or bridge crossing for much of the journey.

That correlation is the main unanswered transit question. BGP can move traffic around a failed carrier session only if another carrier session remains reachable. If one local fibre cut severs every leased wavelength before the paths diverge, a perfectly configured router sees three upstreams disappear at once. If two upstream circuits terminate on one edge router or one power distribution unit, a device or facility failure can have the same effect. If all tower backhaul converges on one aggregation building, that building becomes the network's real centre regardless of how many towers are drawn on a map.

The 2025 transaction makes the route question more urgent. Washington Broadband said its wired network had been built to connect directly with Spectrum infrastructure and that Spectrum bought all of that wired plant. The sale notice does not say whether Washington Broadband retained any strands, indefeasible rights, transport contracts or building access needed by the wireless system. It also does not say whether Charter supplies post-sale backhaul to any retained tower.

The public interconnection record also needs careful interpretation. Presence at the Seattle Internet Exchange can shorten paths to participating content and access networks, reducing reliance on paid transit for some traffic. It does not provide a route to the entire internet, and a peering session is not a backup transport circuit from Yakima. A Portland facility listing may support geographic diversity, but only if the transport to Portland avoids the failure domains of the Seattle path and remains active.

What would settle the matter is not a list of vendor names. Washington Broadband could state that it has at least two physically diverse long-haul exits, identify the cities where they diverge, disclose whether they use separate entrances and routers, and say how often failover is tested. It could describe the number of tower rings or protected microwave paths without exposing sensitive exact routes. It could publish the outcome of a planned upstream withdrawal and an aggregation-site power test. Those facts would turn visible routing options into evidence of recoverable service.

Until then, AS19116 supports the controlled topic of peering and transit but not a claim of end-to-end redundancy. The company has demonstrated an active edge. It has not publicly demonstrated that the edge can still be reached after the first fibre path fails.

Wireless goes down when either end loses power

Fixed wireless is often described as an escape from vulnerable cables. It is also a chain of powered electronics. A customer antenna normally receives power from inside the home. The indoor router and the small power unit described in Washington Broadband's FAQ need electricity. At the other end, tower radios, switches and backhaul equipment need site power. Aggregation routers and internet-edge equipment need power too.

A local outage can therefore break service even when every radio and fibre remains intact. If the tower has batteries but the customer's home does not, internet access stops at the premises. If a customer runs a router from a small battery but the serving tower exhausts its own reserve, the connection stops at the site. If both survive but the aggregation building or an intermediate relay loses power, the path still fails.

Washington Broadband's FAQ specifically lists power outages around the valley among causes of degraded or failed wireless service. That is candid and technically plausible. The company does not publish backup duration for its towers, whether all sites have batteries, which sites have generators, how fuel is replenished, or whether low-battery alarms are monitored centrally. Its older about page says network monitoring is continuous and gives staffed support hours extending into evenings and weekends. Monitoring can identify a failing site; it cannot extend battery runtime or clear a road for a generator crew.

The local dependency is not abstract. The City of Yakima identifies Pacific Power as its electric provider, while Yakima County's utilities plan says most of the county is served by Pacific Power and describes Benton Rural Electric Association service in part of the region. The same county plan describes several transmission feeds into the valley. A tower footprint spanning multiple communities may cross utility boundaries, but the public network material does not map sites to power providers.

Yakima County's emergency-management guidance tells residents to prepare for wildfire, winter storms and power failure. Its multi-jurisdictional hazard plan also treats severe weather, severe winter storms, wildfire, floods, landslides and extreme temperatures as hazards of concern. These documents do not predict a Washington Broadband outage. They identify credible conditions that can remove commercial power, damage a pole route or restrict access to a remote site.

Federal resilience guidance explains why runtime alone is not enough. CISA's infrastructure-dependency primer notes that communications and energy depend on one another and that shared geography can create correlated failures. Its implementation guidance recommends battery and generator alternatives, system interconnections and continuity planning. These are general principles, not evidence that Washington Broadband lacks them.

A CISA communications power guide is more specific: backup systems need appropriate sizing, fuel, monitoring, regular starting and load tests. Its Midwest derecho case study shows how restoration can be delayed when responders lack site contacts, power information or road access, and recommends multiple backup power and communications paths. Yakima is not Iowa, and Washington Broadband is not a public-safety radio system. The failure mechanics still apply to a rural tower.

The useful disclosure would divide sites by protection class: battery-only runtime, fixed-generator runtime, portable-generator connection, renewable supply, remote monitoring and refuelling access. It would state the minimum runtime rather than the best protected site. It would also explain whether customer phone service depends on a powered antenna and router, because a cloud-hosted voice platform can remain healthy while the local access path to it is dark.

Power resilience is expensive precisely because it must exist at many small places, not only in one central room. Thirty-five towers, if that count represents active sites, create 35 sets of batteries, environmental enclosures, surge protection and maintenance visits. The rural reach claim becomes credible when the weakest of those sites can survive long enough for a crew to arrive.

Weather and foliage can erase usable reach before equipment fails

Not every failure is binary. Washington Broadband's FAQ says heavy snow or fog can affect a link and that trees, new construction or antenna misalignment can block the sight line. These are company statements about its own service, and they are a useful guide to where capacity can become unusable without a total tower outage.

Radio links have a margin between the signal needed for the selected modulation and the signal actually received. Distance, obstructions, moisture, interference and alignment consume that margin. A link can continue to pass traffic by shifting to a slower modulation, which preserves availability but reduces sector capacity. When many links fall back during the same weather event, customers can experience congestion even though no device is broken.

Trees create a slower version of the problem. A successful winter installation can become marginal when leaves return. An orchard or shelter belt can grow into the path. A new metal building can introduce a reflection or blockage. An antenna mount can move under wind or thermal stress. Washington Broadband's site survey reduces initial risk, but resilience depends on keeping path records and revisiting links whose margin deteriorates.

The FCC's guidance for a provider disputing a fixed-wireless availability challenge illustrates the evidence available to an operator. It calls for path analysis, signal levels, interference ratios, distance, antenna height, base-station characteristics, customer-equipment details and a narrative explaining how the advertised speed can be achieved at the challenged place. It also says a provider lacks availability if capacity limits prevent a standard installation within ten business days.

That framework separates three questions that broadband discussions often mix. Availability asks whether the location can be connected. Performance asks what an installed connection delivers over time. Resilience asks whether service survives a defined failure. Washington Broadband's broad footprint and current plans address the first two at a marketing level. They do not answer the third.

Local support labour is part of the topology

Washington Broadband makes local service a central promise. Its current home page presents a team that has served the valley since 1994. The about page says every team member has at least five years with the company and describes the staff as local and bilingual. The older about page says the network is monitored continuously, gives weekday and weekend support windows, and accepts support email at all times.

That continuity matters. Fixed-wireless troubleshooting is geographic knowledge as much as a script. An experienced technician may know which ridge blocks a path, which access road turns impassable in snow, which tower sector serves an orchard and which customer antenna tends to shift in wind. Bilingual support can shorten diagnosis when a customer and field crew need to distinguish a household power problem from a wider outage.

The public claim still lacks the denominator. The company does not state its employee count, number of installers, on-call rotation, average repair time, maximum simultaneous site incidents or dependence on contractors. It does not identify whether 35 towers are maintained by two field technicians or ten. Long tenure is evidence of experience, not evidence of enough hands during a regional event.

The sale and succession plan sharpen the labour risk. The founder's announcements say he is seeking retirement and a buyer for the remaining wireless business. Ownership transfer can preserve a local team, add capital and deepen operations. It can also change staffing, vendor relationships and the unwritten knowledge held by a few people. A buyer needs an asset register, route records, tower access agreements, credentials, spare inventory and maintenance history that can outlive the founder.

Field capacity also has to cover customer equipment. Each new installation involves a path test, antenna mount, cable entry, power unit and router handoff. Each failed radio, damaged cable or shifted antenna consumes a truck visit. Tower work can require specialised climbing, safety and radio skills. General NTIA workforce material highlights tower technicians, installers, splicers and pole workers as distinct broadband occupations. The reference describes national workforce needs, not Washington Broadband's staffing.

The strongest local-support disclosure would publish service targets and actual distributions: time to answer, time to identify a tower incident, time to dispatch, and time to restore by failure class. It would state whether after-hours restoration is performed by employees or contractors and how many concurrent tower failures the team can support. It would identify critical spares held in Yakima, such as customer radios, sector radios, switches, power supplies, batteries and backhaul equipment.

This is not human-resources detail detached from infrastructure. A spare radio in another state is theoretical capacity. A generator without an available driver is theoretical runtime. A second path that no one has tested is theoretical redundancy. Local support labour is one of the paths through which the network recovers.

The post-sale economics favour upgrades but narrow the room for error

Washington Broadband says it has held nominal prices for decades, offers no contracts and imposes no data caps. The current residential range from $39 to $99 preserves a simple ladder while raising maximum speeds. For rural households with few terrestrial options, that proposition can be valuable even when fibre or cable elsewhere in the region offers a higher top speed.

The economics are less visible. No reviewed public source gives wireless revenue, subscriber count, churn, tower lease expense, transport cost, payroll, equipment age or capital spending. "Thousands" could describe a resilient customer base or a small base spread across a costly geography. "Hundreds of square miles" could reflect efficient high-site coverage or expensive long repair routes. A 35-tower system can have very different costs depending on how many structures are owned, leased or shared and how many customers each sector supports.

The cable and fibre sale may have released capital and removed the maintenance burden of wired access. It may also have transferred revenue, eliminated an upgrade path for some wireless customers and changed the terms on which towers reach Spectrum facilities. The company has not published the sale price or a post-sale financial statement, so none of those effects should be assigned a number.

The June 2025 wireless update gives management's own incentives. It says higher speeds are needed to meet demand, to keep the network competitive under the BEAD programme and to increase the value of the business before sale. Those incentives align with customer upgrades but are not identical to resilience. A new radio can raise peak throughput and company value while leaving the same utility feed and backhaul route underneath it.

Washington State is deploying substantial public support through BEAD, with reliability and affordability among the programme's objectives. Washington Broadband does not appear in the reviewed public material as a confirmed awardee, and the company should not be associated with a grant without a specific record. The programme matters because subsidised alternatives can reach areas where existing service does not meet the required availability, speed or reliability terms.

For an incumbent regional ISP, the rational response is not merely to advertise a bigger number. It is to document which locations are truly serviceable, invest where sector and backhaul capacity are tight, and make recovery performance visible. That can protect the operator from incorrect overbuilding assumptions while giving residents evidence stronger than an assertion.

The regional-ISP economics topic is therefore evidence-supported but unresolved. Washington Broadband has a real network and a meaningful local franchise. Its post-sale durability depends on whether retained wireless revenue can finance licensed spectrum if obtained, 100/20-capable upgrades, diverse transport, protected power, local spares and succession at the same time.

Six failure drills would reveal the real network

The most useful way to assess Washington Broadband is not to ask whether it is "reliable" in the abstract. Reliability becomes testable when the failed component is named and the recovery path is observed.

1. A customer loses power while the tower stays online

The customer's outdoor radio, power unit and indoor router stop unless the premises has backup power. The operator should make clear that network-side batteries do not keep household equipment alive. For phone customers, the disclosure should explain what equipment needs power for emergency calling and what battery options exist. The key measurement is how long a typical supported customer setup runs, not whether the cloud phone platform remains reachable elsewhere.

2. A distribution sector fails at one tower

A radio or switch failure can remove service for every customer on that sector while other sectors at the site remain healthy. Recovery depends on remote diagnosis, a compatible spare, site access and a trained technician. If neighbouring towers can serve some customers, the operator should know which antennas can be reassociated remotely and which require realignment. A reported spare count and median sector-restoration time would establish this layer.

3. Commercial power fails at a remote site

The site should transfer to battery or generation, alarms should identify the event, and the operator should know the remaining runtime. A long event tests fuel, road access and simultaneous failures across the valley. The relevant result is the minimum supported runtime and the percentage of sites meeting it. One well-protected core site does not compensate for an unprotected relay.

4. A backhaul route is cut

The tower radios can remain powered and still have nowhere to send traffic. A ring should carry traffic in the other direction; a protected microwave hop should take over; or customers should be shifted to another site. The test must remove the normal physical path, not merely disable one routing session over the same cable. Capacity on the backup path also matters: a route that stays up at a fraction of busy-hour demand may preserve messages while collapsing video and work traffic.

5. One upstream carrier or edge router disappears

AS19116's visible Cogent, Ziply and Lumen paths give Washington Broadband options. A controlled withdrawal should show how quickly routes converge, whether IPv4 and IPv6 both recover, and whether customer sessions survive. A separate test should remove an edge router or facility power feed. Passing the carrier test while failing the facility test would reveal a shared local dependency.

6. A regional event creates cuts, outages and staff demand together

Wildfire, wind, winter weather or a vehicle strike can combine power loss, damaged plant and blocked access. The company must prioritise sites, communicate with customers, deploy spares and maintain the remaining network with a finite team. The CISA derecho study shows why contact lists, site power information and access planning matter before such an event. For Washington Broadband, the proof would be a dated exercise or incident report stating affected sites, restoration sequence, backup performance and lessons carried into the next maintenance cycle.

These drills also identify who bears each consequence. A rural household can lose remote work, school access, telehealth and ordinary communication. A small business can lose payment and cloud systems. A Washington Telco customer may lose the local path to a functioning voice service. Former cable and fibre customers, by contrast, now depend on Spectrum's network and should not be counted as evidence of retained wireless resilience.

No public evidence reviewed here shows that Washington Broadband fails these drills. No public evidence shows that it passes them either. Publishing even aggregate results would be more persuasive than another undifferentiated uptime adjective.

Unofficial signals show stakes, not network truth

Local discussion boards contain both criticism and appreciation of Washington Broadband. Some writers describe fixed wireless as an option of last resort or report that a property could not be served; others say they used the company for years or discuss it alongside Starlink, Ziply and Spectrum. These comments are useful as a map of customer concerns. They cannot establish network-wide speed, uptime or coverage because addresses, plans, equipment, dates and household Wi-Fi conditions differ.

One Yakima discussion says a household exhausted Washington Broadband as an option and remained on another rural wireless service. That suggests sight-line or serviceability constraints can matter at the edge of the footprint; it does not prove why Washington Broadband could not serve the address. Another local thread includes a former customer who later switched providers. That confirms consumer choice exists at some addresses, not that one network is categorically better.

The evidence that would settle these anecdotes is ordinary operating data: completed and failed site surveys by community, trouble rates by tower, busy-hour results by plan, outage duration and reasons for disconnection. A failed installation because no tower is visible belongs in availability statistics. A slow evening on a qualified link belongs in performance statistics. A multi-hour outage after a power cut belongs in resilience statistics.

The company should also reconcile old public pages. Seven, 12, 20 and 35 towers may each have been correct when written, but an undated visitor cannot know that. Older business tiers and current residential tiers may reflect different products, but the distinction is unclear. Cleaning up those inconsistencies would not prove resilience. It would make independent testing easier and reduce the temptation to use anecdote as a substitute.

Unofficial comments should therefore remain signals only. They suggest what customers care about and where to look. They cannot prove tower count, congestion, route diversity, power runtime or service quality.

What proof should come next

Washington Broadband already discloses more than many small operators about its history and strategic transition. The next set of disclosures should be operational rather than promotional.

First, publish a current serviceability surface. An address checker should return the technology, available tiers, whether a site survey is required, expected installation timing and any distance limitation. Community summaries should distinguish locations passed by a usable radio path from the entire area visible on a broad map. Current FCC filings should align with those results, and capacity-constrained sectors should not be reported as available for a tier they cannot accept.

Second, define the retained asset boundary after the Charter sale. Customers and a prospective buyer need to know whether Washington Broadband owns or leases tower structures, shelters, backhaul, aggregation rooms and long-haul transport. The answer can be aggregated to protect sensitive details. The key is to separate owned assets, contracted dependencies and services supplied by the purchaser of the wired network.

Third, describe physical diversity. State the number of aggregation domains, the proportion of towers with two backhaul paths, whether alternate paths avoid the same conduit or pole route, and whether Seattle and Portland connectivity uses separate long-haul corridors. Name the failover tests and publish their dates. Three upstream contracts are encouraging; physical separation is the proof.

Fourth, publish power classes and restoration performance. Report minimum battery duration, generator coverage, refuelling arrangements, alarm monitoring, annual load-test completion and median restoration time by incident type. Include the customer-premises boundary so users understand that their antenna and router need local backup too.

Fifth, show capacity as a distribution. Publish the share of serviceable addresses that can buy 100/20 or faster, busy-hour plan performance, sector upgrade thresholds and the number of customers still on legacy tiers. State whether the Priority Access License was granted and which part of the footprint has compatible equipment. Protected spectrum should be connected to installed sites and observed outcomes, not left as an application claim.

Sixth, make local labour measurable. Give the number of field and network staff or at least the ratio of active tower sites to qualified responders. Publish after-hours coverage, contractor dependence, critical spare categories and the number of simultaneous failures the restoration plan is designed to handle. Succession documentation should preserve route, access and configuration knowledge through any sale.

These disclosures would not require a public engineering blueprint. They would allow residents, officials and a buyer to distinguish a long-running network with tested recovery from one that is merely working today.

The evidence grade is medium, with a resilience discount

Washington Broadband clears the basic operating threshold. The company has a current Yakima presence, current residential offers, a detailed post-sale statement, a claimed 35-tower network and thousands of customers. AS19116 is active, originates IPv4 and IPv6 routes, uses route-origin security and has several visible upstream and peering options. This is considerably stronger evidence than a business name, an ASN registration or a marketing page alone.

The physical record is much thinner. Tower locations and status are not reconciled publicly. Address-level reach is not transparent. The current mix of frequencies and installed radio generations is unknown. There is no public backhaul map, ring count, route-separation statement, backup-power inventory, sector-utilisation series, spare-equipment list, field headcount or restoration report. The 2025 wired-asset sale adds an important unanswered boundary around retained transport.

The correct judgement is neither that Washington Broadband is unproven as an operator nor that its visible internet diversity proves rural resilience. It is an established regional wireless ISP whose public network edge is easier to verify than the access and recovery system behind it.

That asymmetry matters most for customers at the edge of the Yakima Valley footprint. A route can remain visible to the global internet while one tower is dark. A tower can remain powered while its backhaul is cut. A backhaul can remain healthy while foliage makes one home unreachable. A second upstream can exist while sharing the same long-haul trench. And every technical backup can exist on paper while a storm exhausts the available crew.

Washington Broadband has spent decades making rural radio links work where alternatives were scarce. After selling the cable and fibre path, it has made wireless the centre of its remaining value and its succession plan. The company now has to show that the centre holds when the usual path does not.