WISP TECNOGER advertises 750 Mbps fibre, but its public record cannot size one relay's blast radius

Summary

• WISP TECNOGER is demonstrably active: it markets fibre packages up to 750 Mbps, holds AS273093 and an IPv6 /32, originated six IPv4 routes and one IPv6 route on 16 July 2026, and declares an operational 10 Gbps port at NAPVE Valencia.
• That evidence does not reveal a recoverable physical network. No public record identifies a current TECNOGER relay, line-of-sight path, tower-access arrangement, alignment margin, backup-power runtime, compatible spare-radio pool, physical upstream handoff, field-crew depth or household count behind one failure point.
• Public BGP observations show two strong adjacent networks, with IPv4 concentrated through Tecnoven and IPv6 concentrated through Gold Data. Those are logical paths, not proof of separate circuits, facilities, power domains or survivor capacity.

The relay that goes dark before dawn

At 4:17 in the morning, imagine a relay on a mountain shoulder above an Aragua valley. The scene is deliberately hypothetical: no public record reviewed for this article identifies such a WISP TECNOGER site. The lattice is still upright. The dish still points roughly across the valley. Yet a tired DC supply has fallen out of tolerance, or wind has moved the mount just far enough that the far end can no longer hold the link.

Nothing in the village below necessarily looks broken. A fibre drop can remain physically intact. An optical network unit in a house can still have mains power and show familiar lights. A local switch can still pass frames inside a building. But if that imagined relay is the only upstream path from a neighbourhood aggregation point, the useful internet path ends at the dark radio. The next relay, fibre cluster or access node in the chain may be powered and healthy while being isolated.

How many households sit behind that single point?

The honest answer is not a number. It is an unresolved dependency. WISP TECNOGER's current homepage invites prospective customers to check whether service is available at their address, but it publishes no node-to-premises map. Its residential page names an ONU and 100 metres of fibre, but not the feeder, splitter, OLT, backhaul or relay that would carry those customers upstream. A recent LinkedIn post uses the broad phrase "thousands of homes", but gives no dated total, no municipality split, no access-technology split and no count behind any physical asset.

One relay might therefore be irrelevant to current residential service, because the company may no longer use radio in that part of its network. It might carry only a private link. It might backhaul a fibre cluster. It might be one protected leg among several. The reviewed evidence cannot distinguish those cases. It cannot establish that the imagined relay exists, much less count the homes behind it.

That uncertainty is the central finding. WISP TECNOGER is visible where modern networks are easiest to see from outside: offers, number resources, route announcements and exchange declarations. It is almost absent where resilience is decided: at mounts, batteries, access permissions, spare shelves, fibre junctions, edge routers, handoff rooms and dispatch queues. The public record proves a live operator. It does not yet prove that a failed physical point can be bypassed or repaired before the households behind it exhaust their tolerance for being offline.

A fibre storefront under a wireless name

The name invites one assumption and the shopfront supplies another. "WISP" commonly signals a wireless internet service provider. WISP TECNOGER's present public offer, however, is emphatically fibre. The homepage lists residential, small-business and enterprise service, describes fibre-optic connectivity and advertises 400 Mbps for $30, 550 Mbps for $35 and 750 Mbps for $45. Each displayed residential package includes an ONU, 100 metres of fibre and unlimited connectivity, with extra fibre charged by the metre. The dedicated residential page repeats the same three packages.

This appears current rather than an abandoned web page. The site's machine-readable homepage record says the page was modified on 26 September 2025; the residential-page record carries the same modification date. The company's LinkedIn profile also says it connects Venezuelan homes and businesses over fibre in the central region.

Those facts support a narrow conclusion: fibre is the company's current public retail message. They do not support the stronger conclusion that every customer access, inter-town span or aggregation path is fibre. An operator can sell fibre to the premises while using radio elsewhere in its transport chain. It can also retain a historical name after migrating away from wireless. No current company page identifies a radio package, frequency, sector, tower, rooftop, microwave hop or subscriber antenna.

There is one weak counter-signal. A publicly uploaded contract copy bearing the WISP TECNOGER brand mentions field inspection, access to roofs and ducts, antennas, UTP and PoE. It also describes provisional equipment during some warranty replacements. Yet the copy contains subscriber information, is not hosted by the regulator and gives a tax identifier that conflicts with repeated reproductions of the authorised-provider list. It can justify asking whether wireless installations remain in service. It cannot establish a current relay inventory, a tower, a service-level result or even a clean legal-identity chain.

The correct treatment is therefore asymmetric. Fibre access is marketed and recently updated. Wireless operation is possible but unlocated. A relay-chain analysis must proceed as a resilience test, not as a description of known TECNOGER plant. Every radio-specific statement in that test is conditional unless the company publishes the missing site and link data.

A live operator with an invisible physical layer

Thin physical disclosure does not mean dormant operation. Several independent signals point the other way.

The company website was updated in late 2025. Its LinkedIn page carried recent vacancies for technical support, administration and other roles. One Turmero support advertisement asked for experience with IPv4 and IPv6, MikroTik, VSOL OLT equipment and SmartOLT, while describing real-time monitoring and remote fault handling. That is a labour-market signal of an active fibre and network-support operation, though it is not proof that every named platform is installed or that the vacancy was filled.

A customer application called WispHome appeared in April 2026. The Google Play address ties the package to a debt-review and payment-reporting function, while a Chrome-Stats mirror records version 1.1.1, an April 2026 update and 15 downloads in its snapshot. An independent AppAgg mirror reports the same purpose and release window. Fifteen downloads are not a customer count; the mirrors can lag the store. The significance is more modest: the company had a newly distributed customer-account tool in 2026.

The network evidence is stronger still. AS273093 was broadly visible at the publication-date observation, and WISP TECNOGER had an operator-maintained exchange profile created in late 2025. These are not the traces of a company that exists only in an old business listing.

But operational presence and physical resilience are different propositions. A route can remain visible while a village access segment is down. A payment application can work from third-party hosting while the provider's access network is impaired. A support analyst can see an alarm without having a safe route to the site or the correct replacement radio. The public signals establish activity; they do not expose the recovery machinery.

Three geographies that must not become one map

WISP TECNOGER's public geography comes in several incompatible forms.

First, the company's contact page gives two customer-facing points in Aragua: Calle 1, Casa No. 31, Sector La Casona in Turmero, and C.C. Galería Las Villas in Villa de Cura. It even publishes coordinates for both. These are useful anchors. They are not labelled as network operations centres, OLT sites, warehouses, towers or dispatch bases.

Second, a December 2025 public reproduction of CONATEL's authorised-provider list names WISP TECNOGER, C.A., RIF J-500680945 and habilitation HGST-00766, with Zamora and Santiago Mariño municipalities in Aragua and Juan Germán Roscio in Guárico. A December 2024 reproduction carries the same three municipalities. The repetition improves confidence that the entry was not a one-off transcription. The original regulator-hosted file was not found, so the copies remain secondary. More importantly, authorised geography is permission, not proof of complete installed coverage.

Third, number-resource records add different cities. The ARIN-derived record for 38.183.212.0/22 shows a Cogent-parent reassignment to WISP TECNOGER with a Maracay customer address. The record for 38.129.89.0/24 gives Valle de la Pascua in Guárico. Those are registry fields attached to address resources. They do not locate routers, customers or service nodes.

Fourth, recent LinkedIn vacancies mention San Juan de los Morros, Turmero and Catia La Mar. A hiring location may mark a real commercial or labour need, but it does not prove a live access network, an office or a field crew at that point.

Finally, the company's exchange profile places a logical interconnection in Valencia. That does not draw a physical route from Valencia to any of the other cities.

Joining these layers would produce a persuasive but fictional network map: offices in Turmero and Villa de Cura, authorised municipalities around them, address records in Maracay and Valle de la Pascua, a Valencia exchange port and vacancies farther afield. The evidence does not supply the lines. There is no attributable inter-city fibre, microwave hop, pole route, handoff building or service polygon. The physical map must stop at separate points with different labels.

That is especially important for the relay question. Mountainous or valley terrain can make line-of-sight engineering consequential in general, but no public path profile places a TECNOGER radio between any two of these locations. The commissioned relay scene is a useful test of what would need to be known, not a picture of a verified route.

AS273093 is real and current

The most authoritative identity chain begins with LACNIC. Its RDAP record for AS273093 marks the autonomous system active, registered on 27 September 2023, and connects it to WISP TECNOGER through contact details that align with the public website. The company also appears in a 2024 LACNIC electoral roll, another narrow corroboration of registry membership.

LACNIC separately assigns the company IPv6 block 2803:2050::/32, also registered on 27 September 2023. On the IPv4 side, two blocks from Cogent's parent address space are reassigned to WISP TECNOGER: the /22 registered in December 2023 and the separate /24 registered in December 2024.

At 08:00 UTC on 16 July 2026, the dated RIPEstat routing-status snapshot reported six IPv4 prefixes covering 1,280 unique addresses and one IPv6 prefix equivalent to a /32. IPv4 was seen by 326 of 326 responding RIS peers; IPv6 by 319 of 321. RIPE NCC's methodology documentation is explicit that these are observations from route collectors. They show broad control-plane visibility at a timestamp, not customer uptime or packet performance.

The announced-prefix dataset resolves the route set: 38.183.212.0/22, its four component /24s, 38.129.89.0/24 and 2803:2050::/32. Counting the aggregate and its more-specifics as separate physical networks would be wrong. They are routing choices over overlapping address space.

Origin authorisation is also in place. RIPEstat reports a valid ROA for the IPv4 /22, permitting more-specifics to /24; a valid ROA for 38.129.89.0/24; and a valid ROA for the IPv6 /32, permitting more-specifics to /48.

These are meaningful strengths. Registered identity, current announcements, broad visibility and valid origin authorisation all reduce uncertainty about whether a real network is operating. None says where the edge routers are, how they are powered, which customers use the addresses or how a failed relay would be bypassed.

Different prefixes tell different stories

The publication-date routing view does not show one uniform external path.

RIPEstat's AS-neighbour observation found two strong left-side neighbours: AS269730 and AS394684. It also found AS273904 on the right side of AS273093. Registry records identify AS269730 as TECNOVEN SERVICES CA, AS394684 as GOLD DATA USA INC, and AS273904 as AM. CONNECTIONS, C.A..

The terms left and right describe the direction in which the adjacent AS appeared in collected paths. They do not settle whether a relationship is paid transit, peering, customer service, resale or another arrangement. This article therefore describes them as observed adjacent networks, not contractual partners.

The BGP-state snapshot shows how differently the routes appeared. After repeated origin prepends are disregarded for the purpose of identifying the immediately preceding external AS:

• 38.183.212.0/22 had 380 collected paths: 377 immediately through AS269730 and three through AS394684.
• Each of the four component /24s had 381 paths: 378 immediately through AS269730 and three through AS394684.
• 38.129.89.0/24 had 380 paths, all immediately through AS269730 in that snapshot.
• 2803:2050::/32 had 364 paths: 329 immediately through AS394684 and 35 through AS269730.

This is an unusually clear protocol and prefix asymmetry. Public IPv4 visibility was heavily concentrated through Tecnoven, while public IPv6 visibility was heavily concentrated through Gold Data. The separate IPv4 /24 showed no Gold Data path in the timestamped view.

What can that support? It supports the claim that AS273093 had more than one observed external AS relationship and applied, received or participated in different routing outcomes by prefix and protocol. It suggests that loss of one relationship need not have identical effects on every route.

What can it not support? The counts are not traffic shares. Three paths out of 380 do not mean three circuits, three percent of traffic or three units of reserve. A collector path is a view through the internet, and many views can converge on one physical handoff. Conversely, a single adjacent AS can deliver service over several protected circuits. The snapshot also cannot show a dormant or private backup.

The asymmetry therefore sharpens the resilience question rather than answering it. If AS269730 disappears, can every IPv4 route move to AS394684? Is there enough capacity? Does the separate /24 have an accepted alternate policy? If AS394684 disappears, does IPv6 converge through AS269730 at useful scale? Are the two relationships delivered to separate routers and facilities? No dated failover result, contract or physical diagram settles those questions.

The prefix timeline also needs restraint. In the July observation window, the aggregate and four component routes shared a gap between the 9 July 08:00 and 10 July 00:00 collection points, while the separate 38.129.89.0/24 showed additional intermittent intervals. That pattern is not an outage report. It may reflect collector visibility, a route change, a withdrawal or another data condition. There are no customer probes, packet measurements, incident notices or restoration records with which to classify it. Calling the interval a TECNOGER service outage would repeat the same error as turning an AS path into a fibre route.

For resilience, a genuine route event would still be only the beginning of the inquiry. Investigators would need to ask whether customers lost traffic, which access areas were affected, whether one protocol behaved differently from the other, what physical or policy change occurred, and how much capacity remained. The public timeline can identify a period worth examining. It cannot supply the incident.

BGP is not a recovery map

Two common mistakes are tempting here.

The first is to treat the parent of an address block as the current physical upstream. WISP TECNOGER's IPv4 space sits inside Cogent's 38.0.0.0/8, and the registry records contain Cogent-maintained route objects. Yet Cogent was not the immediate external AS in the dated public path snapshot. Address provenance, routing-policy registration and live path observation are different things. None alone identifies a physical circuit.

The second is to draw the AS path on a geographic map. A path ending AS269730 AS273093 says that collectors observed Tecnoven immediately before TECNOGER in the control plane. It does not say whether the connection is fibre or microwave, where the handoff is, which company owns the local loop, how many routers are involved, or whether the delivery shares poles, ducts, buildings or power with the Gold Data relationship.

This distinction reaches all the way back to the hypothetical relay. A local relay can fail while AS273093 remains globally announced through healthy edge equipment. Conversely, all customer access links can be intact while an edge or handoff failure removes the routes. BGP sees reachability between autonomous systems. It does not enumerate the physical points between a household ONU and the place where another network accepts TECNOGER's traffic.

The existence of multiple logical paths is good news compared with a single observed adjacent AS. But resilience requires a chain of physical and operational facts that BGP cannot supply: separate delivery, adequate survivor capacity, working convergence policy, independent power and a healthy customer path to the surviving edge.

The 10 Gbps fact that cannot carry the conclusion

WISP TECNOGER's PeeringDB profile supplies the most specific public interconnection-rate figure in the record. It lists one operational 10,000 Mbps interface at NAPVE VLN, with IPv4 and IPv6 addresses and route-server participation. The row was created on 29 September 2025.

The exchange itself is identified by PeeringDB as NAPVE VLN in Valencia. NAP VE's own description names Valencia alongside Caracas and Maracaibo, while its policies say multilateral peering through route servers is the default and bilateral peering is permitted.

This is useful evidence. It establishes an operator-declared 10 Gbps exchange interface, not merely a generic ambition to peer. It also fits the broader route record: AS273093 has two strong adjacent networks and valid dual-stack announcements.

It is still not total capacity. A 10 Gbps port can carry local peering, route-server traffic, private interconnection or some combination; the profile does not publish actual traffic. It does not show the rate of any transit circuit, the capacity of customer aggregation, the path used to deliver the port to TECNOGER, or the amount held in reserve. A port labelled operational can also sit behind one entity router, one delivery circuit or one power domain.

There is an intriguing overlap. Tecnoven's PeeringDB profile also lists NAPVE VLN, at 100 Gbps, as well as NAP VE ports in Maracaibo and Caracas. That makes Valencia a plausible venue for some interaction, but not a proven venue for the observed AS269730-AS273093 adjacency. Gold Data's PeeringDB profile lists a much wider exchange and facility footprint, including a facility entry in Venezuela. None of Gold Data's facilities can be assigned to TECNOGER.

Most importantly, adjacent-network scale is not TECNOGER reserve. Tecnoven's 100 Gbps exchange port and Gold Data's external footprint may strengthen those networks. They do not reveal what TECNOGER bought, where it connects, how much it uses or whether its two logical paths survive the same regional fault.

The 10 Gbps figure therefore belongs in a narrow box: declared interface rate, normal-state status, one exchange. It cannot answer the household question, prove two physical upstreams or establish useful capacity after a relay, edge or delivery failure.

Six tests for a recoverable relay

A recoverable wireless link is not simply a radio that works on a clear day. It is a chain in which geometry, access, alignment, power, replacement equipment and upstream delivery remain manageable when something goes wrong. The following tests are general engineering questions, not claims about known TECNOGER assets.

1. Is there a verified line of sight with measured margin?

The current ITU-R P.530 recommendation sets out propagation data and prediction methods for terrestrial line-of-sight systems. Its scope summary covers clear-air and rainfall effects and the mitigation needed to design availability. Applying that framework requires link-specific inputs: endpoints, heights, frequency, path profile, climate and equipment characteristics.

No such TECNOGER record was found. There is no published pair of relay coordinates, antenna height, frequency, channel width, path clearance, fade margin, modulation history or availability calculation. Without them, a reviewer cannot distinguish a comfortably engineered link from one that works only under favourable conditions. Nor can the reviewer tell whether vegetation growth, construction or a shifted mount would consume the remaining margin.

2. Can a crew reach and legally access the site?

A radio can be perfectly diagnosable from a monitoring screen and still remain offline because the technician cannot reach the mount. The public record does not identify whether any current TECNOGER radio sits on an owned tower, a leased tower, a rooftop, a utility structure or a third-party site. It gives no gate access, key-holder, climb-permission, landlord, safety or after-hours arrangement.

The branded contract copy mentions access to subscriber roofs, ducts and antennas, but that is not a tower-access framework and its provenance problems prevent strong reliance. The two contact points in Turmero and Villa de Cura cannot be treated as crew origins. A small distance on a map would not settle repair time anyway; the relevant variables are dispatch readiness, road access, weather, permission and safe working conditions.

3. Can alignment be restored and verified?

The opening's misaligned dish is plausible as a generic failure mechanism, but no TECNOGER alignment incident is documented. To recover such a link, a crew would need the correct mounting knowledge, safe access, a compatible measurement method and a reference for the intended path. The public record contains no antenna make, mount type, polarisation, alignment tolerance, commissioning record or stored baseline.

An operator may hold all of this privately. The evidentiary point is that "rapid maintenance", as claimed on the about page, cannot be translated into a restoration estimate without the link and labour details. Remote monitoring can identify degraded signal, but it cannot tighten a bracket in high wind.

4. Does power last long enough for response?

The website's 24-hour attention language concerns people, not energy. No source identifies mains feeds, rectifiers, batteries, UPS systems, generators, fuel, site load or tested runtime at any TECNOGER access, relay, aggregation or edge location. The residential offer also does not state whether the customer ONU has backup power.

Power dependency can create two different failures. A central relay or aggregation site can go dark and isolate many powered households. Alternatively, the outside network can remain healthy while individual homes lose mains and their ONUs switch off. Without site and customer power data, neither the affected population nor the restoration sequence can be estimated.

5. Is the replacement equipment compatible and nearby?

The public contract copy's provisional-equipment language is the only specific spare-related signal, and it appears in a warranty context. It does not identify network radios, antennas, PoE injectors, power supplies, SFPs, OLT cards, mounts, cables or quantities. A provisional customer device would not restore a failed backhaul radio or edge router.

Compatibility matters. A spare with the wrong frequency range, connector, licence, firmware, mounting pattern or power requirement may not be operationally spare at all. No public inventory, storage location, replenishment time or failure-rate history exists. The correct public status is not "no spares"; it is "spare capacity unknown".

6. Does the surviving upstream path begin before or after the failed point?

This is the test that joins the physical relay to BGP. AS273093 has two observed adjacent ASes, but a second AS relationship protects households only if traffic from those households can still reach it. If both external sessions sit behind the same failed relay, fibre cut, router, building entrance or power system, logical diversity arrives too late in the chain.

The inverse is also possible. TECNOGER may have separate protected delivery that is simply not public. The 10 Gbps NAPVE port may be reached over a resilient service. Prefixes may converge cleanly after one relationship fails. None of that should be denied without evidence.

ITU-R's availability recommendation for fixed wireless links recognises that real links may use linear or redundant topology depending on provider needs. The decisive word is not redundant but real: the topology, capacity and test result must be tied to the operator's actual links. For TECNOGER, those records remain private or absent.

Across all six tests, the public answer is consistent. Normal operation is supported. Recoverability is unmeasured.

The human part of the relay chain

Networks are repaired by people, and the public labour signals are more substantial than the physical ones.

WISP TECNOGER's about page says it has trained personnel, an operations centre, 24/7/360 attention and rapid maintenance. LinkedIn shows 11 visible employee profiles and a self-entered company-size range of 51-200. Recent posts advertise roles in several locations. The Turmero technical-support vacancy describes remote fault follow-up, real-time network monitoring and advanced network configuration, with knowledge of IP addressing, MikroTik, VSOL OLT and SmartOLT.

These details support an operating support function. They also show why headcount cannot be read from a platform. Eleven visible profiles are not 11 employees; the 51-200 range is not audited. Neither figure says how many people can splice fibre, align a microwave dish, climb safely, repair power, configure the edge or respond outside normal hours. Contractors may add capacity. Vacancies may reveal gaps or growth. The evidence does not decide which.

The WispHome application adds a customer-service channel for debts and payment reporting. It does not appear to be an outage or dispatch system. A customer who can report a payment is not necessarily able to see an incident, estimated restoration time or affected area. No public status page, outage archive or repair-performance series was found.

The field-response question is therefore one of concurrency. One trained team with the right part may repair one accessible fault quickly. The same organisation can struggle when a power event, fibre damage and a misaligned link occur in different places at once. Public marketing says the company is always available; public evidence does not show the rota, vehicles, test equipment, site permissions, specialist coverage or spare stock needed to convert availability into simultaneous restoration.

The restoration clock also has several separate stages. An alarm must be detected, classified and assigned. A remote operator must decide whether the fault is at the customer edge, optical access, radio path, site power, edge routing or an external handoff. A field team then needs permission, transport, safe conditions and the correct equipment. If the fault crosses a landlord, utility or adjacent network boundary, another queue begins. A fast first response does not guarantee a fast repair, and a promised published contact points does not reveal mean time to restore. TECNOGER publishes no dated series for any of these stages.

That gap matters economically. Regional providers often compete through local attention and lower prices. The current $30-$45 packages are attractive retail statements. The cost of resilience, however, sits in capacity that is not normally sold: a second route, idle ports, charged batteries, duplicate radios, stocked optics and technicians who can be dispatched before the next installation. None of those reserves can be measured here.

Capacity has four incompatible meanings

WISP TECNOGER's public record contains several large-looking numbers. They answer different questions.

The 750 Mbps package is a marketed subscriber rate. It does not state the minimum delivered rate, contention, busy-hour performance or the capacity of the PON tree behind it.

The 10 Gbps NAPVE interface is a declared exchange port rate. It does not show actual traffic, a transit commit, the rate of the delivery circuit, or the proportion available to customer internet service.

The 1,280 unique IPv4 addresses in the routing-status snapshot are address space. They are not 1,280 customers, devices, radios or megabits. The IPv6 /32 is vastly larger mathematically, but its size says nothing about active endpoints.

The hundreds of collected BGP paths are views from route collectors. They are not independent circuits and cannot be summed into capacity.

A resilience assessment needs a fifth number that is not public: usable throughput after the largest relevant failure. That calculation would require normal traffic load, the rate and physical independence of each external path, internal aggregation constraints, prefix convergence, customer distribution and the capacity of the surviving access and backhaul network.

The same distinction applies inside the fibre layer. An ONU in a 750 Mbps offer says little about the OLT port, split ratio, feeder, optical budget or occupied capacity. No OLT count, PON standard, port inventory, splitter map or utilisation series was found. For possible wireless links, no channel width, link rate, modulation, airtime, sector load or fade margin was found.

The categories should remain separate:

• Marketed: three residential fibre rates.
• Registered: an ASN, IPv6 allocation and two IPv4 reassignments.
• Announced: six IPv4 routes and one IPv6 route.
• Observed: broad route visibility and two strong external AS adjacencies.
• Declared: one operational 10 Gbps exchange interface.
• Unknown: installed access capacity, sold load, reserve, backup power and failure-usable throughput.

The network may be well provisioned. The public numbers do not demonstrate it.

The household number remains unknowable

The article began with one dark relay and a simple question: how many households lose service?

No available denominator survives scrutiny. The website's household counter is not a usable published total. LinkedIn's "thousands of homes" phrase is broad marketing, not a dated account count. App downloads are not customers. IPv4 addresses are not subscribers. Municipal authorisation is not homes passed. A 750 Mbps plan is not a connection count. A /32 is not a population.

Even a verified company-wide subscriber total would not be enough. A relay blast radius is a dependency count, not a customer total. It requires knowing which ONUs, radio clients, businesses, public sites or downstream nodes use that point and whether an alternate path exists. The same customer may depend on several serial components; several customers may share one passive or active component.

Nor should "household" be expanded into invented social consequences. Internet failure can interrupt work, education, payments, communications, cameras and cloud services, but no public evidence identifies specific TECNOGER customers or critical institutions behind a particular asset. The safe conclusion is narrower: any premises whose only usable path crosses the failed point would lose external connectivity until routing, power, alignment or equipment is restored.

This is why physical transparency matters. Aggregate network statistics can make a regional operator look either larger or more redundant than it is. A small, sanitised dependency map could answer more: node classes, approximate service clusters, protected versus unprotected links, and the number of connections behind each failure domain, without publishing sensitive coordinates.

Until such a record exists, assigning even a range to the hypothetical relay would be fiction.

What proof would change the answer

WISP TECNOGER does not need to publish sensitive engineering drawings to demonstrate resilience. A bounded, dated disclosure would materially raise confidence.

For the access layer, it could separate active fibre and fixed-wireless connections by municipality, state how many OLT or radio-site failure domains exist, and publish ranges for premises behind each. A sanitised map could omit exact coordinates while showing whether major clusters have alternate paths.

For wireless links, it could publish link classes rather than secrets: access versus backhaul, protected versus unprotected, monitored signal margin, inspection interval and whether a compatible spare is stocked. A dated summary of path revalidation after storms or construction would show that line-of-sight is maintained rather than assumed.

For site recovery, the useful evidence would be access rights, after-hours procedures, safe-work constraints, spare-equipment classes and the number of concurrent field incidents the organisation can support. Actual customer names and technician identities are unnecessary.

For power, the company could give tested runtime bands for customer-independent active sites, the date of the last load test, and whether generators or battery swaps can be delivered before depletion. "24/7" support would then connect to a physical clock.

For the external edge, a useful disclosure would distinguish each logical AS relationship from its physical delivery: separate or shared carrier, separate or shared entrance, separate or shared router, separate or shared power. It could publish the committed rate and a range for peak utilisation without exposing commercial terms.

Most valuable would be a dated failure exercise. Withdraw one path, record which prefixes converge, measure the time, and state the throughput available on the survivor under representative load. Test a site-power loss, a relay replacement and an OLT or feeder incident. Publish the result and remediation without revealing exploitable detail.

These are ordinary operational artefacts of a recoverable network. Their absence from public view does not mean the work is absent inside the company. It means an outside assessment must stop short of claiming resilience.

Active, routed and still physically unproven

WISP TECNOGER is not merely a name in a thin directory entry. It sells current fibre packages, maintains public contact and customer-service surfaces, recruits technical staff, holds its own ASN and IPv6 space, originates broadly visible routes with valid RPKI, and declares a 10 Gbps exchange interface in Valencia.

The evidence grade falls at the point where packets become infrastructure. No source identifies the relay in the opening. No source counts the households behind it. No source shows whether current radio links exist, how they are aligned, who can access them, how long their power lasts, which compatible spares are nearby, where upstream relationships become circuits or how much capacity remains after a failure.

Two observed adjacent networks are better evidence of logical choice than one. They are not a picture of two independent roads. Fibre retail offers are stronger evidence of present access technology than the company name. They are not proof that radio has disappeared from transport. A 10 Gbps port is a meaningful installed interface. It is not an end-to-end resilience figure.

The verdict is therefore split. WISP TECNOGER's normal-state operation is supported at a Medium evidence level. Its physical recovery chain remains Weak. The company may operate a robust network, a fragile relay chain or a mixture in which some towns and prefixes are better protected than others. The public record cannot choose among them.

For the household below the hypothetical dark relay, that distinction is everything. A recoverable network is not defined by how many routes are visible when all systems are healthy. It is defined by whether the remaining path, power, equipment and people can still carry the household's traffic after one physical point fails.