{"id":"17640311-40b9-4666-9417-13045c759096","arxiv_id":"2411.18633","paper_version":1,"verdict":"REJECT","confidence":"MODERATE","novelty_score":5.0,"correctness_risk":"high","formal_verification":"none","parameter_count":6,"one_line_summary":"A geospatial techno-economic model estimates that FTTnb fiber broadband is viable for only about 48% of Sub-Saharan Africa's population, with per-user costs and emissions far higher in low-density areas.","lead":"This paper models the cost, carbon emissions, and social carbon cost of building fiber-to-the-neighborhood networks across 44 Sub-Saharan African countries, using two spatial optimization algorithms and a life-cycle assessment. It reports that per-user costs and emissions are dramatically higher in sparsely populated areas, and concludes that about 48% of the region's population is currently viable for such networks.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The 48% viability figure is arithmetically inconsistent with the paper's own geotype table: Deciles 1-4 sum to about 553 million (48%), but the text says Deciles 1-5 are viable, and those sum to about 688 million (60%).","rationale":"In good faith, the paper's qualitative finding that lower population density leads to higher per-user cost and emissions is plausible and consistent with prior techno-economic work. However, the headline quantitative claim does not survive an internal consistency check. The abstract's 48% and 550 million figures match only Deciles 1-4 in Table 3, while the discussion and conclusion specify a viability threshold that includes Decile 5. This makes the central claim ambiguous at best and arithmetically wrong at worst, independent of any external data or modeling assumptions. The reader's identified weakest assumption, the unstated adoption rate in Equation (2), is a serious secondary problem for per-user cost and emission values, but the 48% mismatch is more directly load-bearing because it attacks the paper's headline result without requiring any additional parameter. Since the reader already recommended rejection, this stress-test confirms that verdict rather than moving it.","tokens_in":24303,"tokens_out":7527,"duration_ms":69265,"concrete_test":"Recompute from Table 3: sum the populations of Deciles 1-4 and Deciles 1-5 and express each as a percentage of the reported total of 1,154,766,136. The sums are 553,543,408 (47.9%) and 688,034,765 (59.6%). Then check the stated viability threshold: if viable is density above 106 per km2 (so Decile 5 is included), the reported 48% and 550 million figures are arithmetically wrong; if viable is density above 172 per km2 (so only Deciles 1-4 count), then the text saying Deciles 1-5 are viable is wrong. Either way, correcting the arithmetic or explicitly redefining the threshold would settle whether the headline claim can stand.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central claim in the abstract and conclusion is that 48% (about 550 million) of the Sub-Saharan African population can be viably served by FTTnb within ten years. The body defines viability by population density: the discussion says the approach is unviable in places with less than 106 people per km2 (Decile 6-10), and the conclusion states that the first five population deciles (Decile 1-5) are viable. But Table 3 gives cumulative populations of 553,543,408 for Deciles 1-4 and 688,034,765 for Deciles 1-5, which are 47.9% and 59.6% of the stated total of 1,154,766,136, respectively. The 48% and 550 million figures match only Deciles 1-4, not Deciles 1-5. Thus the paper either silently excludes Decile 5 (population density 107-171 per km2), contradicting the stated 106 per km2 threshold, or the threshold includes Decile 5 and the viable share should be about 60%, not 48%. This is not a parameter-uncertainty issue; it is an internal arithmetic inconsistency in the paper's strongest quantitative claim.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"This manuscript presents a geospatial techno-economic and life-cycle assessment of Fiber-To-The-Neighborhood (FTTnb) deployment across 44 Sub-Saharan African countries. The authors combine WorldPop population data, existing fiber maps, road networks, and two spatial optimization algorithms (Minimum Spanning Tree Prim and Prize Collecting Steiner Tree) to compute total cost of ownership, greenhouse gas emissions, and social carbon cost per user across ten population-density deciles. The headline finding is that FTTnb is viable for 48% (about 550 million people) of the Sub-Saharan African population within ten years, with per-user emissions and costs far higher in sparsely populated regions. The paper also reports aggregate investment costs of US$25-26 billion and compares them to SSA GDP.","tokens_in":24593,"tokens_out":6245,"duration_ms":49338,"significance":"If the quantitative claims held, the paper would be a useful first-order benchmark for broadband infrastructure planning in Sub-Saharan Africa, and the integrated treatment of cost, emissions, and social carbon cost across granular geotypes is a genuine contribution. The authors make their code and data available, and they explicitly compare an idealized MST design with a road-constrained PCST design, which is a reasonable way to bracket network costs. However, the central viability figure is internally inconsistent with the paper's own table, and key parameters (notably the adoption rate) are not reported, so the headline numbers are not currently reproducible. The qualitative relationship between population density and per-user cost/emissions is structurally sound and likely robust, but the specific quantitative conclusions need substantial revision.","major_comments":[{"comment":"The paper's central claim that 48% (about 550 million) of the SSA population can be viably served is arithmetically inconsistent with Table 3. The cumulative population of Deciles 1-4 is 553,543,408 (47.9% of 1,154,766,136), while Deciles 1-5 sum to 688,034,765 (59.6%). The text in §V says the approach is unviable below 106 people/km2 (Decile 6-10) and the conclusion says Decile 1-5 are viable. Since Decile 5 has a minimum density of 107 people/km2, the stated threshold includes Decile 5, which would imply roughly 688 million (60%), not 550 million (48%). The 550 million figure corresponds to Deciles 1-4, contradicting the 106 people/km2 threshold. The abstract, conclusion, and discussion must be reconciled; as written, the paper's strongest quantitative claim is not internally consistent.","section":"Abstract; §VI Conclusion; Table 3"},{"comment":"Eq. (2) defines Userskm2 = Pop(km2) × ADr, and all per-user TCO, emissions, and SCC results are derived by dividing network totals by this quantity. However, the paper never states the value of ADr used in the scenarios. Table 1 lists the main settlement population, fiber node buffer, and many cost parameters, but no adoption rate. The only clue is an example (\"an adoption rate of 0.5%\") and a statement that \"the model is set to different adoption rates,\" without specifying the actual values. Because every per-user figure (e.g., US$0.29 vs. US$36 annualized TCO; 0.18-9.6 kg CO2 eq./user) scales linearly with 1/ADr, the results are unverifiable and could change by orders of magnitude under a different take-up assumption. The authors must state the adoption rate(s) and, ideally, provide a sensitivity analysis.","section":"§III(a), Eq. (2); Table 1"},{"comment":"The MST Prim results are presented as a primary least-cost design, but the text concedes that \"the distance connecting the nodes is not necessarily realistic as the algorithm calculates the Euclidean distance between the nodes.\" Unlike the PCST design, which uses road data from Overture, the MST design connects nodes by straight lines that ignore terrain, existing roads, and right-of-way constraints, systematically underestimating fiber length, trenching, cost, and emissions. The headline ranges in the abstract combine MST and PCST values (e.g., 0.18-9.6 kg CO2 eq./user), so the lower bounds are unrealistic. The authors should either restrict headline claims to the road-constrained PCST results or explicitly report MST as an idealized lower bound with appropriate caveats, rather than treating it as a realistic least-cost design.","section":"§IV, 'Fiber design', Fig. 8-9; §III(b)"},{"comment":"The term \"viable\" is never defined by an explicit economic or environmental criterion. The paper labels populations in high-density deciles as viable and low-density deciles as unviable, but this is essentially a restatement of the population-density classification: settlements above the 20,000 main-settlement threshold are connected, and the resulting high-density deciles are then called viable. No benchmark (e.g., TCO per user relative to GDP per capita, a monthly affordability threshold, a payback period, or a maximum allowable SCC) is applied. Consequently, the 48%/60% viability conclusion is circular with respect to the chosen deciles and does not follow from the cost and emissions calculations themselves. The authors should define a transparent viability threshold and apply it to the computed TCO and emissions per user.","section":"§V, Research Question 2; §VI Conclusion"}],"minor_comments":[{"comment":"The reported MST Decile 1 annualized regional emissions are given as 0.0015 kg CO2 eq./user in the text, while the abstract and conclusion cite 0.015 kg CO2 eq./user; please clarify which value is correct.","section":"§IV(b), Emission Results"},{"comment":"Table 1 lists \"Fiber node buffer km 2\" with value \"2\" but the unit is ambiguous; the text says a two-kilometer buffer, so the row should read \"km\" or \"km²\" as appropriate.","section":"Table 1"},{"comment":"The sentence \"This variance justifies the rational need for operators to build FTTnb to the most populated areas first\" is awkward and should be reworded.","section":"§V, Research Question 2"},{"comment":"In the Conclusion, \"covering ~550 million people\" conflicts with Table 3; after fixing the arithmetic, update this number consistently throughout the abstract, discussion, and conclusion.","section":"§VI Conclusion"},{"comment":"Reference [46] is cited for the 20,000 settlement threshold and several cost parameters; given its central role, consider adding a brief description of how these values were transferred to the Sub-Saharan African context.","section":"§III(e), Table 1"}],"recommendation":"major_revision","confidential_remarks":"The paper's qualitative finding—that per-user costs and emissions rise steeply as population density falls—is plausible and consistent with prior work. The main problems are internal consistency and missing parameter disclosure, both of which are fixable in revision. I would not reject outright, but the headline claim must be corrected and the adoption rate and viability criterion must be specified before publication. The authors report funding from Pozibl Inc and declare no conflicts; given the company's role, it may be worth asking for a more detailed conflict-of-interest statement, though this is not a technical issue."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Two things to know before you read this one. First, it is the first sub-national, continent-wide assessment of fiber-to-the-neighborhood costs, emissions, and social carbon cost for Sub-Saharan Africa, and the modeling infrastructure is genuinely reusable: WorldPop population, GADM boundaries, AfterFibre core fiber, Overture roads, a transparent LCA framework, and a linked GitHub release. Second, the headline number is wrong on the paper's own arithmetic. The abstract says 48% (about 550 million) of SSA is viable. Table 3 shows Deciles 1-4 sum to 553 million (47.9%), while Deciles 1-5, which the text explicitly calls viable via the 106 people/km2 threshold, sum to 688 million (59.6%). The 'first five deciles, covering ~550 million' sentence in the conclusion is internally contradictory. This is a load-bearing inconsistency, not a rounding issue.\n\nWhat the paper does well: the central qualitative relationship—per-user cost and emissions fall steeply as population density rises—is almost certainly correct and matches prior techno-economic work. The authors are honest about the MST baseline's Euclidean-distance limitation and use PCST on road networks as the realistic design. They also state limitations (raw material extraction excluded, parameter uncertainty) and ship code and data, so the analysis is checkable.\n\nThe soft spots, in rough order. (1) The adoption rate ADr appears in equation (2) but no value is reported anywhere. Every per-user cost and emission number scales inversely with it. This is unverifiable as written and should have been in Table 1. (2) The abstract's 'Annualized TCO per user is 12-90 times lower' attribute is actually the emissions ratio from the body; the TCO ratios in the discussion are roughly 25-125 times. The abstract mixes the two. (3) The viability threshold of 106 people/km2 is implicit and derived from the decile structure; the 48% claim is partly a restatement of that threshold. The reader's circularity concern has some force, though the cost/emissions results aren't fit to the target. (4) The headline ranges blend MST and PCST, which are very different designs; a single primary scenario or explicit scenario ownership would help.\n\nWho this is for: infrastructure policy analysts and techno-economic modelers. The qualitative message is worth taking seriously, but the quantitative headline should not be quoted until the adoption rate is stated, the decile arithmetic is reconciled, and the abstract is corrected. I'd send it to peer review with major revisions; the method, data release, and policy relevance justify referee time. If the authors fix these, it becomes a useful reference.","headline":"The first FTTnb cost/emissions assessment for SSA has a solid qualitative story and reusable code, but the 48% headline is arithmetically inconsistent with its own Table 3, and the adoption rate is missing.","tokens_in":25104,"tokens_out":3351,"would_cite":false,"duration_ms":29732,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":[],"model":"deepseek-v4-flash","headline":"Fiber broadband is viable for 550 million Sub-Saharan Africans","keywords":["Fiber-to-the-Neighborhood","Sub-Saharan Africa","broadband viability","spatial optimization","Steiner tree","life cycle assessment","carbon emissions","total cost of ownership"],"falsifier":"Recompute the model for the entire region with a stated adoption rate, for example 1%, and compare the resulting per-user TCO and emissions for Decile 1 and Decile 10 with the paper's reported figures; if the numbers change by more than the paper's Monte Carlo range, the unstated adoption rate is the controlling factor. A direct field test is to collect actual per-household fiber connection costs from one rural SSA operator and compare them with the predicted US$33-36 per user in Decile 10.","tokens_in":24096,"feed_emoji":"📡","tokens_out":8531,"duration_ms":72297,"temperature":0.7,"pith_summary":"This paper aims to establish where, and at what cost and carbon price, Fiber-To-The-Neighborhood (FTTnb) broadband can be viably built across Sub-Saharan Africa. It combines population-density geotypes with two least-cost network routing algorithms to produce per-user total cost of ownership, life-cycle greenhouse-gas emissions, and social carbon cost estimates for 44 countries. The central result is that 48% of the regional population, about 550 million people, lives in areas where FTTnb is economically viable within ten years, while the remaining 52% would require subsidies or alternative technologies. This matters because fixed fiber is the future-proof backbone for wireless access, and investment decisions currently lack sub-national cost and emissions evidence.","feed_headline":"Fiber broadband is viable for 550 million Sub-Saharan Africans","feed_subtitle":"Per-user costs and emissions run 12-90 times lower in dense cities than sparse rural areas","key_machinery":"The central machinery is a pair of Steiner-tree routing algorithms: Prim's Minimum Spanning Tree (MST) and the Prize-Collecting Steiner Tree (PCST). These choose the least-cost set of fiber links between population settlements along road networks (PCST) or direct Euclidean lines (MST), and the resulting route distances feed the total-cost-of-ownership and life-cycle-assessment equations that produce all per-user cost, emission, and social-carbon results.","core_discovery":"The paper finds that building FTTnb in sparsely populated areas (below 9 people per square kilometer) costs 12-90 times more per user and emits 12-90 times more CO2 equivalent per user annually than in areas above 958 people per square kilometer, with the exact multiplier depending on the routing algorithm. Across the region, total investment is about US$25-26 billion, or 1.2-1.3% of SSA's annual GDP. Because only the first five population deciles (above roughly 106 people per square kilometer) can be connected at reasonable per-user total cost of ownership, the authors conclude that about 550 million people, 48% of the total population, can be viably served by FTTnb within the next ten years.","pith_inferences":["The unstated adoption rate is the single most sensitive parameter for the per-user metrics; publishing a value or a sensitivity sweep would let planners convert population into expected connections and re-benchmark the 550-million figure.","The same modeling chain—geotype demand, Steiner routing, and life-cycle assessment—could be re-run for other low-income regions such as South Asia or Central America to produce comparable universal-broadband viability maps.","The 48% viability estimate likely depends on the 20,000-person settlement threshold; lowering that threshold to include smaller towns would raise the viable population share but also raise costs, so a threshold sensitivity analysis would test robustness.","Because FTTnb stops at the neighborhood, the last-mile wireless access costs and emissions are excluded; including them could change which geotypes are truly affordable and shift the viability boundary."],"forward_implications":["Building FTTnb only in the first five population deciles (above roughly 106 people per square kilometer) would reach about 550 million people for roughly US$25-26 billion, about 1.2-1.3% of SSA's annual GDP.","Per-user costs and emissions rise steeply below about 106 people per square kilometer, so operators and governments should prioritize fiber push into towns and dense rural clusters before considering sparse regions.","In sparsely populated areas (below 9 people per square kilometer), per-user emissions are 12-90 times higher with MST and 49-80 times higher with PCST than in areas above 958 people per square kilometer, implying a much higher carbon price per rural connection.","The social carbon cost of rural FTTnb deployment is 12-93 times higher per user (MST) and 49-85 times higher (PCST) than urban deployment, so environmental cost should be part of subsidy decisions.","The choice of routing algorithm materially changes viability estimates: PCST routes along real roads and skips some nodes, yielding longer routes and higher emissions than MST."],"supporting_citations":[{"why":"Provides the main settlement threshold (20,000 people) and the SSA-specific simulation approach for affordable universal broadband that this model extends.","marker":"[9]"},{"why":"Supplies most cost-model parameters (civil construction, installation, transport, Opex) and the fiber node buffer used in the TCO calculations.","marker":"[46]"},{"why":"Provides the rural fixed-access cost methodology and the OLT and ODF capital cost values used in the Capex model.","marker":"[32]"},{"why":"Supplies the population raster used to compute population densities and geotype deciles across SSA.","marker":"[89]"},{"why":"Provides the existing core fiber network data used as the starting nodes for the optimization algorithms.","marker":"[92]"},{"why":"Supplies the road network data that the PCST algorithm uses to route fiber links realistically.","marker":"[93]"},{"why":"Provides the material inventory (PCB, plastics, steel) and associated emission factors for the terminal-node manufacturing phase.","marker":"[95]"},{"why":"Supplies international shipping emission factors and power-per-node values from a fiber LCA case study used in the operations phase.","marker":"[96]"},{"why":"Provides government conversion factors for electricity and material emissions used in the operational and EOLT phases.","marker":"[97]"},{"why":"Supplies the social cost of carbon value at a 2.5% discount rate used to monetize emissions.","marker":"[87]"}],"fun_headline_variants":["Fiber broadband viable for 550 million in Sub-Saharan Africa","SSA fiber: per-user costs up to 90x lower in dense cities","Only 48% of SSA population can get viable fiber","Rural SSA fiber: 90x higher costs and emissions per user","Dense SSA cities make fiber broadband sustainable"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The per-user cost, emission, and social-carbon figures all multiply population density by an adoption rate whose value the paper never states, so every per-user number depends on an unstated take-up percentage.","fun_headline_variants_meta":{"raw":{"variants":["Fiber broadband viable for 550 million in Sub-Saharan Africa","SSA fiber: per-user costs up to 90x lower in dense cities","Only 48% of SSA population can get viable fiber","Rural SSA fiber: 90x higher costs and emissions per user","Dense SSA cities make fiber broadband sustainable"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000409,"raw_usage":{"total_tokens":2115,"prompt_tokens":933,"completion_tokens":1182,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":549,"completion_tokens_details":{"reasoning_tokens":1092}},"tokens_in":549,"tokens_out":1182,"duration_ms":11205,"temperature":1.0,"reasoning_tokens":1092,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-12T20:33:30.063934+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Recompute the model for the entire region with a stated adoption rate, for example 1%, and compare the resulting per-user TCO and emissions for Decile 1 and Decile 10 with the paper's reported figures; if the numbers change by more than the paper's Monte Carlo range, the unstated adoption rate is the controlling factor. A direct field test is to collect actual per-household fiber connection costs from one rural SSA operator and compare them with the predicted US$33-36 per user in Decile 10.","supporting_citations":[{"cited_title":"Policy options for broadband infrastructure strategies: A simulation model for affordable universal broadband in Africa,","cited_arxiv_id":null,"evidence_quote":"Provides the main settlement threshold (20,000 people) and the SSA-specific simulation approach for affordable universal broadband that this model extends."},{"cited_title":"Policy choices can help keep 4G and 5G universal broadband affordable,","cited_arxiv_id":null,"evidence_quote":"Supplies most cost-model parameters (civil construction, installation, transport, Opex) and the fiber node buffer used in the TCO calculations."},{"cited_title":"A cost study of fixed broadband access networks for rural areas,","cited_arxiv_id":null,"evidence_quote":"Provides the rural fixed-access cost methodology and the OLT and ODF capital cost values used in the Capex model."},{"cited_title":"Carbon emissions of 5G mobile networks in China,","cited_arxiv_id":null,"evidence_quote":"Provides the material inventory (PCB, plastics, steel) and associated emission factors for the terminal-node manufacturing phase."},{"cited_title":"Government conversion factors for company reporting of greenhouse gas emissions,","cited_arxiv_id":null,"evidence_quote":"Provides government conversion factors for electricity and material emissions used in the operational and EOLT phases."}],"review_version":1}