{"id":"bdec9bf0-658d-42cb-b6e3-67bcf79e4595","arxiv_id":"2502.08303","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":1.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"A community status report showing that lattice QCD collaborations collectively need over 10 PB of storage and increasingly plan public release via the International Lattice Data Grid.","lead":"This proceedings paper summarizes the status of lattice QCD gauge ensemble generation and data management reported by 15 collaborations at the Lattice 2024 conference. It aggregates storage requirements, public availability plans, and ILDG usage to help the community plan data infrastructure.","discovery_kind":"review","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The '>10 PB' sum in Section 3 mixes projected and existing storage; excluding ETMC's projected 3 PB drops the total below 10 PB, making the headline claim definitionally fragile.","rationale":"The reader's weakest_assumption correctly identifies that Table 1 relies on unverified self-reports from a self-selected set of collaborations. My review goes further and finds a more concrete, internal issue: the aggregate '>10 PB' figure is not a well-defined quantity because the Storage column mixes current tape/disk holdings, compressed current data, and projected future requirements, and the threshold is crossed only when at least one major projected entry (ETMC, 3 PB) is included. This is checkable from the paper itself and directly affects the strongest claim, so it is more load-bearing than the general sampling concern. That said, the reader's conditional verdict already captures the need for caveats, and my concern does not invalidate the paper's useful role as a proceedings snapshot; it strengthens the case for CONDITIONAL. I do not see grounds to reject or accept outright, so the verdict remains UNCHANGED. Minor issues such as the 16-vs-15 collaboration count in the abstract/Section 1 also support conditional acceptance but are less consequential than the storage misaggregation.","tokens_in":20705,"tokens_out":4744,"duration_ms":50448,"concrete_test":"Recompute the Section 3 total after separating Table 1 Storage entries into 'existing holdings' versus 'projected/planned requirements', using the labels explicit in Section 2 prose (notably ETMC 'expected ... 3 PB' and OpenLat 'expected ... 0.5 PB'). If the sum over existing holdings alone is below 10,000 TB, the abstract's '>10 PB' claim depends on including projected future capacity and should be reworded or separately reported. As a cross-check, sum the column with and without the ETMC entry and confirm the threshold flips.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central claim that 'total storage requirements reported by the collaborations are expected to exceed 10 PB' rests on summing the Storage column of Table 1, but the column entries are not commensurable. Reading Section 2 prose against Table 1: ETMC's 3,000 TB is an expected future requirement ('we expect storage requirements to reach 3 PB'), OpenLat's 500 TB is 'expected to be required,' and HAL QCD's 400 TB includes both stored configurations and derived gauge-fixed rotated copies. Meanwhile Jlab/W&M's 3,500 TB, HotQCD's 1,500 TB (compressed fp32), and CLS's 1,400 TB (tape) are current holdings. Summing all 15 Table 1 entries gives 11,595 TB ≈ 11.6 PB, but removing the single projected ETMC entry (3,000 TB) yields 8,595 TB, below the 10,000 TB threshold asserted in the abstract and Section 3. The summary does not disclose which entries are projections, so the headline number is not robust to a minimal, defensible reclassification. This is load-bearing because the paper's main quantitative contribution is this aggregate storage estimate; if the threshold is only crossed by including future projected requirements, the statement 'the lattice QCD community faces a storage requirement above 10 PB' overstates the current, verifiable burden and conflates existing data with planned production.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"This proceedings paper summarizes a dedicated parallel session at Lattice 2024, organized by the International Lattice Data Grid (ILDG), in which representatives of lattice QCD collaborations reported on their gauge ensemble generation programs, data management practices, publication and embargo plans, ILDG engagement, and storage requirements. The paper provides a short narrative for each of the 15 collaborations included in the written proceedings and collects the reported information in Table 1. Its headline conclusions are that most collaborations plan to release ensembles publicly after an embargo period, that there is strong interest in ILDG infrastructure, and that the total reported storage requirements are expected to exceed 10 PB.","tokens_in":20940,"tokens_out":8001,"duration_ms":85015,"significance":"As a descriptive community survey, the paper is a useful archival record: it identifies who is generating which ensembles, where the data are stored, which formats and metadata conventions are used, and what sharing policies are planned. The per-collaboration entries and the consolidated table are the kind of reference information that the lattice QCD community and funding agencies can use for storage planning. The paper does not contain new derivations, machine-checked proofs, or parameter-free predictions, so its value rests on accuracy, internal consistency, and transparency of the reported data. The central quantitative claim, however, is currently fragile because Table 1 mixes current holdings with projected requirements; with the projected ETMC entry excluded, the tabulated sum falls below the 10 PB threshold asserted in the abstract and Section 3. With a corrected and clearly labeled table, the paper would be a solid and genuinely useful community reference.","major_comments":[{"comment":"The headline claim that 'total storage requirements reported by the collaborations are expected to exceed 10 PB' is not supported by the table as currently defined. Summing all 15 Storage entries gives 11,595 TB, but the entries are not commensurable: Jlab/W&M (3,500 TB), HotQCD (1,500 TB), and CLS (1,400 TB) are current holdings (Sections 2.2, 2.3, and 2.14), OpenLat's 500 TB is 'expected to be required' (Section 2.9), ETMC's 3,000 TB is explicitly a future expectation ('we expect storage requirements to reach 3 PB', Section 2.11), and HAL QCD's 400 TB includes 320 TB of derived gauge-fixed rotated copies (Section 2.6). If the single projected ETMC entry is excluded, the sum is 8,595 TB, below the 10,000 TB threshold stated in both the abstract and Section 3. Please separate current stored data from projected total requirements, for example with two columns or a clear footnote, and rephrase the aggregate statement so that it does not conflate existing data with planned production.","section":"Section 3 and Table 1; cf. Section 2.11"},{"comment":"The paper moves from a self-selected set of collaborations that responded to an open call (Section 1) to the statement that the lattice QCD community 'faces a storage requirement above 10 PB' (Section 3). The table contains no information about non-responding collaborations, and no caveat or uncertainty is attached to the aggregates, which are entirely self-reported estimates. At minimum, the summary should state explicitly that the >10 PB figure applies only to the participating collaborations and that the estimates were not independently verified; otherwise the scope of the central claim exceeds the evidence presented.","section":"Section 1 and Section 3"},{"comment":"The legend defines Public as 0, 1, or 2, but the HotQCD, FASTSUM, and PACS rows contain the compound value '2,1'. An entry cannot simultaneously mean 'currently public' and 'after an embargo period', and this ambiguity makes it impossible to verify the Section 3 statement that most collaborations plan public release after an embargo. Please split the column into current status and future sharing policy, or define compound codes explicitly.","section":"Table 1, Public column"}],"minor_comments":[{"comment":"The abstract says '16 lattice QCD collaborations', while Section 1 refers to 'the 15 collaborations that contributed to this proceedings contribution' and Table 1 has 15 rows; please reconcile these numbers, for example by noting that 16 collaborations presented and 15 contributed written summaries.","section":"Abstract and Section 1"},{"comment":"Section 2.10 states that RC* currently has '80 Tb of data stored', but Table 1 lists 60 TB for RC*; please align the two numbers and specify whether the discrepancy is due to units, timing, or a table error.","section":"Section 2.10 and Table 1"},{"comment":"The Storage column would benefit from a footnote stating whether listed values are current disk/tape holdings, projected end-of-project requirements, or include derived quantities; in particular, HAL QCD's 400 TB combines ~80 TB of primary configurations with ~320 TB of gauge-fixed rotated copies, and this should be transparent.","section":"Table 1 and Section 2.6"},{"comment":"The reference list contains numerous garbled strings and author names, for example Ref. [1] with 'zero.alt' artifacts and Ref. [7] with a corrupted ILDG working-group author name; in addition, there are typographical errors such as 'We plan to to make' in Section 2.6, 'anistropy' in Section 2.4, and 'why' for 'way' in Section 2.10. These should be cleaned before publication.","section":"References and text"}],"recommendation":"major_revision","confidential_remarks":"The paper is a community survey rather than a technical derivation, so the evaluation hinges on the integrity of Table 1 and the alignment between the abstract, the table, and the summary. The >10 PB claim is the main quantitative contribution, and it is currently sensitive to the inclusion of one projected entry; this needs to be fixed before the paper can be accepted. The ambiguous Public codes and the 16-versus-15 count are further signs that the table and abstract need careful revision. I would support acceptance after a major revision that makes the storage data commensurable and clearly labeled."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"The short version: this is a competent, useful proceedings survey of lattice QCD ensemble generation and data management as of Lattice 2024, the sequel to the Lattice 2022 report, and it doesn't pretend to be more. Its one quantitative headline, storage requirements above 10 PB, is directionally right but more fragile than the text admits.\n\nWhat's new: the 2024 snapshot. Fifteen collaborations state their sharing policies, formats, and storage numbers in one table, and that table is the real deliverable. The paper usefully documents the near-universal plan to release ensembles publicly after an embargo, the strong interest in ILDG, and the trend toward finer lattices and physical-mass ensembles. Anyone writing about community data infrastructure, or anyone asking for money to store lattice data, will cite this paper.\n\nThe soft spots, in proportion. The 10 PB figure mixes incompatible things. The Table 1 storage column sums to about 11.6 PB, but Jlab/W&M's 3.5 PB, HotQCD's 1.5 PB, and CLS's 1.4 PB are existing holdings, while ETMC's 3,000 TB is a projection (\"we expect storage requirements to reach 3 PB\") and OpenLat's 500 TB is anticipated for a program still in its early stages. Remove the ETMC projection and the sum sits around 8.6 PB, below the claimed threshold. The word \"expected\" in the summary does some hedging, but the table doesn't mark which entries are current vs projected, and the abstract doesn't hint at the mix. Read carefully, the claim is about forward-looking requirements, which is defensible; read casually, it overstates the current, verifiable burden. One explicit sentence would fix it.\n\nTwo smaller things. The abstract says 16 collaborations; the body and footnote say 15. Minor, but it is exactly the kind of slip that makes readers distrust the table. The Public column also mixes codes like \"2,1\" without a note; you can infer \"some public now, some after embargo,\" but it is not stated.\n\nBottom line: this is a survey, not a research contribution, and it labels itself as such. The self-reported numbers carry no independent verification, and a single collaboration's projection moves the headline. For its actual purpose, community coordination, ILDG planning, and funding advocacy, it does the job. Send it to review, and ask for the projected-vs-existing disclosure and a consistent collaboration count. I would cite it if I were writing about ILDG or lattice data policy.","headline":"A useful community survey whose >10 PB headline is real but fragile: it holds only if projected future storage is included, and the paper never flags which entries are projections.","tokens_in":21511,"tokens_out":6669,"would_cite":true,"duration_ms":63067,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":["12.38.Gc"],"model":"deepseek-v4-flash","headline":"Lattice QCD storage needs top 10 PB, survey says, with most collaborations planning public data releases after an embargo.","keywords":["lattice QCD","gauge ensembles","data management","International Lattice Data Grid","data sharing","storage requirements","FAIR data","ensemble generation"],"falsifier":"Compare the survey's per-collaboration storage figures against the actual disk and tape inventories at the computing centers that host the ensembles; if the sum of independently verified totals falls clearly short of 10 PB, or if the largest reported dataset cannot be located, the paper's headline claim fails. A second check is to revisit the same collaborations in two years and count how many have actually uploaded configurations to ILDG versus how many planned to.","tokens_in":20515,"feed_emoji":"💾","tokens_out":6520,"duration_ms":62696,"temperature":0.7,"pith_summary":"These proceedings report a community survey conducted at a dedicated session of the Lattice 2024 symposium. Representatives of 16 lattice QCD collaborations described their gauge ensemble generation efforts, focusing on publication plans, data management, and storage needs. The paper's central claim is that most collaborations intend to make their ensembles publicly available after an embargo period, that there is strong interest in the International Lattice Data Grid (ILDG) as the sharing infrastructure, and that the total reported storage requirements exceed 10 PB. A sympathetic reader would care because these multi-year simulation datasets are the raw material for precision calculations in QCD, and the community is at a decision point about how to preserve and share them.","feed_headline":"Lattice QCD storage needs top 10 PB, survey says","feed_subtitle":"Most collaborations plan public gauge ensembles after an embargo, and the data grid is the preferred route.","key_machinery":"The carrying object is the ensemble survey table (Table 1), built from structured self-reports of public status, ILDG usage, ensemble count, configuration count, and storage in terabytes. The other central piece is the ILDG itself, a federation of regional grids that standardizes storage interfaces and a searchable metadata catalog, together with the QCDml metadata scheme for describing gauge configurations. These tools supply the shared vocabulary that turns fifteen separate collaborations' data practices into a single community-wide picture of storage demand and sharing intentions.","core_discovery":"On its own terms, the paper establishes that the lattice QCD community has reached a point where open data is a planned norm rather than an exception. The 15 written collaboration reports collected in Table 1 show a spread of storage footprints from roughly 15 TB to 3.5 PB per group, and the sum of the self-reported totals exceeds 10 PB. Most groups mark their sharing status as public after an embargo, and a majority express interest in, or active use of, the International Lattice Data Grid, with a handful already uploading configurations. The survey also documents a push toward physical quark masses, larger physical volumes, and finer lattice spacings down to about 0.04 fm, each of which increases the value and the cost of preserving the ensembles.","pith_inferences":["The 10 PB figure is likely an underestimate because it excludes collaborations that did not respond, derived datasets such as gauge-fixed copies, and analysis output; one report, for instance, counts a 320 TB gauge-fixed set separately from its 80 TB raw configurations.","The storage distribution is highly skewed, with one collaboration reporting 3.5 PB alone, so the overall number is fragile to a single self-report error; a systematic audit of a few of the largest figures would harden the survey.","If actual ILDG uploads lag the stated plans in the next two years, the gap would measure the frictions, such as embargo policies, storage space, and metadata effort, that an attitudinal survey cannot see."],"forward_implications":["If the reported plans hold, most existing and future gauge ensembles become publicly available after an embargo, enabling independent checks of published results.","A storage requirement above 10 PB means funding agencies and regional facilities must treat persistent data preservation as a first-class resource alongside computing time.","The preference for ILDG implies a federated access point where a researcher can search one metadata catalog and download ensembles from multiple regional grids.","The trend to finer lattice spacings and larger volumes will push storage needs further, since each ensemble at 0.04 fm costs many times more bytes than coarser ones.","Standardized metadata with provenance tracking will make ensembles reusable for years after the generating collaboration moves on."],"supporting_citations":[{"why":"Report of the previous Lattice 2022 data session, setting the template for this one.","marker":"[1]"},{"why":"Initial paper establishing the International Lattice Data Grid concept.","marker":"[2]"},{"why":"Early account of using the ILDG for sharing configurations.","marker":"[3]"},{"why":"Operational guide for joining and using the ILDG.","marker":"[4]"},{"why":"Technical description of building the ILDG federation and its components.","marker":"[5]"},{"why":"Defines the QCDml metadata schema used to mark up ensembles.","marker":"[7]"},{"why":"The FAIR principles that the ILDG metadata and middleware are designed to follow.","marker":"[8]"},{"why":"Status report on ILDG 2.0 presented at the same session, the infrastructure the plans refer to.","marker":"[9]"}],"fun_headline_variants":["Lattice QCD storage to top 10 PB, open data planned","Data grid preferred as lattice QCD shares ensembles","Lattice QCD: >10 PB storage, most groups to share","Open data becomes norm for lattice QCD ensembles"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The aggregate storage total and the sharing trends rest entirely on unverified self-reported numbers from the collaborations that chose to answer the open call, so a biased sample or a wrong estimate would change the conclusions.","fun_headline_variants_meta":{"raw":{"variants":["Lattice QCD storage to top 10 PB, open data planned","Data grid preferred as lattice QCD shares ensembles","Lattice QCD: >10 PB storage, most groups to share","Open data becomes norm for lattice QCD ensembles"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000352,"raw_usage":{"total_tokens":1829,"prompt_tokens":770,"completion_tokens":1059,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":386,"completion_tokens_details":{"reasoning_tokens":990}},"tokens_in":386,"tokens_out":1059,"duration_ms":8278,"temperature":1.0,"reasoning_tokens":990,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-08T05:37:28.365974+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Compare the survey's per-collaboration storage figures against the actual disk and tape inventories at the computing centers that host the ensembles; if the sum of independently verified totals falls clearly short of 10 PB, or if the largest reported dataset cannot be located, the paper's headline claim fails. A second check is to revisit the same collaborations in two years and count how many have actually uploaded configurations to ILDG versus how many planned to.","supporting_citations":[],"review_version":1}