Evidence makes the score.

Seven visible components

Each survey rates litter, unified Sidewalk Grime (embedded dirt, gum, stains, oil, residue and discoloration), curb and gutter conditions, trash overflow, graffiti and stickers, tree pits and the public realm, and Obstructions. Higher numbers mean cleaner conditions. The weighted average is rounded to the nearest whole point.

Obstructions: usable pedestrian space

Assess actual interference with pedestrian passage from vehicles, scooters, bags, bulk items, excessive merchandise, construction materials or abandoned objects. Objects merely being visible is not a negative condition. Do not penalize ordinary street trees, approved furniture, hydrants, accessibility infrastructure or normal utilities merely for existing. Describe observable interference; do not make unsupported legal accusations.

Model update and historical evidence

StreetQ model 2 combines overlapping surface categories and assigns 10% to actual obstructions. Litter remains 25%; surface grime becomes 20%; gutter 15%; trash management 15%; graffiti 5%; tree pits/public realm 10%. Historical observations retain their original JSON and computed totals. Historical Sidewalk Grime = (3 ? grime + 2 ? gum/stains) / 5, preserving the former 15% and 10% contributions. Historical Obstructions is unknown, never assumed clean. Old offline submissions retain the original model until edited and fully reassessed. Mixed-era block totals use the stored original observation scores.

Indexed is not scored

Canonical geography covers NYC; indexing does not imply cleanliness evidence. Surveyed coverage counts active blocks with at least one approved observation. Rankings require at least three approved observations and at least 50% current, decayed confidence. Unscored blocks remain available on the map and eligible for their first survey.

Fresh evidence matters more

Only approved observations count. Each observation has a 30-day half-life: a survey from 30 days ago carries half the weight of one taken today. The StreetQ Score is the recency-weighted average of eligible observation scores. Future-dated observations are excluded.

Confidence, not false precision

Confidence is 100 × (1 − exp(−effective observations / 3)), rounded to an integer. Effective observations are the sum of recency weights. Confidence decreases as evidence ages. It is an evidence-volume indicator, not a statistical probability or an independent-verification guarantee. No observations means no score.

Review before publication

New photos remain private until a moderator checks relevance, obvious privacy concerns, duplication, and plausibility. We strip image EXIF metadata but this does not remove faces or text visible in the image. Submitters should avoid those details; moderators reject unsuitable photos.

Public data is context

311 requests describe reported conditions, not proven violations. SweepNYC records indicate a mechanical broom visit, not a guaranteed cleanliness outcome. These signals are shown separately and do not currently enter the score. Unmatched records are not assigned to a block by guesswork.

Corrections and challenges

Use “Think this score is wrong?” on a block page, or upload newer evidence through a survey. Moderators review independently. Paying for cleaning or joining a coalition never changes a score directly or buys favorable moderation.

What the scale means

90–100: exceptional. 80–89: very clean. 70–79: clean. 60–69: acceptable. 50–59: below average. 40–49: dirty. 0–39: filthy. Unsurveyed blocks remain neutral gray.

Sample data and uncertainty

The launch preview is clearly labeled. Sample counts and history are illustrative; no official scores, actual cleaning results, or verified violations are implied. Canonical NYC street geometry is imported separately and starts without a score. Percentiles are withheld until coverage and confidence support a meaningful comparison.

Source documentation

• NYC Street Centerline
• NYC 311 service requests
• DSNY SweepNYC
• OpenStreetMap attribution and license