The real question isn't whether investing is risky. It's which risk you can't undo.
A careful board is right to ask hard questions before spending $1.2 million. So let's ask them — and answer them with numbers, honestly, including where the evidence is thin. There are two risks on the table, and only one of them is reversible.
The numbers on this page come from a full Monte Carlo simulation — every variable, distribution, and assumption is laid out (and runnable) on the companion page, How We Modeled the Risk →
Two risks — and they are not the same size or shape.
Every careful decision-maker weighs the risk of acting. The trap is forgetting to weigh the risk of not acting just as hard. Here, the two are sharply asymmetric: spending is a bounded, recoverable bet; letting the network lapse is an unbounded, one-way door.
If the District invests (~$1.2M)
The risk is bounded — and recoverable.
Worst case: payback takes longer. You still own the asset.
- ~99% chance the county comes out ahead (Balanced case) — and a ~88% chance the District itself recovers the $1.2M; even in a deliberately pessimistic world the county still wins ~82–97% of the time (our three scenarios, below).
- The money buys a durable public asset on the same seven towers — it isn't spent and gone.
- If it underperforms, the District still has the network, the customers, and the option to adjust. Nothing is destroyed.
If the District lets it age out
The risk is unbounded — and irreversible.
Once it's gone, it does not come back on command.
- The seven local providers — all resellers who own no towers — lose their product and fold. That competitive layer does not return.
- Customers lock into a captive satellite that prices by how few options you have — and don't switch back.
- Rebuilding later costs many times more than maintaining now (~$40K/home vs. ~$1.2M). The sunk public money is simply lost.
This is the heart of it for a fiduciary: genuine risk-aversion means avoiding the irreversible downside, not the small recoverable one. Three things follow from that — and together they flip which choice counts as “cautious.”
1. A “pause” is how a public asset gets quietly lost. Networks like this rarely die by a vote to sell them. They die by neglect — stop reinvesting, let the gear age out, and the decay itself manufactures the conclusion that “we have no choice but to let it go.” A pause isn't a neutral wait; it's the first step down that path.
2. The District already owns this network — so abandoning it is the radical move, not funding it. Twenty-five years of towers, fiber, customers, and trained crews, largely paid for. In a decision like this, the irreversible act isn't spending ~$1.2M to keep an asset you own running — it's throwing that asset away and betting you'll never need to rebuild it (at ~$40K/home if you do). Maintaining it is the conservative default; letting it go is the gamble.
3. The reversible option is the one stewards are supposed to favor. This isn't only our view. When the World Bank and RAND advise on long-lived infrastructure under uncertainty, the rule is the same: favor the reversible, flexible option. The right question for a board isn't “will this pay off?” — it's “which choice are we least likely to regret in twenty years?” Keeping a working network you already own is the low-regret answer. INDEPENDENT (real-options theory; World Bank / RAND decision-making under uncertainty)
The cautious move and the bold move have switched places: spending to maintain what you own is caution — letting it go is the irreversible bet.
Below, each of the District's three core questions, quantified — with the honest limits flagged.
“What if we upgrade and it keeps declining anyway?”
This is the fair, central worry. So we built a cost-recovery model on the District's own Tarana cost workbooks and billing data, ran it 200,000 times across the real uncertainty (operating margin, growth, and whether a major reseller like NCI stays or leaves), and asked one thing: does the $1.2 million pay for itself?
“Does that include any O&M?” — the question that came back when we described these simulations to the Board (hear it on the meeting audio →).
Yes — explicitly. Every payback figure here is net cash after operations & maintenance, never gross revenue: the model charges wireless an audited ~45% cash margin (about 55¢ of every dollar assumed consumed by crew, sites, backhaul, and admin) and then stress-tests that margin from 28% to 58% in every run — because the District has never published wireless's standalone O&M split, and our records request asks for it. Don't take our word: read the full answer and drag the O&M lever in the live calculator yourself →
Undiscounted simple-payback by scenario — all inside the gear's ~10–11-year life
OUR MODEL Scoped to the 878 customers the $1.2M actually serves (it's a per-customer build — a reseller's exit removes its radios and its revenue together, so the penalty is small). These bars are deterministic scenarios at the proposed $40 rate and charge only the radios actually built (~$1.06M); on the full ~$1.18M workbook basis the planned tier runs ~6.9 yr discounted (see the financial analysis). They are an undiscounted simple-payback illustration; the headline numbers come from the fully discounted, stress-tested three-scenario model — whose median discounted payback is ~9–10 years (inside the gear's ~10–11-year life), where the District earns its capital back ~88% of the time in the Balanced case (~19–97% across the three scenarios) and the county comes out ahead ~99% (~88–100% across them). Two modeling choices, stated: the 28% margin floor is a cash margin (the District's only published figure, 24.9%, is a net margin struck after non-cash depreciation), and organic subscriber growth is centered at flat — the documented decline is mostly one reseller, modeled separately as a discrete exit, not as an ongoing trend. Source: recovery_model.py, District Tarana workbooks + SAO-audited margins.
Two facts make the “it'll keep declining” premise weaker than it looks. First, the recent decline isn't broad demand erosion — it's one reseller. More than half of the past year's drop is a single out-of-area-acquired provider (NCI/Core Fiber) restructuring; every other provider held far steadier, and wireless revenue over the last 12 months on record is actually higher than the first 12. Second, the network is full — the District's own memo calls it “severely congested with no possibility for expansion.” You can't lose customers you're not allowed to sign up. The upgrade is what lets it take customers again.
Why customers actually leave (what the upgrade should target)
INDEPENDENT Reasons customers cite for switching ISP (Opensignal). The point for a skeptic: the upgrade's payoff is reliability and capacity headroom (stop turning neighbors away), not a speed race — which is the more defensible, durable retention argument.
“Will rural rates really go up without the network?”
You don't have to take it on faith that a monopoly charges more. The dominant fallback — Starlink — tells you it prices by captivity in its own words: it adds surcharges where the network is full and people have few options, and gates discounts to areas with “excess network capacity.” Okanogan is in the surcharge zone.
Starlink prices by how trapped you are — its own pricing, documented
INDEPENDENT / PRIMARY Capacity-tiered $90 vs $120 pricing — CNBC (Feb 2023). City surcharges $750 (Pacific NW) to $1,500 (Alaska) — Broadband Breakfast / SatelliteInternet (2025–26). Discounts “gated to excess network availability” — Starlink's own promo language. These are list figures in a market still in flux; verify live before quoting a single number.
The public network is the thing that keeps that pricing power in check for the households it serves. The mechanism is simple and evidenced: where there are 3+ providers, people pay about $5/month less than where there are only 1–2 INDEPENDENT (Consumer Reports, 22,000 bills). A local fixed-wireless plan runs about $50–$75/month with little or no equipment cost; residential Starlink runs ~$120/month plus a ~$499 dish, and business/farm satellite runs 5–10× a residential WISP plan. Remove the local option and the rural majority is left negotiating with a single provider that, by its own pricing, charges more where you have nowhere else to go.
The price gap doesn't just exist. It widens every year.
Here's the part that's easy to miss, and it's one of the strongest reasons to keep the public network: its pricing edge is not static — it compounds. As a non-profit public utility the District prices at cost, not for profit (and RCW 54.16.330 requires its rates be nondiscriminatory and the telecom line self-supporting). A for-profit ISP charges what earns its shareholders a return. Those two pricing rules pull apart a little more every year.
Projected monthly rate — a cost-based public rate that holds, vs. a for-profit bill that raises a little every year. The shaded wedge is the customer's growing value.
ILLUSTRATIVE Both lines start at the District's proposed $40 to isolate the effect of rate increases alone — in reality the monopoly fallback already starts far higher (Starlink ~$120). The public line holds flat, which is conservative: by law a PUD can only recover cost, and the durable share of its costs falls as debt retires. The for-profit line rises ~5% a year — in line with real broadband hikes (AT&T raised home-internet prices in 2023, 2024, and 2025), and below the 6.1%/yr the FCC documented for cable as a long-run analogy (DA 14-672); at that pace a bill doubles in about 14 years. Over 15 years the wholesale gap compounds to roughly $3,000–$4,000 per connection at the floor. We chart the rate the PUD controls (wholesale); each ISP adds its own retail markup — but because the cost-based floor underneath doesn't ratchet, the public option's bill still climbs far slower.
And the commercial ratchet isn't a theory — it's the documented history of this industry. The FCC's own annual price survey found expanded-basic cable climbed from $22.35/mo (1995) to $64.41/mo (2013) — a 6.1% compound annual increase, about 2.5× the ~2.4% inflation of those years. Same playbook on the broadband side today: AT&T raised home-internet prices in 2023, 2024, and 2025, and cable “fees” (broadcast, “network enhancement,” and the like) have multiplied into $2.50–$10/month add-ons. Pricing to a target return means the bill only goes one way.
A for-profit bill vs. inflation — expanded-basic cable, 1995–2013
PRIMARY FCC Media Bureau, Report on Cable Industry Prices (DA 14-672, 2014): $22.35 → $64.41, 6.1%/yr vs. ~2.4% CPI. The shaded wedge is the gap between what a profit-priced service charged and what inflation alone would explain — a gap that widens with every passing year.
Now stop using national stand-ins — look at this exact network's own rate book. Forget what a reseller charges; go straight to the source: the rate the District itself charges — at cost, as a non-profit public utility — written into its own published rate resolutions. Line them up — 2009, 2015, 2017, 2022 — and the number simply doesn't move; it hasn't been raised since.
The PUD's own wholesale wireless rate — per connection, from its rate resolutions
| Monthly rate | 2009 Res 1484 | 2015 Res 1606 | 2017 Res 1643 | 2022 Res 1768 |
|---|---|---|---|---|
| Legacy wireless — ½ / 1 / 2 Mbps | $18/$26/$34 | $18/$26/$34 | $18/$26/$34 | carried as legacy |
| Broadband wireless — 3/1 → 20/10 Mbps | — | $10/$17/$24/$31 | $10/$17/$24/$31 | $10/$17/$24/$31 |
PRIMARY Okanogan PUD wholesale rate resolutions, each linked: 1484 (eff. 7/1/2009), 1606 (eff. 7/1/2015), 1643 (eff. 10/1/2017), and 1768 (eff. 3/1/2022, hosted here; still in effect). The legacy tiers are unchanged 2009–2017; the broadband tiers are unchanged 2015–2022; the $40 setup charge is the same in 2022 as in 2009. (Res 1606/1643 are also held by records request; Res 1768 is a scanned PDF, its figures cross-checked against the prior schedules.)
First, it answers the question of whether the 2010 rebuild reset the rate. It didn't. The District rebuilt this network on a multi-million-dollar federal loan around 2010 — and did not raise the wholesale rate when it did: the 2009 figures are unchanged in 2015 and 2017, and the broadband tiers added in 2015 hold through 2022 (and remain in effect). The District renews the plant underneath without ratcheting the price — the cost-based rule working as designed.
Second — and this is the number that actually matters — here is what the District proposes to charge for the upgraded network, from its own planning materials. Its Tarana presentation states that the wireless network historically carried “3–4 speed tier options, with wholesale prices ranging from $17 to $39 a month,” and that for the new system it will offer one plan: $40 a month for 250/50 Mbps. PRIMARY (Okanogan PUD Tarana presentation, obtained by records request.) That figure is the one that counts because it's the rate the District itself controls — the wholesale price every independent ISP pays to ride the network. The base barely moves — about the old top-tier price — for several times the speed, sold as a single, cost-based plan. Next to inflation it's not flat — it's a cut. The District's 2009 top wireless tier was $34/month (for 2 Mbps; the slide above rounds the old range to “$17–$39,” and we anchor on the lower, documented $34); in today's dollars that is about $53 (CPI‑U, ×1.56), so a $40 plan runs roughly a quarter cheaper in real terms than what the District charged back then — while carrying 250 Mbps in place of 2. What each ISP then charges its own customers we can't tell you, and won't pretend to — that markup is their call. But the thing that protects a rural household is the existence of that low, cost-based wholesale floor; a captive-market monopoly offers no equivalent.
This isn't unique to Okanogan — it's what cost-based pricing does everywhere it's tried. The one independent academic study of the question — Harvard's Berkman Klein Center — found community-owned networks were cheaper than the local commercial provider in 23 of 27 markets, and (the part that matters here) their prices were “clear and unchanging,” versus commercial “teaser” rates that jumped sharply after about a year. Chattanooga's public utility took gigabit service from ~$300/mo to ~$70 while raising base speeds for free. That's the cost-based model doing exactly what its structure predicts: prices that hold, and fall in real terms, instead of climbing.
Put the two risks together and they answer each other. Because the District never raises its rate to protect a margin — its rate is cost-based, and its costs fall as debt retires and equipment depreciates — every year the commercial bill climbs, the public plan becomes a better relative deal. That isn't only an affordability story; it's a retention story, and retention is the one thing the Risk 1 payback model leans on. So the odds the upgrade pays for itself don't decay over time — they improve. A network that holds its price in a market that doesn't tends to hold its customers, too.
It also turns the Starlink worry on its head. A board can hear “but Starlink is coming” as a reason to retreat. It is the opposite. The entire purpose of a public network is to be the check that keeps a monopoly from pricing rural customers by captivity — which Starlink, by its own surcharges, is already doing (Risk 2). Walk away from the network and you don't escape that risk; you hand it the field, removing the one thing disciplining rural prices at the exact moment the monopoly is demonstrating it will use its power. The competition isn't the argument against the upgrade. It's the argument for it.
A rate that never chases a margin gets safer to bet on every year — and the monopoly's arrival is the reason to keep the check, not to drop it.
A small, captured market is where prices have only one way to go.
An investor-owned company exists to grow returns for its shareholders. In a big, growing city it can do that by adding customers. In Okanogan it can't — and that's exactly what makes the price risk higher here, not lower.
The demographics trap the pricing. The county grows slowly — its small towns (Tonasket, Oroville, Conconully) have been flat or declining — about 1 in 4 residents is over 65, and incomes run below the state's. A provider can't meaningfully grow its subscriber count in a market like that. So once it has captured the customers who are here, the only lever left to grow revenue is the one pointed straight at those captured customers: price.
And no second provider rides in to discipline it. The federal GAO is blunt about why: in low-density, rugged terrain the market “offers little profit potential,” so “the market does not support private broadband investment.” The very smallness that keeps a competitor from entering is what removes the check on the one already here. Nationally, only about 1 in 5 rural Americans has more than one wired option. PRIMARY (GAO)
Here's the pattern to watch, because it's documented: new equipment comes in priced to win customers; once the build is sunk and the market is captured, the price climbs. Independent research on 140 U.S. plans found the introductory rate averaged $62/month and the standard rate it reset to averaged $83 — a $22/month jump, just for staying. The honeymoon ends. INDEPENDENT (New America)
Starlink's residential price since launch — the newest gear, one direction
PRIMARY / INDEPENDENT Top tier shown. On top of the monthly hikes: the kit rose $499→$599; one-time “demand surcharges” up to $1,500 appeared in captured cells; a new $10/mo hardware fee was added. The June 2026 increase was the first applied to existing, long-captured customers — which Starlink justified as supporting “ongoing investment.” Sources: CNBC, Broadband Breakfast, SatelliteInternet (2020–2026).
Now look at who the District would be handing the county to. Starlink launched in 2020 with the newest gear on the market — and has raised its price every couple of years since, layering on a bigger dish, congestion surcharges, and a monthly hardware fee, and in 2026 raising the rate on captured customers for the first time. That's not the exception — that's the cycle, in real time. And it matters for the local picture too: if NCI — an out-of-area, investor-owned operator — builds its own gear, it starts the same clock. Today's attractive prices are the honeymoon. The hikes come after the public network is gone and there's nothing left to hold them in check.
And it isn't a small leak — it's the county's money, leaving for good. A dollar spent on the local network largely stays here: at-cost rates, local providers, local jobs, a public asset. A dollar spent on out-of-state satellite leaves the county entirely — and never circulates back.
And the true loss is bigger than that headline, because of what local money does. Dollars spent with the seven local providers don't just stay in the county — they recirculate: the providers' payroll, their install and repair crews, their offices, and the local taxes they pay all get spent again inside Okanogan. Neutral government models (the BEA's regional multipliers, USDA's rural-economy data) confirm that local spending generates additional local activity beyond the first dollar. An out-of-state satellite bill does none of that — it exports the entire amount the moment it's paid.
We'll be straight about the limit: a local ISP buys its equipment and bandwidth from outside the county too — even friendly industry research finds only about a third of a rural provider's economic impact stays rural. But a third of something is far more than a third of nothing, and what stays — the local jobs, the payroll, the accountability — is exactly what disappears if these businesses fold. Lose the network and you don't only lose a service; you lose local employers and the spending that rides on them.
“What's the risk of bad outcomes — people losing service, providers folding?”
This is where the asymmetry bites hardest, because the downside is the kind you can't take back. Two things make it concrete and quantified: how fragile the local providers are, and how much more it costs to rebuild than to maintain.
The seven local providers own no towers. The District runs an open-access wholesale network — the providers buy capacity and resell it to homes — so every customer-facing ISP here is a reseller whose entire product is that platform. When a wholesale platform degrades or is withdrawn, asset-light resellers don't limp along — they exit fast. The clearest precedent: when federal regulators withdrew the wholesale-leasing rules in 2004, AT&T and MCI stopped serving residential customers within weeks INDEPENDENT. For most of our providers, wireless is the majority of what they buy from the District (Highland 77%, Will Connect 69%, CommunityNET 68%). Let the platform age out and you don't lose one network — you lose the whole competitive ecosystem on top of it, and it doesn't reassemble.
Maintain now vs. rebuild later — the cost asymmetry
INDEPENDENT Rural fiber runs ~$13K/location nationally and $40K–$200K+ in extreme terrain like Okanogan's (state BEAD data; the District's own ReConnect award works out to ~$40K/home). 60–80% of that is sunk civil-works labor — trenching and pole work — that a rebuild pays in full again. Maintaining touches the cheap layer (the radios); losing it forces repayment of the expensive one.
And there's a closer cautionary tale than any benchmark: iProvo — a public open-access network built for ~$39 million, whose local-provider ecosystem collapsed and was eventually sold for $1, with the city left holding the debt. The competitive retail layer never came back. INDEPENDENT
“Are we even allowed to take this risk?” — and “have we before?”
First: yes — it's the board's to decide. Some treat a public utility as if it were legally barred from any uncertain investment. It isn't. Under Washington law a PUD has broad authority to “improve, repair, operate… and add to” its telecommunications network (RCW 54.16.330), read liberally by the courts. There is no statute requiring a PUD to grow demand before it invests — the opposite is the design: PUDs exist to build where the private market hasn't. The “grow-first” condition is a business-judgment policy choice, not a legal mandate. The risk is the board's to weigh — and to take.
Second: this District has taken far bigger risks — and they paid off. The “we can't take that chance” instinct sits oddly against the District's own history. The network exists because the District took risks orders of magnitude larger than a $1.2M refresh:
The risk profile has inverted. The founding bets were leaps into the unknown — and they created the network now in question. Today's decision isn't a leap; it's maintaining the paying asset those leaps built. A board that could shoulder a $9.2M build into unserved country can surely weigh a $1.2M refresh of the network it already owns.
And this time the District isn't starting from dirt — it's starting from strength. The 2010 build was a leap precisely because almost nothing was in place yet. Every one of those open questions has since been answered:
In 2010 the District invested into uncertainty — and was vindicated. In 2026 it would be investing into certainty: an established reseller market, demand proven to the point of congestion, and an asset it already owns. If the first bet was prudent, this one is far easier to defend.
The fiscally-conservative choice is to invest.
Lay the three risks side by side and the conservative call flips from where instinct first points.
The risk of acting is a bet with a known, bounded downside (you spend $1.2M, recovery takes longer than hoped) with a ~88% chance of paying for itself (Balanced case) — and a ~99% chance the county comes out ahead either way — backed by an asset you keep regardless. The risk of not acting has an unbounded, irreversible downside: the provider ecosystem gone, the rural majority handed to a captive satellite, the sunk public money lost, and a rebuild bill in the tens of millions.
A steward whose duty is to protect ratepayers should fear the irreversible loss more than the recoverable expense. That's not advocacy talking — it's the standard logic of decision-making under uncertainty: when one path is reversible and the other isn't, you preserve the option you can't get back. Funding the upgrade is the cautious move.
We're not asking the Board to ignore risk. We're asking it to weigh both risks honestly — and the records say the bigger, less reversible one is doing nothing. Ask the Board to weigh it in the open →
What's solid, and what's inference.
Strongest / independent: the cost-recovery probabilities are from our own model on the District's verified Tarana workbooks and SAO-audited margins (every variable, distribution, and exclusion is published — and you can re-run it yourself); the one-reseller concentration of the decline is from the District's per-provider billing (PRA). Starlink's capacity-based pricing ($90/$120; $750–$1,500 surcharges; discounts gated to “excess capacity”) is independent/primary reporting (CNBC, Broadband Breakfast, SatelliteInternet, 2023–2026). The competition price gap (~$5/mo) is Consumer Reports (22,000 bills). The reseller-collapse precedent (CLEC/UNE-P, iProvo) and the rebuild-cost asymmetry (60–80% sunk labor; $40K–$200K/home extreme-terrain fiber) are independent.
Directional / vendor-sourced (flagged in text): the next-gen-wireless churn-reduction case studies are operator marketing — we present them as directional, not proof. Satellite dollar figures are in flux (2026) — treat as a dated range. The “~$100M rebuild” is an order-of-magnitude estimate (2,700 × ~$40K), to show scale, not a precise bid.
What we deliberately do NOT claim: that an upgrade is guaranteed to reverse the decline (no independent case proves it does or doesn't); that broadband prices “rose” as settled fact; or that losing the network means service “dies forever.” The honest case stands without any of those.
The full financial case → · The one-reseller decline → · Why the network protects rates → · Primary documents →