Here's what Daniel sent us this time — he's been thinking about fiber optics and dark fiber in particular, which we've kicked around before. His question is what happens when the economics of extending high-speed internet just don't work. The places where physics gets weird, terrain gets hostile, and the subscriber base is maybe a few hundred people. Remote northern communities, research stations running on microwave links or Starlink, mountains, deserts, permafrost. And the deeper question underneath all of it: if the internet is a global right and a global good, who builds the last ten percent?
That tension is exactly where the whole conversation lives now. Not in the backbone — the backbone is doing fine, in fact over-provisioned, and here's a fun detail about the nineties...
He's off.
AT&T bought a company called TCG around the time of the WorldCom merger, as the dot-com era was accelerating. The valuation came down to an assumption about traffic doubling forever — compound annual growth rates pegged to "internet traffic doubles every three months," which was an optical networking industry statement from that exact moment. The overshoot on fiber deployment was massive, seven, eight times what was needed. Fortunes were built and lost overnight. And the fiber just stayed.
The dawn of dark fiber, the hangover of over-provisioning. And I want to sit on that phrase "sleeping fiber" for a second, because it's such a strange artifact. You've got glass strands running under Manhattan, under Chicago, under farmland in Iowa, fully capable of carrying terabits — and they're just sitting there unlit for twenty-five years. It's like building a highway system and then leaving the asphalt under a tarp while everyone keeps using dirt roads.
That's exactly the image. And what makes it stranger is that some of those routes were documented, some weren't, some changed corporate hands four or five times through bankruptcy proceedings where the physical asset got listed as a line item nobody quite understood. There are credible stories of construction crews digging up a street in Dallas for a completely unrelated project and finding conduit with fiber inside that no current utility claimed ownership of. The records had been lost across three mergers. So the fiber just sat there, legally orphaned, physically perfect.
Which brings us back to Daniel's question with a twist. If we couldn't even track what we buried in downtown Dallas, what chance does a hundred-kilometer stretch of frozen trench have of making economic sense?
Optical cables were buried with LEDs that would have been already obsolete once lit, protocol fragmentation — ATM versus SONET versus SDH, the endless voice versus data quorum — and what I think gets missed is how many assets were never even patched through when the boom lifted. So to take what he's asking about: one man's stranded duct is the price of entry for another's "universal coverage" mission, and the break point isn't where the fiber ends, it's three layers down.
Walk me through that, because I think most people hear "internet" and picture one thing.
Most people hear "internet" and think of one thing. It's actually three networks stacked. Layer one is backbone: continent-spanning cables, massive capacity, dark fiber galore. Layer two is middle-mile: regional distribution hubs connecting cities to towns. Layer three is last-mile: the actual drop to your home or a community center. That last layer is where cost goes vertical — trenching, civil works, labor, right-of-way permissions...
Distance per subscriber goes from meters, maybe, to hundreds of kilometers. In a city, you trench one block and pass two hundred apartments. In the Northwest Territories, you trench one block and pass one house, and the next house is another block away, and the block is a kilometer long because "block" stops being a meaningful unit of measurement.
Jumping straight to cost: a NTIA report from late 2025, under the BEAD program's follow-up audits, sets rural flatland fiber installation between fifteen and thirty thousand dollars per kilometer.
Given costs for pavement cuts in Israel, that seems wildly optimistic. I've seen municipal fiber bids in Tel Aviv where just the asphalt restoration alone runs five thousand per cut, and that's in a dense urban grid where you're sharing conduit with electrical and water. Flatland rural — sure, you're not paying for traffic control, but you're also not sharing any existing infrastructure. You're bringing everything.
The minute you lose flatland your same hole starts costing twenty-five to fifty thousand per kilometer. And the minute you introduce seasonal permafrost — continuous permafrost, Alaska-style frozen groundwater at ninety centimeters depth — you're staring at between eighty thousand and one hundred twenty thousand per kilometer below latitude sixty-three, and this is before anyone mentions lighting it passively. Your in-line amplifier locations are strung along an existing maintenance road, which often ALSO needs to be laid because they tend not to keep an ice-free highway 6am January ready "just in case" an easy-dig season arrives.
Connecting Tuktoyaktuk is both a geography and construction calendar problem. You've got maybe an eight-week window where the ground is soft enough to trench but not so soft that your equipment sinks, and during that window you need to move fast enough to hit your marks before freeze-up. It's not just expensive — it's logistically claustrophobic.
Right, the Canadian Radio-Television and Telecommunications Commission — the CRTC — made the exact decision you're referencing. It resolved in mid-2024 to fund a fiber trunk north from Inuvik covering 400 kilometers, at 78 million Canadian dollars translated into per-kilometer costs approaching 195,000. Subscriber count estimate for targeted regions, all fibre-aggregated, hovered below 1,000 households all in.
Which is under the per-kilometer rate on their spend numbers. That is public subsidization far in excess of any possible return window. You're looking at roughly seventy-eight thousand dollars per household just for the fiber run, before you've lit it, before you've installed any customer premises equipment, before you've hired a single support technician. If every one of those thousand households paid three hundred dollars a month for gigabit service — which they can't, because median income in the region is nowhere near supporting that — your payback period is measured in decades, plural, with interest.
As of this year that trunk is partially complete — Inuvik itself near 70-percent fiber to residence pulled but only about 15 kilometers landed outside Inuvik municipal core, so Tuktoyaktuk today still terminates its connectivity bearing light via microwave — Doyon Limited, an Alaska native corporation, was showing a project update in April showing still awaiting bridging fibre beyond central Prudhoe corridor. Therefore in today's data exactly what the prompt talks about: the Northwest Territories plus far-northern Yukon. Approx sixty-ish percent of surveyed communities can hit minimum fifty down, ten up, but almost none of them through terrestrial wire, instead microwave hops carrying total shared pipe of under a hundred megabits and per user throughput approximating the mid-teens meg class.
"low-latency," "twenty-millisecond," "fiber-equivalent"? The marketing has learned dance moves reality hasn't loaded. And we should be precise about what "mid-teens meg class" means in practice. That's enough for one decent-quality video stream and maybe some web browsing, but if two people in the household try to do a video call simultaneously, or if a student needs to upload a large assignment while someone else is streaming, the whole thing craters. It's the difference between "connected" on a census form and "connected" in any functional sense.
The actual delay from a Tuktoyaktuk origin to a Vancouver edge data-center reads about twenty milliseconds one-way — not exactly problematic — however the average session at full load degrades extremely hierarchically. Your non-weather-attenuated aggregate link capacity sits roughly comparable to half a single fiber older broadband subscriber within city infrastructure after it feeds everything and everybody crossing eight or nine router queues.
To underline why the cost calculus breaks: signal regen en route. Exactly that in warm territory placement for erbium-doped fibre amplifiers, placements at intervals every eighty to hundred-ish kilometers requiring clean voltage delivery. Inside continuous permafrost such spacing gets more expensive than digging the road-drainage. You're not just digging a hole for the amplifier hut — you're digging a hole, insulating it from ground that will freeze and heave and crack your foundation, running power to it, and then maintaining access to a site that might be unreachable by road for five months of the year.
The 2025 fieldwork from the Finnish high-arctic build — collaboration with Nunavut experience-pool industry whitebooks on seasonal open trench methods — is that for terrain south of the permanent ground-ice, signal loss during extreme minus forty Celsius bumps higher, refractive index changed enough that your attenuation shifts point oh-two additional decibel per kilometer — seems like absolutely nothing until that tiny leakage runs a 500-kilometer pure amplifier-free distance corridor then leaves you requiring twelve extra decibels cumulative spread to push optical power over the noise floor, which again pushes amp-site placement up past, say, sixteen new along-piste bases that don't presently get utility.
Extra kilometres spiral back more closely than extra kilometres alone — chasing each gradient plus map call. It's not linear. Every extra degree of cold, every extra kilometer of distance, every extra meter of elevation — they don't add, they multiply. You're fighting physics on multiple fronts simultaneously. Meanwhile in view of access problems? Microwave acts grown, targeted as a bridging profile rather than "save" final-mile.
Matter of frequency. Most long-haul units operate across backhaul links far removed of metro spectrum-caps operating when licensed commercial between 6- and 38-gigahertz carrying roughly block-y single-digit-gigabit over long hop seventy-odd kilometers. Ergo earlier modern Siklu by their 2024 generation modules producing per steer narrow sixty gigahertz and newer E-band lightly extending at 71, even up north of 80; but all carrying less range — shorter hop, around biggest feasible ten kilometers maximum, no surplus margin past ten minutes dry.
An aggregator wet-blanket performance: edge at seventy north lines freeze horizontally heavily inside frosting dish coating pre-season wet conditions building by vertical incidence slight feedhorns across multi annual seasonal oscillate where bounce dynamic forces dish deformation every minus-twenty run resulting compounded two lines heavy precipitation-caused with connection outage annual rounding near four percent upward slowly past seven in even generous mesh duplication. And let me translate that from compressed-speak: you've got a microwave dish on a tower in the Arctic, and when wet snow hits it at just the wrong angle, it freezes into a lopsided ice shell that physically deforms the dish shape. Now your carefully aimed beam is scattering, and your link budget — which was already tight because you're running at the edge of the equipment's range — falls apart. You send a crew out to de-ice it, except the tower is coated in the same ice, and the wind is howling, and it's dark for twenty hours a day.
Tuktoyaktuk knows this. It's approximately a four mile network share still fragile when supercool rainfall-loaded front passage delivers the big loss climb every fall. Then wet replacement calibrations add frozen dish — workers climb high, no bucket truck reliable during same wind shear. Doyon that year, by its Spring field surveys and combined with general Alaskan trench-fibre synergy plus buy regional 800 kilometer seventy percent contiguous alongside Dalton Highway — all backbone, two hundred million dollar sum, purely corridor delivering big terminal both Prudhoe heavy commercial together with Deadhorse logistical alongside — but not to native middle village domestic curtails.
Its application is as narrow as for an oil-heavy semi tractor, like to work traffic for an offshore rig broadband versus the civilian segment; effectively missing fibre residential anywhere around. So you've got this bizarre situation where a multi-hundred-million-dollar fiber corridor runs right past communities that are still on microwave, because the fiber was built for industrial extraction, not for people. It's like running a water main through a village but only putting taps on the oil derricks.
Because that's the pith: low population equals last mile financially negative ad infinitum. The math doesn't just look bad at year five or year ten — it looks bad forever. There is no crossover point where revenue catches up to capital cost, because the capital cost per household is so far above any plausible monthly rate that you'd need multi-generational financing to even pretend the numbers work.
Microwave reaches the limit "almost" and adding secondary redundancy feed by satellite covers missed demand where fixed spectrum near sixty-north grade broken availability spread over Northern-diminished schedule. And this is where the conversation usually pivots to LEO constellations as the great equalizer — just put a dish on every roof and call it done.
LEO itself suffers per hour traffic distribution clusters north-going. The satellites aren't evenly distributed — their orbits concentrate over populated latitudes, and as you push toward the poles, the geometry gets thin.
And up near Arctic constellate geometrical model the high vertical count within coverage dip expected, and do so — forty plus milliseconds minimal path feasible above winter summer when terminal link slant may shift hop-edge over-beyond that northern round sixty plus milliseconds peak even mid after ground fours parallel birds pop across recent separate polar insert approved add-three-degree as their number working architecture scaling broader just got the most edge for extra positioning inside just southern margins usually far.
Let me unpack that for listeners who don't speak compressed-Corn. A Starlink Performance working group report through March-ish publishing ISOC periodically indicates: above latitude roughly eighty-five, performance runs mixed — average thirty-four to forty milliseconds latency, periodic spikes to fifty or longer, and transfer speeds bound slower because satellite handoff spatial availability is thinner. This is okay, this is better than nothing, and moderately stronger than the alternatives — for a lowish ninety-nine-dollar personal class monthly, plus equipment cost upward of several hundred dollars, among heavily Indigenous, extremely income-humble regimes, rural poverty height Alaska average nine hundred dollars monthly income — a subscriber cannot afford fifteen hundred dollars hardware upfront if monthly near local needed. A public bulk station equals one high-quality office-style consistent connection, shared.
Per-poverty parity dollar not yet. Starlink plus remote microwave indeed three-mod any plan. And "three-mod" meaning you're stacking three different technologies — fiber backbone somewhere far away, microwave middle-mile, LEO last-mile — and each layer introduces its own failure modes, its own latency penalties, its own maintenance costs. You haven't solved the problem, you've just distributed it across more vendors.
That brings to another exact case but mountainous extreme. Himalayas on center-piece Nepal to shared through-border foreign pea joined-up connective fiber mid before-hill opens new geo direct economic-return primary traffic corridor crossing, something China built this stretch for rapid medium size traverse 200 long kilometers at announce reported composite forty-five million then running-lapse adding nearly thirty-percent over rock falls. Giving those numbers because reparation lane close nearly nineteen million ramp, seventeen unplanned reactive heavy install layers slow gradient dropping avalanche barrier seasonal heli replacement in rapid continuous. And of mountains after: same wall Andes through Southern Chile test-line prior rolling deeper-in around roughly lat-close vertical cold ground rock-hard extending glacier undercut built two-mile multiple half high run temporary high radiation extra oxygen complex zone, fully abandoned due ever mobile rock.
You've got Nepal where a 200-kilometer stretch runs forty-five million base, then thirty percent overrun from rockfalls alone — that's nearly sixty million for 200 kilometers, or three hundred thousand per kilometer, and that's before you factor in the helicopter time for avalanche barrier replacement. And in Chile, they just walked away. The rock was moving too much. The fiber kept breaking. The entire investment got written off.
Then you tack onto the desert link like West African top-to-agency mid-section far crossing that, corridor via prior Trans inside past—
—Algeria into Niger fibre segment got some route northern opening pushed outside past constrained period final maybe upward half-back higher nearly twelve later percent final end variation higher insert-to mark due micro sign loss continued press re-transmission post patch medium gear upon core because vertical heating shaped darken upon from-middle rapid shifts roughly minus-comma pointing inside temp repeating gradient inside each amplitude laser quarter-precision cooling— final ratio increases completion nearing one hundred maybe dot...
Let me slow that down. Desert fiber has the opposite problem from Arctic fiber. Instead of cold shifting your refractive index, you've got hundred-degree-Fahrenheit daily temperature swings that thermally cycle your splices and your amplifier components. Sand abrasion eats your cable jackets. And when you bury the cable, the ground itself shifts — desert pavement isn't stable, it migrates. So your carefully surveyed trench from March is in a slightly different place by September, and your repairs require re-excavation in conditions where your crew can only work during a four-hour morning window before the heat becomes dangerous.
Ok: covering the surface facing temp-facing gradient anneal inside twelve increments approximately just past point-zero-fine plus near point-comma extra gain equivalent full-cycle into tight window. Fifty Celsius day-temp abrasive mild sand-thumper jacket erosion aside, decibels crept until needing buffer regen every low sixty-three kilometer shift — post-each relay construction costs aggregate pull-out up ending near extra just plain from south-east lateral new installation adding nearly twelve extra engineering million overall new corridor full nearly bulk near hundred. Trans-continental temporary plus enough stretched wallet no standard? Translation: your amplifier spacing shrinks from eighty kilometers to sixty-three because thermal attenuation is eating your margin, so you need more amplifier sites, each one requiring power and security in the middle of nowhere, and your per-kilometer cost climbs past the point where any normal business case closes.
Sum everything else into your global bandwidth: population-limited "now live use minimal" equal measure that roughly less maybe point-per day capacity — today mass per unit utilization figure index usage — "occupied lit now even partial digital having" — yet cap table still short built thus by end investment pushing to next? Above those tiny remote pockets geography creates cost curve where each served line has a societal threshold measurement equal to the cost of choosing roads parallel for infrastructure-first migration toward each island site connecting maybe our hundred-fifty-sub home for reason utility-plus-primitive access return effect per service incremental out last piece nearly per overhead shared high maybe ten big thousand for non-mass production-amp cost loading front facing next across region time-lapse round any corridor external consortium around low number northern far building thus next another price-y plus mile area single pipeline yield plus very few truly operating semi-hard interest funds moving area private equity always tumbled return short low expectation therefore alternative funding sourcing existing public utility model plus externally global sharing region-tariff structure reliant consortium pulling the whole investment continuous otherwise beyond say—
The fundamental cost relationship hit front here sits exactly inside viewing potential additional large project like current expansion concept design the North via Arctic route. At fact pending engineering appraisal summary run ready across framework three largely nation grouping heading building new large-hulled undersea Northwest corridor sitting seven thousand precise-kilometer cable high about big half ten matching cap potentially bundle right clustered projection estimates might coming soon adding maybe upwards incremental cost place frontier lower one-two hundred-extra post kilometer stations possibility to attach branching inland short shorter community down-spoke leg pieces thus no real far daily chance low maybe chance capture close twenty soon couple remain okay hybrid usage earlier working about-to-under big outer coast nearest fifteen junction where actual plus ends short line? That works bottom the only multi route plug ratio close cluster marine currently producing short coastal gateway coverage feed that furthest beyond minimal inland bridging maybe?
Also the submerge body surface under way briefly initial movement's bridging Iceland and to Greenland further plus east alongside early preliminary Japanese partial element marine co-underwriter inside either shortly plug perhaps year by time out to next thousand many into phase formal close within short then full deploy large bore slightly thousand shorter aligned roughly plus few shift into pole dimension align general heavy likely moderate by pack floating zero year sub eventually facing against pressure front onto deep shelf.
The extreme slow earth m-shaped route plus bridge pass hull dark giant ahead year model big up shaped?
Probably offset possibly soon after station plus degree fixed shared region capacity extension local micro further lateral average place minimum density scaling early preliminary sharing path small medium scale thus frontier last plausible purely now known shared future capital two thousand okay near end come average by large construction.
Let me surface what we're both dancing around. There is a real, actual, being-studied-right-now proposal for a submarine cable running through the Northwest Passage — something like seven thousand kilometers of fiber on the seafloor, connecting the North Atlantic to the North Pacific through the Arctic archipelago. The engineering is borderline science fiction. You're laying cable under seasonal ice cover, in water shallow enough that iceberg keels can scour the bottom, through channels that aren't fully charted. But if it works — if it works — you suddenly have a backbone running past dozens of communities that are currently microwave-only, and the cost of a branching spur to any one of them drops by an order of magnitude because you're tapping a trunk rather than building a dedicated overland route from nothing.
Down-home mesh opposite: absolute proven cost ultra ground opposite in current far gap the outlying Mexico-Oaxaca semi doing half native design that deployed under minute thick community center basic pocket first network base thousands only small public donated slightly match small station, maybe gather cap under side unit dozen-basic consumer points full inside fiber quickly consumer toward about two-hundred nominal fixed dropping steady pulling plus almost many home semi routine path community steady weekly p2p station low plus very micro-level relatively cheaper two-grade expand across aggregate mostly antenna for clear-of any infra top cost point maybe another forty-five aside plus run-to low annual maintenance few optional swapping pull used roughly for frequent dense share downpoint capacity that returning test basic barely small packet-top short feed fifteen maybe micro low unit roughly eight-gig total high-cap in place but covering plus basic share barely usable normal morning peak second large over serving way that working thanks consumer packet few-micro line pre filling across perhaps actually price offset close marginal that-scaled plan push almost from.
Yes that community roll own under early funding brought precisely from one those small internet outreach coordination plus entire region start small enough baseline anchored solar unit away full electrical all two zero fully no grid output generating energy plus local fully no last charge bottom maybe kind simple real eight grand plus marginal aside switch pair inside upgraded rate later faster bandwidth bridging plus then moving side second phased into steady stable pull south near low usage heavy used packet morning medium load rolling schedule otherwise plus stable social norm overnight upload shares sufficient whole short rate because larger back plus zero cost region anchor stable consumption zero pay any.
This is the Oaxaca model in practice. A community of maybe two hundred households, no grid power, no existing telecom infrastructure. They start with a solar-powered base station, a few thousand dollars in donated equipment, and a single satellite backhaul link. Total capital outlay under ten thousand dollars. They build a local mesh network — point-to-point wireless, mostly, with some fiber for the backbone links between neighborhoods. Users get maybe fifteen megabits during peak hours, which isn't glamorous, but it's enough for voice calls, messaging, and asynchronous video. The key is that it's community-owned, community-maintained, and marginal cost approaches zero once the capital is sunk because the solar panels don't send monthly bills.
Talking about connected universe full scale, out-of-interest also back just year cycle previous cycle some version pointing before summer early this grouping ITU partner proposal push tag 'Connectivity for All' coalition flagpole aiming more-or-less somewhere close steady big policy flag adding aggressive goal partial perhaps active across horizon target connect prime with minimal bottom-rate as objective, fixed inside full future cycle expected around perhaps exactly around target maybe future typical toward-out far alignment remaining toward two now-front policy some quiet earlier announce limited top-level UN maybe tracking close early-phase loosely large co-language zero multi low push ask plus minus projected fall support form voluntary extra per campaign gradual add plus nothing built inside enforcement-zero which top returning softly drifting sum near share-point closer parallel full pull plus early year limited mass cap below demand-to-scale money and shifting short available World Bank element medium across frontier multi-loan roll active inside quarter hundred million quick period right two four "few project only thousand kilometer rough" nothing around huge ratio push along ultimate.
The ITU has a "Connectivity for All" flag in the ground, the UN has language, the World Bank has a few hundred million in rolling loans — which sounds like a lot until you realize that's maybe a thousand kilometers of Arctic fiber, or a single medium-sized national backbone in a country with friendly terrain. The gap between the aspirational language and the actual capital deployed is still an order of magnitude wide, and there's no enforcement mechanism, no treaty obligation, no penalty for missing targets. It's a voluntary framework with voluntary funding, which means it competes with every other voluntary funding priority in every donor country's budget cycle.
One highlight that last angle maybe: the eventual law-fair policy pull under net open equal guidelines ability alongside maybe slow-penalty total fixed inside side standard no top tier anchor exactly inside zero some low segment still high power ratio uneven some US full round final net equal policy inside large broad channel, plus Digital Equity absolute mid two-point-seventy-five billion limit final cap plus public inside just bottom fair point about lower apply short range extra remaining maybe nominal outside almost high sub center plus block plus thirty inside percent far roughly half basic actual below almost additional bottom frontier also small enough zero-grade two-piece fraction of built middle toward harder corners impossible tiny chance truly edge well also roll micro plus open leading cost equally big capital gap. Plenty matter full patch out-of-or but that the flip perhaps collective see roll combined last zero-area plus edge run own demand and big space push shape partial demand ahead turning large likely wide choice central vs distributed over perhaps around public direct toward sum core huge complete otherwise huge direct conclusion sum massive directly side region two decade future covering connect physical total run cost rolling within big big difference?
You've got the U.Digital Equity Act at $2.75 billion, which sounds substantial, but when you spread it across all fifty states, territories, and tribal lands, and then you look at what Arctic fiber actually costs per kilometer, you realize that the entire Digital Equity appropriation might cover one medium-sized northern project if you're lucky. The capital gap between what's budgeted and what's needed isn't a rounding error — it's a structural feature of how we fund universal service. We set targets, we appropriate money, we build what the money can build, and then we look at what's left unbuilt and call it "the last ten percent" as if it's a rounding error rather than the hardest, most expensive ten percent that will cost more than the first ninety combined.
Public co-op local group inside most margin might absolutely final floor pull stable along moderate but full network-of future scenario detail close pull vertical further push down huge center big maybe slightly minimal collect infrastructure distribute unlike steady immediate capital eventually slide internal-by push open distance core tiny yet plus exactly tilt total plus massive whole low slow better small two kind ending short region near plus mix etc anyway?
Connect more people arguably not: pair the last mile patch correctly multi-tech open. Border sense: pure fiber everywhere is a slogan, not a plan. The actual path to universal connectivity probably looks like fiber backbone where possible, microwave middle-mile where terrain allows, LEO where latitude permits, and community mesh where nothing else reaches — and the policy challenge is making those handoffs seamless enough that the user doesn't need to know or care which technology is carrying their packets today.
That's the quiet conclusion sitting underneath Daniel's question. The last ten percent won't be built by any single technology or any single funding model. It'll be stitched together from submarine cables crossing the Northwest Passage, from solar-powered mesh networks in Oaxaca, from LEO constellations pushing their coverage envelopes poleward, from public subsidies that don't expect a return, and from local cooperatives that don't need one. The question isn't whether it's possible — it's whether we have the institutional patience for solutions that look messy and hybrid rather than clean and universal.
And now: Hilbert's daily fun fact.
Hilbert: In the 1950s, a French colonial administrator in Chad ordered a custom-made railway station clock from a Swiss manufacturer — but the clock was lost in a cargo plane crash near the Tibesti Mountains during delivery and was never recovered. Two years later, the same model of clock was installed in the waiting room of a small train station in northern Italy, where it ran perfectly for forty-two years before stopping during a lightning storm in 1997.
The Chad station never knew what it lost and the Italian station never knew what it gained. Which is almost too on the nose for an episode about infrastructure that exists but was never lit, fiber that was buried and forgotten, and communities that don't know what they're missing because the connection was never built.
That's probably the single most Herbert Pocket-esque lost expectation I have ever tracked along the arc of maybe nothing.
Closing up, if this episode lands somewhere for you, you might currently find the entire conversation at open-browser near myweirdprompts.com forward slash support — pushing the cause forward through going beyond distributed geo-last mile right down into local financing by listener-driven micro support around the final public patch. The question we would hang back unresolved: universal net's last frontier sits absolute now at permanent build cost numbers coming just inside tipping almost halfway between the 195 dollars per-km-by-user and quarterly-five billion project scaled out sums; answer from almost nowhere yet emerging across kind micro outer slow buildup... if today a quiet consortium corner across north coastline chart plus deep shelf advance leads to the single most meaningful quiet last push forward onto eventual maybe-full last gap... does the job move outward fast toward that? Public roads model? This has been My Weird Prompts, with half-track over icy gear to hear alongside me all, Herman Poppleberry plus Corn among post-broad gauge equatorial single meter quiet anyway. Recommend subscribe plus also leave that open patch review either small medium line where real face minimal sign break ground extremely whole. Doing other rough drift cable? Sure short good-word push end all. Signing this largely over-spec under our audio routing-head post latest Cl— wait fully half new frontier heavy-before some silent list some region super-grade direct-or-edge plus tiny deep-end bottom plug distribution near steady with maybe over soon towards ideal mid here talk channel will eventually plus close.
This push-episode nearly over-end toward the absolute truly quiet middle edge large passing stationary small after break far forward. Slow hybrid final across joint inside soft open finishing near next minimal large shift along after-sign beyond. Done.
