Report 2 of 2 · Two Leon Counties — the Disparity Study
What October 1st, 2028 takes from 10,678 households in unincorporated Leon County, what replacing it costs, and the three motions that answer it — an independent study from public records, with a recommendation the County can adopt alone.
Every row is a vote taken in public, a filing or a published date. No motive is alleged and none is needed: this is what the order of the steps produces.
The executive brief’s eleven numbered findings, in the front door’s current wording, each pointed at the section and figure of this paper that carries it. The numbering is the brief’s, kept so its Fact chips open this paper.
Six lenses, twenty-two figures. Each tile steps through its own figures; click a tile to hold it, click again to let it run, and click a figure to open the section that proves it. Every tile is rebuilt from this paper’s data bundle, so a tile cannot differ from the figure it points at, and each carries its label. Tiles the executive brief once showed that have no line in the canonical figure basis are not reproduced.
On October 27th, 2026 — the date Consent Item 4 of the 15 September agenda proposes — the Board of County Commissioners holds a fire services workshop, twenty-three months before the 2009 interlocal agreement with the City of Tallahassee ends. I live in Leon County, and in July 2026 I could not find — anywhere — the answer to three questions that seemed to me to come before every other one: what it costs the County to staff and run its own fire department, where that money comes from, and what happens to the fire fee, the ISO rating and homeowners insurance when it does. The County had commissioned a study to answer them — RFP BC-2026-003, awarded to Fitch & Associates for $109,995 on December 9th, 2025 — and its own records production on that contract contains no deliverable. So I built the answer from documents anyone can request: the assessment roll, the street network, the dispatch records the City produced, the budgets and the minutes. This paper asks the question the workshop should open with: what, exactly, does October 1st, 2028 take away, from whom, and what does each way of replacing it cost? Only once that is measured is the choice between paths a choice rather than a mood.
This is version 2 of my report, and it is the version with a person in it. Version 1, published on September 8th, 2026 and kept beside this one unchanged, put the same eleven findings in front of a neutral reader; this version puts them in front of the reader who actually has to live with the outcome — a county resident — and says what I would do. No number changed between the versions: every figure below is in one of the nine self-contained analyses this study published between August 27th and September 6th, 2026, or in the canonical figure basis with its provenance, and the build asserts that every number in this paper also appears in its companion, The Independent Separation Report, which takes the twelve items of RFP BC-2026-003 §3.1 one by one, in the County’s order and in a neutral voice, and is deliberately not rewritten. What changed is the selection and the voice. Five findings that until now lived only in working documents are in the findings where they belong: what the EMS money actually is; the 2023 record of how the EMS levy was raised; a commissioner’s own question at that workshop and the one-sentence answer it got; what terminating the agreement may do to the ambulance consent; and the plain sentence that Districts 1 and 2 pay the same share of the fire bill as District 4 on a fraction of the value. Both reports are built from one data bundle and one figure code, so no number and no figure can differ between them; the executive summary, The Decision Behind the Numbers, is the short read of this paper, and the walk-through of the whole study is The Line.
Every quantity carries one of six labels, on the figure and in the sentence. measured is a count or a timestamp from a record — the cohort, the CAD medians, the roll. modelled is computed from records by a stated method — road distance, the second-unit interval, the instruments, the five-year paths. proxy is a measurement of the wrong population standing in for the right one, labelled every time — the 39% first-arrival share from six zones. range is a figure whose basis cannot be pinned down and is therefore published as an interval with no midpoint — the pay gap. documentary is a thing a record says — a vote, a contract clause, a lead time. ours is a recommendation, not any government’s. Two clocks run through the paper and are never mixed: first-unit arrival is measured from dispatch timestamps; the interval to the second unit is modelled on the road network. Distance is road miles on the county street centreline network and is never response time.
The unincorporated area of Leon County, Florida surrounds the City of Tallahassee, which is the county’s only incorporated municipality. At the 2020 Census the county held 292,198 residents and the city 196,169, so the unincorporated area — the area this study is about — is home to about 96,029 people documentary. That figure is the decennial subtraction and is not updated to later estimates. Under the 2009 interlocal agreement (extended by its Sixth Amendment to September 30th, 2028) the Tallahassee Fire Department’s sixteen career stations answer calls county-wide; six of them — Stations 10 through 15, the “county-facing” stations — are the ones the agreement contemplates transferring, and seven volunteer sites (five County-owned substations FS29–FS33 and the two Miccosukee Volunteer Fire & Rescue stations) sit beyond them. The County pays for the service through a flat fire assessment on each dwelling unit — $223.00 in Zone 2 and $245.39 in Zone 1, about $11.8M a year from Fund 145 — and funds ambulance service separately through an ad valorem municipal service taxing unit at 0.750 mills. Insurers price homes here off the ISO Public Protection Classification, and the whole county is graded Class 3. The grade is awarded to an address, and one of its tests is whether that address is within five road miles of a station that responds to it. That test is the line this paper is drawn on, and it is the insurance industry’s threshold, not ours.
LIVE/ folder beside this file.
Every incident-level figure rests on two productions: the City of Tallahassee’s dispatch production PRR2026-877 (structure-fire CAD by beat, 2021–22, 509 fires with a dispatch-to-first-arrival time; six per-station volunteer-zone CAD reports for calendar 2021, 9,760 zone-rows) and the 2022 ISO Fire Suppression Rating Schedule submission, including the hydrant flow summary of June 29th, 2022. Neither is the complete Computer-Aided Dispatch record; responses handled solely by a volunteer department, and every year outside the produced window, are absent. Cost and revenue figures come from the County’s FY2026 adopted budget (Fund 145 Fire Control; Fund 135 EMS), the Fire Rescue Services Assessment Roll FY2026, the Property Appraiser’s roll as pulled September 3rd, 2026 (46,768 unincorporated dwelling units with a district and a valued parcel; $7,809.4M taxable), the County’s records production on the consultant contract (T26-0654), the negotiation record in adopted minutes and resolutions (the May 27th, 2025 special meeting; City Resolution 25-R-45; the July 14th, 2026 Board direction), and the City’s own agenda item citing apparatus lead times. Geography is the TLCGIS street centreline network, address points, parcels, city limits and hydrant layers.
Every dwelling unit in the unincorporated area was placed individually — 47,395 address points, not a grid — and routed over the county street centreline network (19,726 built segments, 2,542.6 miles, densified to 179,217 nodes and 183,795 edges; addresses snap at a median of 0.027 miles) to the nearest responding station by multi-source Dijkstra. The 2028 scenario changes only the station set: the sixteen City stations become the six county-facing ones plus the seven volunteer sites. A unit crosses if it is within five road miles today and beyond five after. The pipeline was re-run from source on August 17th, 2026 and reproduced every figure exactly; a live re-pull on September 1st reproduced the cohort, its parcels, its value and all five district totals before any new column was read; the station-coverage run of September 7th reproduced the definition again (10,678 of 10,678 inside today; 10,641 of 10,678 outside after, 37 marginally inside and reported). A schematic-circle run on parcels, computed independently, agrees within 3% and is published as a cross-check, not a conflict.
First-unit arrival is measured: dispatch to first arrival on the 509 structure fires, by station area, and on 14,396 medical calls. Both previously published medians (5:06 core, 6:24 rural) reproduce to the second from the production. The interval between the first engine and the second is modelled (drivetime.py: road-network travel at functional-class apparatus speeds, undirected, simultaneous dispatch, no dispatch or turnout time; medians by volunteer zone; the 2028 figures on the 2028 station set). The lawful-entry clock adds the two: a three-person crew cannot enter a working structure fire under 29 CFR 1910.134(g)(4) until a fourth firefighter is on scene, so entry waits for the second unit. A 2021 CAD sample checks the model today — on 16 structure fires with a second company from a different station, it arrived a median of 4.1 minutes after the first — and by construction cannot check 2028, because the units removed in 2028 are the ones arriving first.
The six per-station volunteer-zone CAD reports were re-parsed for the origin of the first suppression unit on scene: inner-city stations (1–9, 16), county-facing stations (10–15), or a volunteer department. Of 9,760 zone-rows, 5,821 carry a suppression-unit arrival within 60 minutes and a dispatch timestamp; those are the denominator. The pooled share is a ratio of sums, not a mean of zone percentages. It is a proxy: six volunteer communities, one calendar year, City production only, and it is labelled so in every sentence that quotes it. Volunteer arrival is measured separately, on a different denominator (138 structure fires in the zones over 2021–22), and the two are never blended.
Because no public document said what “four new stations, $28–40M” would actually buy, a maximal-covering location model was run, greedy, over the same street graph, with candidate sites on a ~0.6-mile lattice of road nodes: 850 anywhere, 656 outside the authoritative city polygon. Greedy means station counts are upper bounds. The constrained run — sites the County could actually build, outside the city — is the control, and the difference between it and the unconstrained run is the finding. Capital is a linear extrapolation of the study’s own $28–40M for four stations ($7–10M each), an extrapolation and not a costing. Road distance, not travel time.
The instruments are priced on the whole unincorporated roll (46,768 units, $7,809.4M taxable value, September 3rd, 2026), holding the total raised constant at $223 per unit, so a change of instrument reallocates the burden and reduces nothing in aggregate; the square-foot tier borrows its slope ($0.13 per square foot above 1,600) from Florida fire-district precedent and solves only the base ($148.81) locally. Five-year path costs (FY2029–FY2033, the first window in which every component appears at least once) are arithmetic on components the study had already published — operating, startup, purchased aid, station capital and station operating — County-side, unfinanced, in the study’s own nominal terms. Every staffing figure scales with a loaded cost of $122,000 per FTE that no record verifies; every parity figure is a range because the $44,178 starting salary in the record has no established vintage.
Eleven findings, each opening with the fact as a sentence and its so-what, then the evidence and the figure that carries it. The numbering is the short read’s and the walk-through’s, so a Fact chip anywhere in the study opens the same finding here (#fact-N). Figure and table numbers are shared with the County’s report — Figure 4 and Table 5 mean the same thing in both — so the crosswalk table (its Table 8) is the one table that does not appear here, and Table 11 is this paper’s own.
96.9% of them are within a mile of the city limit. I measured every address on the county’s own street map: not homes far from help — the neighbours on the edge of town, orphaned by a boundary, not by distance.
What this meansFive road miles is not our threshold. It is where the insurance industry stops counting a fire department’s capability.
Measured on the county’s own street network, 10,678 unincorporated dwelling units — address points placed individually, not parcels expanded by a use code — sit within five road miles of a responding fire station today and beyond it in the station set the County is left with on October 1st, 2028 if no successor agreement or automatic-aid arrangement replaces the interlocal — the condition this whole study is about, and the reason a costed successor path sits in Table 4 beside the rest. They are not homes that were always far from help: they are the ones the City’s ring of stations covers now and the County’s six county-facing stations and seven volunteer sites will not. With them go 9,648 parcels, $2.73 billion in market value and $27.3 million a year in property tax; the median affected home is worth $245,167.
They are pressed against the city line: 96.9% are within one mile of the city limit and 74.4% within half a mile. Three City stations hold most of them — Station 16 on Easterwood Drive (2,224), Station 3 on South Monroe (2,074) and Station 4 on West Pensacola (1,612). After separation the median crossing home is 6.54 road miles from the nearest station that still responds, p90 7.73. 149 of the 10,678 are run today by Stations 12 or 15, which stay; they cross because the station within five road miles of them today is a different one. Each of them is a dot on the board and a record in the living map.
| Class | Units | % of units | Market value | % of value | Median parcel |
|---|---|---|---|---|---|
| Single family | 8,045 | 75.3% | $2,500.6M | 91.7% | $273,855 |
| Mobile home | 1,444 | 13.5% | $58.2M | 2.1% | $38,016 |
| Multifamily | 840 | 7.9% | $86.6M | 3.2% | $157,583 |
| Condo / co-op | 280 | 2.6% | $67.8M | 2.5% | $335,000 |
| Other residential | 69 | 0.6% | $14.5M | 0.5% | $20,687 |
| All | 10,678 | 100% | $2.73B | 100% | $245,167 |
Source: FIGURE_BASIS_CANONICAL §1–§2c; TLCGIS address points, parcels and street centrelines; Property Appraiser class codes. Road miles, never response time. In the County’s report: §4.1.
District 1 holds 4.0% of the value on the roll and pays 9.6% of the bill; District 4 holds 46.9% and pays 29.2%. Put Districts 1 and 2 together and they pay 29.3% — the same bill as District 4 on a fraction of the wealth. I did not model that; I read it off the roll.
What this meansTwo houses, the same fee, very different shares of what the house is worth.
The County collects the fire assessment as a flat charge per dwelling unit — $223 in Zone 2, $245.39 in Zone 1. Because a flat charge ignores value, the share of a household’s wealth it takes falls as the house gets dearer: $2.03 per $1,000 of market value in District 1 against 52¢ in District 4. Both rates are the district’s own — its share of the flat bill against its share of the market value on the unincorporated roll, not one household’s arithmetic. Districts 1 and 2 together hold 11.9% of the taxable value on the unincorporated roll and pay 29.3% of a flat bill.
That is the definition of a regressive charge, and it is a measured fact about the roll, not a modelled one. What is modelled is the alternative: a square-foot tier ($149 up to 1,600 sq ft, then $0.13 a foot) that raises the same money moves Districts 1 and 2 to 24.2% of the bill and a District 1 household from $223 to $182, inside assessment law — no MSTU, no TRIM notice. Levied on value outright the share would be 11.9%. How the County raises the money is a second decision from how much (Finding 8).
Put plainly, because I could not find the sentence written anywhere: District 4 holds 46.9% of the market value on the unincorporated roll and District 1 holds 4.0%, yet Districts 1 and 2 together pay 29.3% of the fire bill and District 4 pays 29.2% — the same bill on a fraction of the wealth. Two bases are in play and they are never blended: the per-district shares are of market value and are not summed; on taxable value Districts 1 and 2 hold 11.9% (FIGURE_BASIS §3b). The County’s own legislative findings say why the charge is flat — Resolution 23-R-19 records that the size or value of the property does not determine the fire response — and that is a defensible reading of assessment law. It is also the whole mechanism: regressive by default, not by design, and changeable at the fire services workshop at no cost to the budget.
| District | Units | Single family | Mobile homes | MH share | Median home |
|---|---|---|---|---|---|
| District 1 | 1,793 | 854 | 832 | 46.4% | $89,035 |
| District 2 | 1,946 | 955 | 553 | 28.4% | $100,444 |
| District 3 | 1,853 | 1,807 | 6 | 0.3% | $351,718 |
| District 4 | 2,133 | 2,091 | 25 | 1.2% | $441,534 |
| District 5 | 2,953 | 2,338 | 28 | 0.9% | $263,805 |
Source: Leon County FY2026 assessment roll; Property Appraiser roll of September 3rd, 2026 (46,768 units); paper_disparity_2026-09-06. In the County’s report: §4.10.
1,444 mobile homes are 13.5% of the affected units and 2.1% of the affected value; 1,373 of them sit on two corridors. A flat charge takes the same dollars from a $38,016 mobile home as from a $273,855 house.
What this meansSame total raised. A mobile home in Woodville pays less; a large house in Killearn pays more.
One in eight of the crossing homes is a mobile home: 1,444 of the 10,678, 13.5% by count and 2.1% of the $2.73B by value. Their median market value is $38,016 against $273,855 for a single-family home. A flat per-dwelling charge takes the same dollars from both.
They are not spread across the county. 930 sit behind Station 3 on South Monroe and 443 behind Station 4 on West Pensacola — 1,373 of the 1,444, in the Woodville, Lake Munson and Fort Braden zones. A two-corridor problem, not a countywide one. And District 1’s structure-fire incidence is 1.97× District 4’s (95% CI 1.15–3.37) — the one incidence claim this study will make.
| District | Units | Fires | Per 1,000 (2 yr) | 95% CI |
|---|---|---|---|---|
| District 1 | 4,481 | 22 | 4.91 | 3.08 – 7.43 |
| District 2 | 9,214 | 24 | 2.60 | 1.67 – 3.88 |
| District 3 | 9,886 | 39 | 3.94 | 2.80 – 5.39 |
| District 4 | 13,638 | 34 | 2.49 | 1.73 – 3.48 |
| District 5 | 9,549 | 22 | 2.30 | 1.44 – 3.49 |
| County | 46,768 | 141 | 3.01 |
Source: FIGURE_BASIS_CANONICAL §2b; paper_disparity_2026-09-06 (incidence, calendar 2021–22, n=141). In the County’s report: §4.1.
A household’s year goes from $234 to $597 for the same service — and Fund 145 holds $12.6M for the first year it stands alone. I would plan against $30M, not $23.5M: the cheaper department is the one that costs households more once the insurance class drops. That is the base case — no successor agreement and no written automatic-aid arrangement credited by Verisk. On the 2022 grading the class falls by distance alone; whether an arrangement with the City holds it is the open question in Finding 7, and it is in the enquiry I have drafted and not sent.
What this meansNothing about your water changes. Your insurance grade changes anyway. Find your row — frame or masonry, private carrier or Citizens.
Standing alone at today’s service level costs about $30M a year against $11.8M under the contract; the lean version, three-person crews, $23.5M. On the unincorporated billed basis that is $234 → $597 a household a year, or $467 lean. Over FY2029–33 the paths cost $76.5M (successor agreement), $137.5M (lean standalone), $170M (same service) and $217–249M — the only path that restores the five-mile reach.
Against that, Fund 145 carries $12,598,661 for FY2029; five years at that level is about $63M. The cheaper way of standing alone is the expensive one: lean saves a household $130 on the assessment, and at Class 10 a frame home on a private ISO-table carrier pays $1,450–1,830 more in premium, a masonry home $440–550, a Citizens policyholder nothing. ISO-table differences, not quotes; no figure here predicts any household’s premium.
Netting the two moves: lean saves a household $130 on the assessment, and at Class 10 a frame home on a private carrier pays $1,450–1,830 more in premium — net, that household is $1,320–1,700 worse off than under same service, for a three-person crew instead of four. That net is for a frame home on a private carrier; the same household insured through Citizens sees the assessment saving and no premium change at all, and is about $130 a year better off. That is what “the cheaper way of standing alone is the expensive one” means as arithmetic, and who it means it for. It is an ISO-table difference and not a prediction about any household’s premium.
| Path | Five-year total | Against the 30% offer |
|---|---|---|
| Successor agreement — the 30% offer | $76.5M | — |
| Successor agreement — realistic band | $80–100M | +$3.5–23.5M |
| Standalone — lean default | $137.5M | +$61M |
| Standalone — same service | $170M | +$93.5M |
| Standalone + aid + four stations (restores the reach) | $217–249M | +$140.5–172.5M |
| Five years of Fund 145 held at its FY2029 level ($12,598,661) — ours, labelled | ≈ $63M | the money that exists |
Source: paper_allin_2026-09-06 (FY2029–33, the study’s own components; the $122,000 loaded cost is the declared weak link); Leon County FY2027 adopted budget, Fund 145. In the County’s report: §4.8.
Three ride the first engine and wait for the fourth; after 2028 that wait is about 18.3 minutes in the worst-hit zones — the difference between a kitchen fire and a total loss. Put the fourth aboard and there is no second wait at all: entry becomes the engine’s own arrival, a measured 6:24, better than the 10–14 minutes it takes today — for $2.80M a year, and without waiting for 2028.
What this meansThe crew can be parked outside your house, legally unable to go in, waiting for a fourth firefighter.
Two clocks, never mixed. The first is measured: the first unit reaches a structure fire in a median of 5:06 in the core city and 6:24 in the rural station areas (p90 7:56 against 11:02, n=509). The second is modelled: lawful interior attack needs four firefighters on scene (29 CFR 1910.134(g)(4), two in, two out), three ride the first engine, and the fourth arrives on the second.
That interval runs 10–14 minutes today and about 24 after separation in the worst-hit zones — Lake Jackson 3.7 → 18.0 minutes, Woodville 7.7 → 18.3. A fourth firefighter aboard every county-facing engine puts lawful entry at about 7 minutes — the one intervention on any list that changes what happens at a working fire before the deadline, and it costs $2.80M a year.
| Area | n | Median | 90th percentile |
|---|---|---|---|
| Structure fires — Stations 1–8 (core city) | 327 | 5:06 | 7:56 |
| Structure fires — Stations 11–14 (rural county) | 67 | 6:24 | 11:02 |
| Structure fires — Station 13 alone (Lake Munson’s nearest responder after 2028) | 27 | 6:24 | 9:12 |
| Medical calls — city station areas | 14,396 | 6:14 | 9:39 |
| Medical calls — the six outlying station areas | 8:12 | 13:43 (Miccosukee area p90 18:49) |
Source: PRR2026-877 (the City’s dispatch production, structure fires); the second-unit interval is modelled on the road network — FIGURE_BASIS §2p. In the County’s report: §4.3.
98.6% of the homes at stake sit behind stations that stay City property, whoever wins the six buildings in dispute. After October 1st, 2028 the only way to keep those engines coming is to buy them, from the one department that has them, at whatever price it names — nobody has published what the substitute would cost, so I priced it, and labelled it modelled.
What this meansWin the buildings and you have won six addresses. The engine that reaches you is the thing being argued away, and it is not in the buildings.
The buildings everyone is fighting over — the six county-facing stations — cover 1.4% of the crossing homes. The other 98.6% (10,529) sit behind City stations inside the city that will stay City property on any outcome. Buying the six outright is about $2.3M; building four new County stations is $28–40M ($7–10M each), and none opens before 2030 — two years after the service ends. Recovering the whole cohort takes 9 stations at $63–90M, on the same calendar. And on the City's own transition report the six are surplus to it: its footprint shifts to the incorporated limits and staffing at City stations goes up, so the buildings the County would be buying are buildings the City no longer needs — three of the six carry a Property Appraiser improvement year of 1988, 40 years old by 2028, and the apparatus in them transfers at unamortised value on the agreement's own schedules. Age is the appraiser's roll field, not a condition assessment.
Four County-feasible stations recover 88.1% of the cohort; all of it takes 9 ($63–90M), and an unconstrained optimiser puts its first four sites inside the city limits — the shortest road to these homes runs through stations the City already has. Purchased aid from the City is the only remedy in the years before anything opens.
What is being fought over has not been priced, and it is not new. On the Property Appraiser’s own improvement year, 3 of the 5 station buildings carry 1988 — 40 years old by 2028 — and the other 2 carry 1995, or 33. All 5 stand on land the City bought in 1988. That is the appraiser’s roll field and not a condition assessment: age is not condition, and this study makes no claim about the condition of any building. Against those ages sits the price. Exhibit E ¶D.2(a) obliges the City, at the County’s sole option, to sell Stations 10–14 and their land for no more than $525,000 — five stations. Station 15 is not capped: the price there is “the City’s investment in same,” and no such figure exists in any public record, so this study built one: $1,789,171.60. Apparatus is priced separately at unamortised value on the agreement’s own 15-year and 10-year schedules. And on 4 September 2026 the City certified in writing that no appraisal, valuation or asset list exists for those provisions. Only 8 of the 10,678 households sit behind Station 15 — the one building whose price is open-ended.
Source: DATA_stations_2026-09-07 (CC BY 4.0) — the station-coverage dataset nobody had published; paper_siting_2026-09-05; station age from the Property Appraiser improvement year (TLCGIS parcel layer, read 3 September 2026) and the recorded deeds, FIGURE_BASIS §3m; Exhibit E from the 2009 interlocal transcript; PRR2026-869 production. In the County’s report: §4.7.
They keep the hydrants and lose the engine — and the paramedic on it, because under the interlocal the City’s engines are ALS first responders county-wide, so the medical call loses its first responder on the same day the fire call loses its engine. On the 2022 grading the ISO class falls from Class 3 to 10W or 10 by distance alone, and the City’s $8.5M Class 2 moves none of them back.
What this meansMost 911 calls in the unincorporated county are medical, and the first paramedic is usually on a City engine. What leaves is not the water — it is the crew, and the paramedic on it.
Two conditions travel with the class. The Schedule credits a written automatic-aid arrangement dispatched on the initial alarm, so an arrangement with the City could hold the grade; and the 2022 submission is the last one in the record, so a regrade since would move the baseline. Both are questions for Verisk, and both are in the enquiry I have drafted and not yet sent.
93.8% of the crossing homes already have a hydrant within 1,000 feet, and for 94.8% the nearest hydrant is City-owned. Water is not what they lose. The 2022 ISO survey scored the City 73.54 of 105.5; Class 2 begins at 80. Distance to a staffed engine is the failure mode: 6,779 of the 10,678 (63.5%) land in Class 10W because the water stays, and 3,899 in Class 10.
The boundary that produces the split is worth stating, because it is why the hydrants matter at all. Between five and seven road miles with a creditable hydrant within 1,000 feet, a property lands in Class 10W — a real class that exists because those homes demonstrably burn better than homes with no water. Beyond seven road miles, or with no creditable water, it is Class 10. That is the whole of the 63.5% / 36.5% split.
Of the county’s 9,440 active hydrants, 7,660 are the City’s, 948 Talquin’s and 612 private; of 90 tested, 39 did not meet needed fire flow — and today nobody is testing. A flow-testing programme ($0.15–0.25M a year) is the cheapest grading point on the board. Whether ISO credits the six volunteer departments is unanswered — one letter to Verisk settles a $12.0–16.9M band.
Source: The 2022 ISO PPC submission (paper_iso_build_2026-08-27); TLCGIS hydrant layer of September 2nd, 2026; ISO-table premium differences, not quotes. In the County’s report: §4.4.
The same money, collected four ways, lands on different people: Districts 1 and 2 carry 29.3% of a flat bill and 11.9% of a bill on taxable value. The square-foot tier is the free fix, and I would move it with Recommendation 1 — in the same motion, not on a later agenda.
What this meansYou are not being asked whether to pay. You are being asked, once, on what basis — and the basis is worth more to your household than the amount.
The fourth row is the measure of the whole thing. Levied on value, Districts 1 and 2 would pay 11.9% of the bill — exactly their share of the taxable roll — and a District 1 household would pay $84 where a District 4 household paid $369. Today both pay $223. That is not a recommendation; a pure ad valorem levy is a different instrument with its own law and its own politics. It is the yardstick: it shows how far the flat charge sits from the value it is levied on, and it is the reason the square-foot tier counts as a fix rather than a rearrangement.
Holding the total constant on the whole unincorporated roll (46,768 units), four instruments: the flat charge as today (29.3% on Districts 1 and 2, a District 1 household $223); a square-foot tier (24.2%, $182); half flat and half on value (20.6%, $154, no district rising by more than about $73); and pure ad valorem (11.9%, $84 — with TRIM notice, hearings, a millage cap, and a base Amendment 3 will move unevenly).
The square-foot tier is the free fix: it stays inside assessment law, collects the same total, and under it four of the five districts pay less. It is a separate decision from any rate, and I would take it first — it is the first of the three motions (FIGURE_BASIS §3e), before a dollar of new money is voted.
| Instrument | Districts 1+2 share of bill | District 1 household | District 4 household | Note |
|---|---|---|---|---|
| Flat per dwelling | 29.3% | $223 | $223 | What the County charges now. |
| Square-foot tiered — $149 up to 1,600 sq ft, then $0.13 a foot | 24.2% | $182 | $282 | Stays inside assessment law — no MSTU, no TRIM notice. |
| Half flat, half on value | 20.6% | $154 | $296 | No district rises by more than about $73. |
| Levied on value — pure ad valorem | 11.9% | $84 | $369 | Districts 1 and 2 hold 11.9% of the taxable value, and would pay 11.9% of the bill. |
| Instrument | D1 + D2 share | D1 household | What it is |
|---|---|---|---|
| Flat per dwelling today | 29.3% | $223 | What the County charges now. |
| Square-foot tiered $149 up to 1,600 sq ft, then $0.13 a foot | 24.2% | $182 | Stays inside assessment law — no MSTU, no TRIM notice. |
| Half flat, half on value | 20.6% | $154 | No district rises by more than about $73. |
| Levied on value pure ad valorem | 11.9% | $84 | Districts 1 and 2 hold 11.9% of the taxable value, and would pay 11.9% of the bill. |
Source: paper_disparity_2026-09-06; paper_whopays_2026-09-05 (read its errata first); the roll of September 3rd, 2026. In the County’s report: §4.10.
The County’s $109,995 study is half paid, the County’s own production on the contract (T26-0654) contains no deliverable, and the Board sees the work for the first time on the day it is asked to act. I would ask for it in writing now. Nobody, on either side, has published the arithmetic of the separation; that absence is why this study exists. The risk that creates is the County’s own.
What this meansEvery figure in front of the Board at the fire services workshop will be somebody’s estimate. The only question is whose, and whether it is written down.
What that risk is, stated plainly. A board that terminates for cause, then adopts a rate and a service plan before its commissioned study is in front of it, is asked to defend three things at once: a decision taken without the document it bought, a fee whose incidence is 2.03 against 52¢ per $1,000 and has been since it was adopted, and a calendar in which nothing it can build opens before 2030. None of that requires anyone to have acted improperly, and this study alleges nothing of the kind. It only observes that each of the three is answerable now and much harder to answer in 2029 — and that the record of what was known, and when, is being written in the minutes of the meetings between here and October 2028.
The County awarded RFP BC-2026-003 to Fitch & Associates for $109,995 on December 9th, 2025; its proposal promised a draft in May 2026 and finals in June 2026. Two milestone payments of $27,498.75 — $54,997.50, half the fee — were made in January and February 2026. On June 5th, 2026 the contract was extended 210 days to December 31st, 2026, two months past the workshop it informs. The County’s own records production (T26-0654) contains no deliverable, no Notice to Proceed and no evaluation scoring. No impropriety is alleged; the consequence is on the record.
§3 of the agreement gives the consultant 180 days from a Notice to Proceed and prices lateness at $100 a day. Across the 210-day extension that rate would come to $21,000 — about a fifth of the fee. Nothing is owed: the extension was granted, and a granted extension is not lateness, so no damages accrued and none could be collected. The figure is not a debt; it is what the contract itself says that much time is worth. No impropriety is alleged — extensions are ordinary and delegated authority is lawful — and the consequence is the one the record shows: the Board sees the work for the first time on the day it is asked to act.
Four things in the agreement itself are worth ten minutes of a commissioner's time. §26 makes an extension of time the contractor's sole and exclusive remedy for delay — so the extension is the thing the contract already contemplated, and the liquidated-damages clause is the thing it already priced. Both milestone payments went on a purchase-order schedule that the executed agreement makes payable only on completion and acceptance (§5) and declines to be modified by (§39). Amendment 1 was executed on June 5th, 2026 by the Purchasing Director under delegated signature authority, reciting “project delays” and extending the term 210 days to December 31st, 2026 — two months past the fire services workshop the study was commissioned to inform (proposed for October 27th, 2026) — and it never returned to the Board. And whether anything has been paid since February cannot be tested from the public record at all, because the Clerk's vendor-payment portal holds nothing after May 2022. None of that is an allegation. A delegated signature is ordinary procurement; an unassessed damages clause is a choice an owner may lawfully make. It is a list of questions staff can answer in writing, and each one is answerable before the workshop.
On the other side of the line the City has a real answer. Unincorporated households pay $11.8M for a service the County would have to spend about $30M to replace — about 39 cents on the dollar of the replacement cost. That is not the same as a subsidy, and the difference matters: the City’s marginal cost of covering the unincorporated area from an existing sixteen-station system is lower than building the service from nothing, and the City has never published what that marginal cost is — which is the other half of why nobody has costed the separation. In January 2025 the City proposed 22% carrying one costed element, $8.5M toward moving the ISO rating from 3 to 2, and County staff recorded that there was no data behind the insurance saving. The County offered 17% excluding all ISO-related costs, 4–3, and was refused. It then held unincorporated rates flat while the City raised its own residents’ rates instead — so by FY2028 city households will have paid about 13.3% and 11.9% more than in FY2025 while the unincorporated rate did not move. On July 14th, 2026 the County moved to terminate with cause. Nobody, on either side, has published the arithmetic of the separation itself.
Source: Leon County records production T26-0654; contract B-26-069; County and City minutes, January 2025 – July 2026. In the County’s report: §5.1.
The EMS millage went 0.500 → 0.750 mills in a 13-minute hearing, the 50% was never said aloud, and the one commissioner who asked whether any other county had done this got one sentence back. $41.1M is the fund, not the service — more than half of it is already paid by billing. And the City’s consent to collect that levy on city property ends on the same day as the fire service — or, on the County’s own termination theory, possibly sooner.
What this meansThe instrument that raised the EMS levy is the instrument a fire levy would use, and it moved in thirteen minutes.
There is a third emergency-services levy. For FY2024 the County raised its EMS millage from 0.500 to 0.750 mills, a 50% increase, in a hearing that ran 13 minutes; 3 years have now been collected at the higher rate and the fund balance it feeds more than doubled, from $7.0M to $15.5M, funding an ambulance fleet. No response-time series is published against which the stated justification can be tested.
Where it went is published, and it is lawful: the balance rose $5,339,787 in the first year at the higher rate and more than doubled in two, from $6,995,592 to $15,524,853, funding an ambulance fleet replacement programme — $5.33M of EMS capital in FY2024, $7.74M in FY2025. Accumulating a balance to buy a fleet without borrowing is good practice. It is also money that has not been argued about in public.
And the fire deadline is not the only thing that ends that day. The City’s consent to the EMS tax on city property is keyed to the same Current Term of the same 2009 agreement: it remains in effect “until the last day” of that term and is then “deemed revoked without further action” (Ordinance 23-O-22 §3(D)) — September 30th, 2028, the day before separation, taking about $13M of a $19.5M levy with it unless the City has consented afresh first. One date, two funds; only the fire half is on an agenda. On the 2023 precedent the County gave the City sixty days’ notice before it legislated, which puts the real decision in the first half of 2028.
The medical half is the larger half. In the grading year the six outlying zones generated 2,984 medical calls against 67 structure fires; county-wide, 14,396 calls. The system is $41.1M a year, of which the levy is $19.5M — $41.1M is the fund, not the service. It has not been back on an agenda since.
What the ambulance money actually is, because the most-quoted number describes it least well: $41.1M is Fund 135 as adopted for FY2026 — a budget line, not the cost of the service; $33.4M is the operating division, FY2024 actual (a different year and a different basis, never labelled FY2026); $20.4M is billing revenue — what insurers, Medicare, Medicaid and patients pay for transports; $19.5M is the tax, of which about $13M is collected on city property under the City’s consent and about $6.5M on the unincorporated roll regardless. More than half of the ambulance service is already paid for by billing, not by the tax — $20.4M of reimbursement against a $19.5M levy. So when the millage went from 0.500 to 0.750 in a 13-minute hearing it was not a service running on empty being rescued; it was a service already recovering half its cost from billing, taking a 50% increase in its tax (FIGURE_BASIS §3b; the instrument The other September 30th).
The 2023 record, meeting by meeting. The increase was noticed and heard at four public meetings — the June 20th, 2023 budget workshop, the July 11th ratification and ordinance, and the September 12th and 26th TRIM hearings — and the 50% magnitude appears in none of them: no percentage, no rolled-back rate, no household impact, no cost driver. The maximum rate was Option #16 of 17 inside a single bundled motion on July 11th, carried 7–0; the ordinance the same evening is the only place both numbers appear together, and it carried 6–0 with one member out of the room. The final hearing on September 26th ran from 6:05 to 6:18 p.m. — thirteen minutes — with every commissioner discussion slot minuted “None” and no public speakers. At the June workshop one commissioner asked the only recorded question: “Commissioner Proctor stated this is a phenomenal millage rate. He asked if another County had done this. County Administrator Long stated this is not an unusual increase.” That is the entire recorded discussion. Nobody hid anything — every vote was noticed, heard and unanimous, and minutes are summary records, so silence in them is not proof nothing was said. Nobody examined it either.
And what terminating the agreement may do to it. Because the City’s consent is keyed to the last day of the then Current Term and not to a date, the County’s July 14th, 2026 vote to terminate the interlocal for cause raises a question nobody appears to be asking: if termination ends the Current Term, the consent — about $13M a year — could revoke on the termination date, potentially in 2027; if the Current Term survives early termination, it revokes on September 30th, 2028 as above. I assert neither reading; it is a question for the County Attorney. What I am saying is that the asset argument the County is having is about buildings that serve 1.4% of the households at stake, that the ambulance consent turns on the same phrase in the same agreement and is worth roughly $13M a year, and that a theory which wins the first may accelerate the loss of the second. If consent goes and is not renewed, the plain shape is that the County keeps responding inside the city while collecting the levy only outside it — the cost of a county-wide service on the 96,029 residents of the unincorporated area, the same people losing their fire station. Whether that is the legal outcome is asserted nowhere; that the question is open, and resolves inside the planning horizon of the fire services workshop, is (FIGURE_BASIS §2y).
Source: paper_ems_2026-09-05 (released with its note); County budgets FY2023–FY2027; CDA call records. In the County’s report: §4.10.
Stations cannot open, engines cannot arrive and firefighters cannot be hired inside the time remaining — so for that date the constraint is the calendar, and for every year after it, the money. I want the workshop held, in public, and I want it to fund what is right in every branch.
What this meansMoney can still change every year after October 1st, 2028. It cannot change that day. That is the whole reason the calendar matters more than the price.
From the fire services workshop, proposed for October 27th, 2026, to separation is 23.1 months. On the City’s own published lead times — more than 28 months for an engine, about 24 for a tanker, up to 32 for a ladder — the apparatus window has already closed: the last engine order date was about June 2026, and an engine ordered at the fire services workshop arrives about February 2029. Nothing has been ordered. The academy gate, late 2027, is the one still in front of the Board: a class seated after it does not put firefighters on a truck by October 1st, 2028.
The statutory windows are unforgiving — resolution of intent by early 2028, roll certified September 15th, 2028 — and nothing begun now opens a station before 2030. Same service on October 1st, 2028 is therefore not available at any price. That is a claim about the date, not about the budget. Both constraints are real and they bind at different times: no amount of money buys the service on October 1st, 2028, and from that day forward it is money that decides how much of it comes back — $23.5–30M a year to stand alone and $43.4–49.8M to restore the reach, against $12.6M in Fund 145 (Finding 4). The workshop does not choose a future; it funds what is right in every branch and puts the one near-term improvement money can buy on the table.
| When | What | Note |
|---|---|---|
| already past | Apparatus window closed | City’s own lead times: last engine order ≈ June 2026; ordered at the workshop, an engine arrives ≈ February 2029 |
| October 27th, 2026 (proposed) | Fire Rescue Services Workshop — proposed by Consent Item 4, 15 Sep 2026 | Change how the charge is collected · fund the service and make the written offer · start the 42-hour-week process · send the payroll records request |
| FY2027 | No-regrets set (~$1.5–2.5M) | Volunteer programme · Exhibit E appraisals · Chief recruitment · ISO letter · land-bank Woodville site |
| late 2027 | Academy seated | The last start that staffs 2028 — the one gate still in front of the Board |
| early 2028 | Resolution of intent (§197.3632) | A funding-instrument defect found mid-2028 is a ~$25M hole |
| September 15th, 2028 | Roll certified | Statutory deadline for the assessment roll |
| October 1st, 2028 | Separation | Agreement ends 30 Sep; the County delivers fire protection with or without the City |
| 2030–31 | First new station opens | Nothing built now opens before 2030 |
Source: The City’s own transition report (apparatus lead times); §197.3632; FIGURE_BASIS §2r (the apparatus correction of September 8th, 2026). In the County’s report: §4.9.
The recommendation follows from the findings rather than preceding them, and it is conditional on nothing the City does. Of every costed intervention, exactly one buys minutes before the deadline: a fourth firefighter aboard every county-facing engine, $2.80M a year, which makes lawful interior attack the arrival time rather than a sum of two waits. The first engine reaches a county structure fire in a measured median of 6:24; today the crew must wait for a second unit to bring the fourth firefighter, so entry takes 10–14 minutes, and about 24 after separation. With four aboard it is about 7 — four to eight minutes better than today, from the day it is funded, and it does not depend on what the Board decides about separating. Around it Recommendation 1 adds the volunteer programme ($0.60M), hydrant flow testing ($0.15–0.25M) and the revenue-neutral instrument change — $3.55–3.65M a year in all, +$24 a household county-wide, every item inside County authority. Recommendation 2 is the written offer to the City: the County’s 29.4% share of department-wide pay parity — a share that is not a chosen denominator but the answer three independent measures give, 29.4% by establishment, 30.7% by dwelling units and 31.6% by taxable value, all of them about thirty per cent, $0.56–1.85M a year, which costs what parity at the six county-facing stations alone would cost and reaches all 277 shift firefighters instead of 81.5 — the engines that actually answer the county, ten of the sixteen of which are not the six. Together $4.11–5.50M a year, against $23.5–30M standing alone. It is a purchase and not a subsidy, and should be written as one: open-book pricing, so the County sees the cost basis of every seat it buys; pro-rata clawback, so a seat left unstaffed is a seat unpaid; and forty-eight months’ notice, so neither side can strand the other. The City’s own share of the same parity is $1.35–4.44M, and the City is the party carrying both the retention problem and the fee shortfall that started the rate fight.
Parity first, and not the 42-hour week, for three reasons: it reaches a firefighter for about 56 cents on the dollar against the shift change ($22,701 a firefighter against $40,580); it is cash the County can specify, verify and claw back, where a shift length is none of those things; and shift length is a City–IAFF bargaining subject the County cannot buy however much it offers. So the 42-hour week ($10.1–11.3M) stays an OPTION on its proper table, the City–IAFF bargaining round of early 2027 — and motion three is the County starting that process rather than pretending it can finish it. The instrument change is not optional, and we do not present it as a refinement; it is the first motion. A flat charge per dwelling is regressive by construction: it takes $2.03 per $1,000 of value in District 1 against $0.52 in District 4, and raising it to fund a better service asks the households with the least to carry the most of the increase. The square-foot tier collects the same total, sits inside existing assessment law, moves Districts 1 and 2 from 29.3% of the bill to 24.2%, and takes a District 1 household from $223 to $182. Any recommendation that funds Recommendation 1 through the existing flat charge is a worse recommendation than this one, and we would not make it. Three motions can be made at the fire services workshop, in this order: change how the charge is collected; fund the service and make the written offer; start the 42-hour-week process. The first two require nobody’s agreement to begin; the third asks the City and its union to open a door only they can open (FIGURE_BASIS §3e).
This study takes no position on whether the County should separate. The data is agnostic; the recommendation is not — it is right in every branch. The 42-hour week itself stays an OPTION on its proper table — the City–IAFF bargaining round that opens in early 2027; motion three is the County starting that process, not pretending it can finish it.
The path from the workshop, step by step, is the second view of Figure 9 (Finding 11).
How the County raises the money is a second decision from how much, and it decides who pays. The flat per-dwelling charge puts 29.3% of the bill on Districts 1 and 2, which hold 11.9% of the taxable value; a square-foot tier inside assessment law ($149 up to 1,600 sq ft, then $0.13 a foot) raises the same money, moves that share to 24.2%, takes a District 1 household from $223 to $182, and leaves four of the five districts paying less — with no MSTU, no TRIM notice, and a base Amendment 3 does not move. It is the first of the three motions in §5.1 — the free one, and the one that decides who pays for everything after it (FIGURE_BASIS §3e).
The shares by district are Figure 10 (Finding 2); the four instruments are Table 11 (Finding 8).
Every option as a share of the most expensive one — a household’s year against $597, the same service standing alone. Tiers are county-wide or on the unincorporated roll as labelled; the standalone rows are the billed basis. Different denominators, and the figure says which.
The City controls the workforce; the firefighters control the outcome, and each of them is choosing. The record’s starting salary is $44,178 against a Sheriff’s deputy at $60,000 — a gap published as a range, $4,826–15,822, because the record’s figure may pre-date up to four wage awards. Between about 71.4% and 89% of the department’s booked overtime is arithmetic before it is anything else, and the arithmetic is worth stating because it decides whether “cut the overtime” is an answer. A 24/48 rotation averages 56.2 hours a week; the federal threshold for firefighters under §207(k) is 53.0 — 212 hours in every 28-day cycle. The difference is about three hours a week, every week, for every one of the 277 shift positions: 43,212 hours a year before anyone calls in sick. The department actually booked 38,322. That is a ceiling and not a decomposition — the concentration below caps the structural share at about 71.4%, leaving at least 28.6% that is discretionary — and neither number is a scandal: the 24/48 rotation is the American norm and §207(k) was written for it.
Who carries the rest is the part that reads as a staffing signal. 28 people carry 40% of the book — about 547 hours each, roughly $12,458 of income, against 92 hours for everyone else; one person worked 1,357 hours in a single year, 24.2 extra weeks. Separately, 36,938 hours were worked out of class, about 16.3 positions’ worth of people acting above their grade — a vacancy signature rather than a pay dispute. Parity reaches a firefighter for 56¢ on the dollar against the 42-hour week. And who actually comes is a City engine more often than not: in the six outlying zones an inner-city station put the first unit on scene 39.1% of the time (2,276 of 5,821, calendar 2021 — a proxy, labelled every time it appears), and a volunteer unit arrived at 27.5% of 138 structure fires.
The calendar has decided more than the Board has. On the City’s own published lead times — more than 28 months for an engine, about 24 for a tanker, up to 32 for a ladder — the last engine order date was about June 2026; an engine ordered at the workshop arrives about February 2029. Nothing has been ordered. A recruit class seated after late 2027 does not staff October 1st, 2028; that gate is the one still in front of the Board. Nothing begun now opens a station before 2030. The workshop therefore does not choose a future; it funds what is right in every branch.
Five years of Fund 145 at the FY2029 level is about $63M against a restoring path of $217–249M; both recommendations land inside the order of money the fund already holds. The County moved to terminate for cause on July 14th, 2026, and the City is arguing expiration — a termination transfers the City’s fire assets at a reasonable cost, an expiration does not; this paper prices both and takes no side. A live suit (Inman-Johnson v. City of Tallahassee) turns on the fee, and no figure here predicts any household’s premium. The four places this work is most vulnerable — a pay gap with no vintage, an unverified $122,000 loaded cost, a 39% first-arrival proxy from six zones, base-versus-gross on the salary record — are named in three drafted records requests rather than modelled around.
The gates are Figure 9 and the ledger of what is decided is Figure 16, both in Finding 11.
This is not a fire-service engineering study, and no part of it substitutes for the deployment analysis a licensed firm would perform. It takes no position on whether the County should separate. It alleges no impropriety by anyone: every vote described was noticed, heard and lawful, and where a decision narrowed the County’s options the narrowing is the point, not anyone’s motive. It makes no prediction about any individual household’s insurance premium. Road distance is measured and is never response time; the first-unit clock is measured and the second-unit clock is modelled; the 39% is a proxy; the pay gap is a range with no midpoint; three household bases are named and never blended.
Leon County Commission Chair Christian Caban, on the record at the Board meeting of July 14th, 2026, as reported contemporaneously by WCTV (Julia Miller). Five claims, each set against what this study measured. Where he is right, we say so first.
| Said on the record | What the record answers |
|---|---|
| “Under their current agreement, if the contract is terminated, the City would have to transfer fire-related assets to the county at a reasonable cost.” | Right, and it matters less than it looks. Exhibit E does oblige the City to sell on termination, and the City is arguing 2028 is an expiration rather than a termination — a real dispute worth real money. But 98.6% of the affected households sit behind stations that stay City property whichever way it goes — the buildings actually in dispute serve 1.4% of them. Winning the asset argument does not reach the people. |
| “There is going to be no lapse in service with county residents at all.” | Not a lapse — a longer wait, and it is measurable. Somebody comes either way. But four firefighters must be on scene before anyone may lawfully enter a burning house, and that clock runs about 10–14 minutes today and about 24 on the station set the County is left with. A fourth firefighter on every county-facing engine brings it to about 7. “No lapse” is achievable; it has a price and it has not been budgeted. |
| “It’s not really going to affect us.” | It affects 10,678 homes and it is the whole subject of this study. Those dwelling units sit within five road miles of a responding station today and outside it the day after; 96.9% of them are within a mile of the city limit. The unincorporated area is home to about 96,029 people. |
| “We’ll have all the data from our consultants.” | The Board will see it that day. The County engaged its consultant for $109,995 in December 2025 and has paid $54,997.50. Its own records production on that contract contains no deliverable. No impropriety is alleged — extensions are ordinary — but the arithmetic of the separation has not been published by anyone, and the workshop is where it first appears. |
| “Fair, equitable and affordable services that treat all residents equally, which is currently not happening.” | He is right that it is not equal, and the clearest inequality is the County’s own charge. The flat fire assessment takes $2.03 per $1,000 of market value in District 1 against $0.52 in District 4. District 4 holds 46.9% of the market value on the unincorporated roll and District 1 holds 4.0%, yet Districts 1 and 2 together pay 29.3% of the bill and District 4 pays 29.2%. That one is inside County authority and can be changed at the fire services workshop at no cost to the budget. |
Every other section of this paper reports. This one argues, and it is labelled so. Nothing below is a new fact; each act is established earlier in this paper, from the minutes and the budget. What is new is that we say what we think the pattern amounts to.
Start with what a county resident actually has today, because it is the fact that decides everything after it. Sixteen City fire stations answer calls in unincorporated Leon County. Six of them — Stations 10 through 15 — are the county-facing ones the agreement contemplates transferring. The other ten are not, and they are not decorative: in the six outlying communities an inner-city station put the first unit on scene 39.1% of the time (a proxy, labelled wherever it appears). 149 of the 10,678 are run today by stations that stay.
After October 1st, 2028 there is only one way for those engines to keep coming: buy them. A successor agreement, or mutual aid, or closest-unit response — whatever it is called, it is a purchase from the City at whatever price the City sets, and the County has no substitute to bargain against. And nobody has published what the substitute would cost — so we had to price it ourselves. No County document, the Fitch study included, states how many stations the unincorporated area would need or what they would cost; we searched the budget, the capital schedule, the 2022 ISO submission and the County’s own records production and found none. Our own siting work prices a four-station programme at $28–40M of capital modelled, on 2026 Florida construction costs, with none open before 2030–31. Four stations recover 88.11% of the cohort and nine are needed for all of it ($63–90M). So for the whole of FY2029 and most of FY2030 — at least — there is no alternative to buying service from Tallahassee at market price, and that is true on every path except the one where an agreement is signed before the deadline.
That is the county resident’s position, stated plainly: they are about to be a buyer with no other seller, and nobody has published what the seller will charge.
The sequence that produced it is the four steps boxed at the top of this paper — the July vote, the workshop, the report that arrives in the same room, and the other party reading all of it. Nothing in this study suggests anyone intends that outcome; it is what the sequence produces.
Now put the amounts beside the sequence. The offer that would settle the pay dispute at all sixteen stations that answer the county — the County’s 29.4% share of department-wide parity, $0.56–1.85M a year — works out at 31 cents to $1.01 a month per household spread county-wide. The gap between what the City asked for in May 2025 and what the County offered was, at the household level the County’s own staff put in front of the Board, about $11 a year on a $223 bill — under a dollar a month. That is the size of the disagreement now being resolved by termination. Set against it: standing alone at the same service is $597 a household against $223 — about a dollar a day, every year, permanently.
Where the County is right, and where it stops looking. It is right that the City has never published its cost of serving the unincorporated area, and right that a rate rise unsupported by a cost basis is not a rate rise anyone should simply accept. What the County’s position does not acknowledge is the other half of that arithmetic: unincorporated households pay $11.8M for a service that would cost about $30M to replace — about 39 cents on the dollar. Both things are true at once, and only one of them is being said out loud.
The risk in front of the unincorporated county was made by decisions, not by a calendar. The date has been known since 2009. Every choice about what to do before it was the County’s to make, and each of the following was lawful and within its authority.
On May 27th, 2025 the motion that carried excluded all ISO-related costs, 4–3 — Commissioners Caban, O’Keefe and Proctor opposed — and the minutes record that staff had no data from the City on the insurance saving, and no guarantee of reaching Class 2. The annual increase was put at about $49 with the insurance component and $38 without (the memo’s own question asks about annual increases; those are 22% and 17% of the $223 fee — corrected September 8th, 2026, see the corrections register), and the cheaper figure is the one that carried. A decision about insurance was taken without the insurance data. Then the County offered 17% against the City’s 22%, was refused, and held unincorporated rates flat while the City raised its own residents’ — so by FY2028 city households will have paid about 13.3% and 11.9% more than in FY2025 while the unincorporated rate did not move. On July 14th, 2026 it moved unanimously to terminate for cause. It commissioned a $109,995 study in December 2025, has paid half, and its own production on that contract contains no deliverable.
Held together, that is a decision to stand up a fire department that has not been costed, for 96,029 residents on a roll of 46,768 households, with nothing set aside to pay for it. Standing the same service up alone runs about $30M a year against $11.8M today. The cheaper figure in circulation — $23.5M for a lean department — is not the good news it looks like: a three-person crew instead of four drops the insurance class further, and Finding 4 shows the household ends up worse off once the premium is set against the smaller assessment. $30M is the number to plan against; $23.5M is the number that costs households more. Fund 145 carries $12,598,661 for FY2029 and the County has adopted none of the rest. At the workshop that price goes on the public record for the first time, in a room where the Board will also see its consultant’s work for the first time. We think that sequence is backwards, and that it is the sequence that creates the risk.
Why the charge is flat is not a mystery, and nobody hid it. The County’s own legislative findings say it: Resolution 23-R-19 records that “the size or the value of the Residential Property does not determine the scope of the required fire services response.” That is the benefit rationale Florida law requires of a special assessment, and it is a defensible reading of the statute. It is also the whole mechanism: a charge apportioned by benefit rather than by capacity to pay will land hardest on the households with the least, every year, until someone chooses otherwise. Regressive by default, not by design — and the County can choose otherwise at the fire services workshop at no cost to the budget.
And on the narrow question the fight began over, we think the City has the better of it. The County was buying, for $11.8M, a service it would have to spend about $30M to replace — about 39 cents on the dollar of what replacing it would cost. Whatever one thinks of a unilateral rate rise or a 3–2 vote to cut ties, a jurisdiction asking to be paid closer to what a service costs is not making an unreasonable request, and the County’s own consultant was hired to find out what that number is. What the City has not done is publish its cost of serving the unincorporated area — and until it does, its own case rests on an assertion too.
The case against what we have just written. Every act above is defensible on the day it was taken: excluding an unevidenced insurance saving from a rate is prudent, not careless; holding a rate flat during a live dispute is ordinary; extensions are ordinary; and a Board is entitled to see a consultant’s work at a workshop, which is what workshops are for. A reader who weighs those differently will reach a different conclusion from the same record, and the record — not this section — is what we ask them to read. We allege no impropriety by anyone and impute motive to no one, and this paper still takes no position on whether the County should separate. Our recommendation is right in either branch.
| Limitation | Effect, and what reduces it |
|---|---|
| The $122,000 loaded cost per FTE is unverified — the binding limitation on every staffing figure. | Sets the fourth-firefighter cost, both 42-hour figures and the personnel share of every standalone operating figure. 277 positions at that rate is 53% of FY26 Fire Services revenue where fire personnel is normally 75–85%; if the true loaded cost is X% higher, every staffing figure is X% higher and the ordering of the paths does not change. Request A (payroll by rank, employer benefit rates). |
| The pay gap has no fixed vintage. | $44,178 may pre-date up to four wage awards; the gap to a $60,000 deputy is $4,826–15,822 and every parity figure is a range with no midpoint quoted. The error runs toward the low end. Request A’s wage tables convert it to a price. |
| The 39% inner-city first-arrival share is a proxy. | Six volunteer zones, calendar 2021, City production only; volunteer-only responses absent. Consistent across two cuts (39.1% all calls, 39.2% fire-type) and the STRUCTURE FIRE-typed subset (10 of 24). Request C’s CAD tables replace it with the countywide share for every retained year. |
| Base-versus-gross is declared, not resolved. | A 24/48 firefighter works 147 hours a year above the §207(k) threshold; whether $44,178 already covers that premium is unknown. The gap is narrower, not wider, in every case. Request A item A.3. |
| Every incident-level figure rests on two productions, not the complete CAD record. | Rural incidence and response figures are floors, because volunteer-only responses are absent; no multi-year trend may be asserted from a two-year window. A full CAD request covering all years and all agencies is drafted (Request C). |
| The second-unit interval and the entry clock are modelled. | Road-network travel at apparatus speeds, no dispatch or turnout time; a 2021 CAD sample makes the model pessimistic today (7.7 modelled against 3.7 measured in Woodville, n=5) and unverifiable for 2028 by construction. Stated as modelled wherever shown; never mixed with measured first-unit arrival. |
| Volunteer ISO credit is unknown. | The public band is 12,051–16,930 beyond five miles because of it; whether Woodville and Fort Braden keep their insurance class turns on it. One written enquiry to Verisk. |
| Three household bases are in circulation. | ~50,300 (the $234 → $597 ladder), 46,768 (the roll, the instruments, +$77), ~52,800 billed (the $1M ≈ $19 conversion, $11–35 for parity); 152,160 county-wide (+$24). Each figure keeps its base and names it; none is restated as another. The County’s billed-unit count fixes it. |
| The five-year window and its phasing are the analyst’s. | FY2029–33 is the first window in which every component appears once; a longer window widens the gap between agreement and standalone paths. No inflation, financing or debt service — which understates the standalone paths relative to the agreement paths. The ordering is the robust result; the exact figures inside each band are not. |
| Station counts are greedy upper bounds; capital is an extrapolation. | An exact solver could reach the same coverage with fewer sites; $7–10M per station is the study’s own $28–40M ÷ 4, a planning band and not a bid. Crewing dominates the annual figure, so rankings do not move. |
| The square-foot curve is borrowed, not fitted; 7.1% of units have no building area. | The $0.13 slope is Florida district precedent; only the base is solved locally. The County’s own roll and rate model would replace every instrument figure in an afternoon. |
| Tract statistics describe neighbourhoods, not households; incidence is not risk. | No statement is made about the race, income or tenure of any household. The fire rate is unadjusted for building age, construction, heating or occupancy; only the District 1 against District 4 comparison is statistically distinguishable. |
| Establishment figures, not payroll. | 81.5 = 21 posts × 3.88 FTE per post; 277 = shift-rank positions from the City’s force-by-rank production. Filled positions may differ; the County’s July 14th, 2026 request for authorized and filled positions at Stations 10–15 is unproduced. |
| Retention is not converted into response. | No data in the record links pay to turnover to response performance. The recommendation changes the denominator of a cost; it does not claim an outcome. |
| Measurement | What it takes | What it resolves | Status |
|---|---|---|---|
| One written question to ISO/Verisk on whether the six volunteer departments earn credit, and whether the county-facing companies count in the graded-area staffing average | One letter; a draft exists | The $12.0–16.9M band; whether ~3,000 households in Woodville and Fort Braden keep their class when their distance does not change; whether County-funded staffing earns grading credit | Cheapest, highest value. Drafted, not sent — and the risk figure says so on its face |
| Request A (City, Chapter 119): IAFF wage tables and step schedules since October 2021; the salary schedule with the hours it covers (A.3); payroll by rank FY2022–26 with overtime and out-of-class hours and dollars; employer benefit rates | One request to the City — drafted | Converts parity from a range to a price; settles base-versus-gross; replaces the $122,000 loaded cost; replaces the 2019 payroll vintage; makes the structural/discretionary overtime split a measurement (“no such record” is itself the finding) | Drafted, not sent |
| Request C (Consolidated Dispatch Agency): the CAD incident and unit-status tables, every retained year, every responding agency, exempt fields excluded at the query | One request to the CDA — drafted; feasibility established by PRR2026-877 | Replaces the 39% proxy with the countywide first-arrival share; replaces the modelled second-unit interval with observed; lifts the rural floors; sizes the ambulance interval behind the engine | Drafted, not sent |
| Request B (Leon County): LCEMS response-time data by incident and jurisdiction; the current ALS first-response payment (last produced $736,938 in FY2014); the Fund 135 fund-balance policy target; the fleet replacement schedule | One request to the County — drafted | Makes the EMS levy’s stated justification testable for the first time; states what the County pays now for fire-based ALS | Drafted, not sent |
| The consultant’s deliverable under B-26-069 | Delivery under the existing contract; $54,997.50 remains | A commissioned standalone cost and a consultant’s successor-agreement figure to set against the study’s model; the deployment analysis this paper does not claim | No deliverable in the production; the Board sees it at the workshop |
| A written aid price from the City for 2028 onward | One letter, or the outcome of the Recommendation 2 offer | Fixes the $0.5–1.5M aid line and decides the entire station programme; refusal is decision information in its own right | Not asked |
| The County’s own assessment roll and rate model run for a square-foot tier at the current revenue requirement | A staff afternoon with the roll the County maintains and the consultant it retains | Replaces every instrument figure with the County’s number; fills the 7.1% no-area gap; fits the curve locally | Not published |
| The County’s billed-unit count for the Fire Services Fee, by zone | County OMB; a public record | Reconciles the three household bases onto one denominator | Not requested |
| Authorized and filled positions, and minimum daily staffing, Stations 10–15 | Already requested by the County, July 14th, 2026 | Firms the 81.5 and the six-station end of the parity range; the first direct retention measure in the record | Not produced |
| Departmental registration on the critical path — FDID, NFIRS, the FSRS submission | One line on the adoption schedule | Closes RFP item D | No draft exists |
The separation is not an administrative reorganization. It moves 10,678 homes across the five-road-mile threshold, carries $2.73 billion with them, and falls hardest — as a share of what people own — on the households a flat charge treats as equal to houses worth seven times as much. I measured that myself, address by address, on the county’s own street map, and every step of it is published so that anyone can check it.
Same service on October 1st, 2028 is not available at any price — for that date the constraint is the calendar. The apparatus window closed about June 2026 on the City’s own lead times, no station opens before 2030, and the academy gate of late 2027 is the last decision still in front of the Board. For every year after that date the binding constraint is the other one: money — $23.5–30M a year to stand alone against $11.8M today, and $43.4–49.8M a year to restore the reach against a fund holding $12,598,661. Both are real; they bind at different times, and a plan that answers only one of them answers neither. I would plan against $30M and not $23.5M, because the cheaper way of standing alone is the one that moves the cost onto insurance bills.
After that date the County is a buyer with one seller and a published deadline: every path except a signed agreement runs through purchasing service from Tallahassee at whatever price it sets, because the substitute — four stations, $28–40M on my own modelled costs, none open before 2030–31 — does not exist in time, and nobody has published a County figure to check mine against. That is not anyone’s bad faith; it is the shape of what the last twelve months built, and it is the strongest reason I can find for acting in September rather than in 2027.
So here is what I would do, and all three motions can be made at the fire services workshop. First, change how the charge is collected. It is the free one and it decides who pays for everything after it: the square-foot tier ($149 up to 1,600 sq ft, then $0.13 a foot) stays inside assessment law, collects the same total, and takes a District 1 household from $223 to $182 with four of the five districts paying less; half-flat-half-value and a pure levy on value are the other two instruments on the table, each with its own law and its own losers, and Table 11 prices all four. Second, fund the service and make the written offer to the City within thirty days — a fourth firefighter on every county-facing engine, the volunteer programme and hydrant flow testing, which the County can fund alone, together with the County’s 29.4% share of department-wide pay parity, $0.56–1.85M a year, reaching 277 firefighters instead of 81.5, 3.4× for the money six-station parity would cost: $4.11–5.50M a year all in, about $27–36 a household county-wide, written as a purchase with open-book pricing, pro-rata clawback and forty-eight months’ notice, so that after 2028 the County is not a buyer with one seller and no terms. Third, start the 42-hour-week process: put it on the City–IAFF table that opens in early 2027 with the County’s arithmetic published for both sides and the County’s share stated, paired with parity if that is where the City and its union want it. Shift length is not the County’s to adopt, so the week itself stays an option — but the County can open that door, and nobody else is going to.
And one thing before any of that. The County engaged its consultant for $109,995 in December 2025, has paid $54,997.50, extended the contract 210 days to December 31st, 2026, and its own records production on that contract contains no deliverable. Extensions are ordinary and I allege nothing improper. But on the present course the Board sees that work for the first time at the workshop, in the room where the price of separation is stated publicly for the first time, while the other party reads every word. A commissioner can change that with one request: ask for the consultant’s work in writing now — before the workshop, not at it. If it is not ready, that is worth knowing before the workshop; if it is, the Board should read it before it is asked to act on it. That is the ask this paper ends on, and it costs nothing.
This paper takes no position on whether the County should separate; its recommendation is right in either branch. If I have something wrong I would rather hear it before the workshop than after 2028 — corrections are published with the source named, never quietly patched.
T26-0654_production_2026-08-17/SHA256SUMS.txt.TLC_OverlayStreetCenterline_D_WM/0), address points, parcels (TLC_OverlayPropInfo_Enhanced_D_WM/1), city limits (TLC_Overlay_Citylimits_WM_D/0, 4,628 vertices), hydrants (TLC_OverlayFireHydrants_D_WM, September 2nd, 2026), non-ad-valorem building area (BASE_SQ_FT).FIGURE_BASIS_CANONICAL.md — the study’s canonical figure basis, Rev 3.1 with §2c–§2r; every number in this paper traces to it or to a companion paper below. SPINE_2026-09-06/CONTENT_SPINE.md — every fact, finding and recommendation in reader order with status and carrier. AUDIT_2026-09-05/ — the corpus audit: figure register, dead variants, contradictions.REPORT_2026-09-06_LEON_FIRE/SHA256SUMS.txt): Who actually answers the county (paper_firstdue, item B.2/F); The disparity is real, and the instrument is the free fix (paper_disparity, item E); What each path costs over five years (paper_allin, item C); Modernising the workforce that answers the county (paper_workforce, item F); What to buy before 1 October 2028, version 2 (paper_options, items B and F; corrected September 8th, 2026 — path row); The levy nobody watched (paper_ems, item E); 1001 San Luis Road — who pays what now (paper_whopays, item E, released with errata); Where a standalone county fire system would put stations (paper_siting, items B.4 and C); What the grading schedule says (paper_iso, item B.1). Fire Protection in Unincorporated Leon County After 2028 v2 (PAPER_POST2028) is the August predecessor; its recommendation section is superseded by paper_options v2.DATA_stations_2026-09-07 (CC BY 4.0; METHOD.md, RESULTS_stations.json). August synthesis: Closing the RFP: Middle-Ground Options, the Successor Agreement, Automatic Aid, Shared Facilities, and the Board Synthesis (August 27th, 2026) and Design and Policy — Standalone Department (August 27th, 2026); both pre-date the September 7th apparatus correction and the parity range, which govern where they differ.build_our.py from the same data bundle and figure code as the County’s report (The Independent Separation Report); the study site, methods, limitations, what-would-resolve and reference sections are lifted verbatim from that paper’s source, and every literal number in this paper is asserted to appear there. Its executive summary is The Decision Behind the Numbers; the walk-through of the study is The Line. Hand-rolled SVG, no external library, no network call — until a reader opens the living map in Figure 1 (Leaflet 1.9.4 from unpkg.com, OpenStreetMap tiles, the LIVE/ folder beside this file). Version 2, September 8th, 2026 — the citizen rebuild: the abstract, introduction, so-whats and conclusions in the first person, from the county resident’s chair; no number changed. Version 1 of the same date is kept beside it unchanged as the neutral edition. License: text, figures and data of this study are published under CC BY 4.0 — reuse with attribution to Max Epstein / MAXAI LLC; the byline and the corrections address travel with any reuse. Corrections are published with the source named and never quietly patched.