Performance Benchmarks
Quantitative measurements of @daniel-murnane/core at runtime. Sourced from public bibliometric data (OpenAlex, ORCID) and self-reported throughput metrics.
Headline Metrics
const stats = await daniel.getStats();
// =>
{
worksCount: 36,
citedByCount: 369,
hIndex: 8,
i10Index: 7,
twoYearMeanCitedness: 6.14,
source: 'OpenAlex (live, see /publications)'
}A live, machine-readable copy is at /data/publications.json.
Throughput
Outputs per unit time, averaged over the most recent operational window.
| Output | Rate |
|---|---|
| Preprints / papers | ~5 per year |
| Conference talks (contributed + invited) | ~6 per year |
Code commits (across gnn4itk, exa-trkx, etc.) | bursty; long-tail distribution |
| Peer reviews completed | ~1 per month |
| Espresso shots consumed | ~3.5 per day |
Latency
| Operation | p50 | p90 |
|---|---|---|
| Email reply | 24h | 48h |
| Code review on a draft PR | 48h | 1 week |
| Detailed feedback on a paper | 1 week | 3 weeks |
| Decision on a new collaboration | 2 weeks | 6 weeks |
Latencies degrade during conference travel; check the status field before benchmarking.
Citation Distribution
The citation distribution is power-law-like, as is typical for research bibliographies. A few highly-cited works (Exa.TrkX, GNN4ITk) drive the bulk of the h-index; many shorter contributions trail behind.
const distribution = {
top5Percent: 'drives ~60% of total citations',
median: '~5 citations',
longTail: 'preprints and CHEP proceedings'
};For the full breakdown, see Publications, which renders the live OpenAlex data sorted by citation count.
Reliability
Production-relevant SLAs based on the last decade of operation.
| Metric | Value |
|---|---|
| Annual paper-shipping uptime | >99% |
| Conference attendance reliability | ~95% (modulo pandemic years) |
| Showing up to a scheduled meeting | high; ~3 sigma below 100% |
| Returning a paper review on time | improving; trending toward 100% |
Hardware Profile
The substrate has known limits and is documented for capacity planning.
const hardware = {
cores: 1, // physical
contextSwitchCost: 'see /api/limits',
sleepCycle: '~7h nightly, hard requirement',
caffeineDependency: 'present, well-tolerated',
upgradePath: 'firmware updates via espresso',
};Comparing Against Baselines
For benchmarking against other researchers, the relevant comparable bibliometric data is published on Google Scholar and ORCID. Note that h-index in particle physics + ML is not directly comparable to single-author fields; collaboration-paper inclusion conventions vary.
Bibliography Sync
The publication-level metrics here are kept current by a CI workflow that pulls from OpenAlex weekly. See scripts/sync-publications.py in the site repository for the implementation.