Scan
Montefiore Medical Center
Score
60/100
17 of 51 tests pass · 40 points still on the table
Adequate
Well ahead of its cohort, with specific gaps rather than structural ones.
Do these three first
The fastest 7 points on this scan
Content Signals declared in robots.txt
No Content-Signal line in robots.txt. Missing: search, ai-input, ai-train.
Do this: Declare a Content Signals position in robots.txt.
FAQPage and QAPage on genuine question-and-answer content
0 pages carry FAQ or QA markup; 1 pages show visible question-and-answer content.
Do this: Mark up the Q&A that exists; remove what does not.
Bylines linked to author entities
0 of 24 substantive pages reference an author by @id; 0 carry a bare string; 24 are unsigned.
Do this: Reference authors by @id rather than repeating a name.
Nine dimensions, most to least left on the table
32 pages crawled
- montefioreeinstein.org
- montefioreeinstein.org/patient-care
- montefioreeinstein.org/patient-care/mychart
- montefioreeinstein.org/patient-care/patients-visitors/billing-insurance/online-bill-pay
- montefioreeinstein.org/patient-care/for-professionals
- montefioreeinstein.org/news
- montefioreeinstein.org/giving
- montefioreeinstein.org/es
- montefioreeinstein.org/research
- montefioreeinstein.org/education
- montefioreeinstein.org/community
- montefioreeinstein.org/patient-care/patients-visitors/managing-your-care/request-medical-records
- montefioreeinstein.org/patient-care/locations
- montefioreeinstein.org/patient-care/patients-visitors/billing-insurance/accepted-insurance-plans
- montefioreeinstein.org/profiles
- montefioreeinstein.org/about/history
- montefioreeinstein.org/supplychain
- montefioreeinstein.org/about/leadership
- montefioreeinstein.org/about/leadership/montefiore-health-system
- montefioreeinstein.org/about/leadership/trustees
- montefioreeinstein.org/404-page-not-found
- montefioreeinstein.org/about/leadership/colleen-m-blye
- montefioreeinstein.org/about/leadership/loreen-babcock
- montefioreeinstein.org/about/leadership/deepesh-chandra
- montefioreeinstein.org/about/leadership/marcos-a-crespo
- montefioreeinstein.org/about/leadership/joseph-a-damico
- montefioreeinstein.org/about/leadership/ruben-diaz-jr
- montefioreeinstein.org/cancer/types/blood-bone-marrow
- montefioreeinstein.org/cancer/types/brain-spine-nerves
- montefioreeinstein.org/cancer/types/unknown-origin
- montefioreeinstein.org/cancer/types/childhood
- montefioreeinstein.org/cancer/types/gastrointestinal
Every test · all 67 tests, in order
67 of 67 showing
AIR-1.1 Can AI crawlers reach your site? PASS 5 pt
All 10 AI agents may fetch content paths.
Your robots.txt file tells crawlers what they may read. A line left behind by a previous site or a copied template can quietly shut out the assistants people now use to find you.
AIR-1.2 Does your CDN let AI crawlers through? PASS 5 pt
All 8 AI user-agents receive full content.
Even when robots.txt says yes, a firewall or bot-protection rule can turn assistants away before they reach your pages. We sent a request as each one and recorded what came back.
AIR-1.3 Is your page readable without a security challenge? PASS 1 pt
No interactive challenges on 32 sampled read-only pages.
A CAPTCHA or bot check on an ordinary page stops an assistant at the door. These belong on forms and logins, not on pages anyone can read.
AIR-1.4 Have you stated what AI may do with your content?No Content-Signal line in robots.txt. Missing: search, ai-input, ai-train. PROBLEM +3
No Content-Signal line in robots.txt. Missing: search, ai-input, ai-train.
There is now a machine-readable way to say whether AI may search, quote or train on your material. Having a position — any position — is the point.
AIR-1.5 Can a person read your AI policy?robots.txt carries a 241-character comment, too thin to read as a policy statement. NEEDS WORK +0
robots.txt carries a 241-character comment, too thin to read as a policy statement.
A few plain sentences in robots.txt saying what you permit. Journalists and lawyers read this file, and right now most sites say nothing.
AIR-1.6 Are you allowing assistants to quote you? PASS 1 pt
No noarchive, nosnippet or max-snippet:0 directives found.
Some sites carry old settings that tell search engines not to show excerpts. Those same settings stop an assistant quoting your page in an answer.
AIR-1.7 Does this page declare its real address? PASS 2 pt
97% of sampled pages emit a valid, self-consistent canonical. 1 have problems.
A canonical link tells a machine which URL is the true one. Without it the same content can be treated as several competing pages.
AIR-1.8 Is your page structured so a machine can follow it?0% of sampled pages carry one main element, a wrapping article or section, and a nav. 32 exceed 12 divs per semantic element. PROBLEM +2
0% of sampled pages carry one main element, a wrapping article or section, and a nav. 32 exceed 12 divs per semantic element.
Meaningful sections rather than an undifferentiated wall of containers. It is the difference between a document and a soup of boxes.
AIR-1.9 Do your headings tell a clear story? PASS 2 pt
94% of sampled pages have exactly one non-empty h1 and no skipped heading levels. 2 have no h1
One main heading, then sub-headings in order. Assistants use headings to work out what a page is about and which part answers a question.
AIR-1.10 Do you publish a guide for AI assistants?/llms.txt lists 66 links across 13 sections, covering 1% of the sitemap. NEEDS WORK +1
/llms.txt lists 66 links across 13 sections, covering 1% of the sitemap.
A file at /llms.txt listing your most important pages. It is early and cheap, and it forces a useful conversation about what actually matters on your site.
AIR-1.11 Can a machine tell what your organization is?142 of 341 entity nodes carry an @id (42%). The WebPage to WebSite to Organization chain is broken at WebSite->Organization. 2 duplicate definitions of WebSite. NEEDS WORK +1
142 of 341 entity nodes carry an @id (42%). The WebPage to WebSite to Organization chain is broken at WebSite->Organization. 2 duplicate definitions of WebSite.
Structured data that links together — your organization, your site, this page — rather than a scattering of disconnected labels.
AIR-1.12 Can a machine be certain it is you?5 sameAs links: 0 tier-1 authority identifiers, 1 tier-2, 4 social. NEEDS WORK +1
5 sameAs links: 0 tier-1 authority identifiers, 1 tier-2, 4 social.
Links from your markup to authoritative records like Wikidata or ROR. Without one, an assistant is guessing which organization of your name it has found.
AIR-1.13 Can a machine find out who you are?0 of 7 organizational facts are present in the Organization markup: none. PROBLEM +2
0 of 7 organizational facts are present in the Organization markup: none.
Founding date, leadership, address, contact details, legal identifiers. Thin About pages are the most common weakness we see.
AIR-2.1 Primary content is present without JavaScript PASS 6 pt
Raw HTML contains 100% of rendered text on average across 32 URLs. The worst template is /es: 100% across 1 pages.
AIR-2.4 Stable, human-readable anchor IDs on section headings0 of 630 subheadings carry a slug-like id (0%); 105 have generated or unstable ids. NEEDS WORK +1
0 of 630 subheadings carry a slug-like id (0%); 105 have generated or unstable ids.
AIR-2.5 Alt text on content images, empty alt on decorative351 of 531 images carry appropriate alt text: 2 have no alt attribute, 11 use the filename. NEEDS WORK +0
351 of 531 images carry appropriate alt text: 2 have no alt attribute, 11 use the filename.
AIR-2.6 Key facts exist as HTML text, not only in images or PDFs19 of 32 sampled pages are thin relative to their images or lead with a document download. PROBLEM +1
19 of 32 sampled pages are thin relative to their images or lead with a document download.
AIR-3.1 JSON-LD is generated from mapped fields, not hardcoded PASS 2 pt
Schema values track visible content across 3 templates.
AIR-3.2 BreadcrumbList on every page below the homepage PASS 2 pt
97% of pages below the homepage carry a BreadcrumbList with at least two steps. 1 have a single-item trail.
AIR-3.3 Vertical-specific schema types2 of 4 expected healthcare types are present: MedicalOrganization, MedicalCondition. Missing: Physician, MedicalWebPage. NEEDS WORK +2
2 of 4 expected healthcare types are present: MedicalOrganization, MedicalCondition. Missing: Physician, MedicalWebPage.
AIR-3.4 FAQPage and QAPage on genuine question-and-answer content0 pages carry FAQ or QA markup; 1 pages show visible question-and-answer content. PROBLEM +2
0 pages carry FAQ or QA markup; 1 pages show visible question-and-answer content.
AIR-3.5 isAccessibleForFree, about, and mentions with entity references0 of 24 substantive pages declare isAccessibleForFree; 0 reference subject entities by identifier. PROBLEM +1
0 of 24 substantive pages declare isAccessibleForFree; 0 reference subject entities by identifier.
AIR-3.7 SearchAction on the WebSite nodeThe SearchAction template returns 20 characters of results for a test term. NEEDS WORK +0
The SearchAction template returns 20 characters of results for a test term.
AIR-3.8 Schema validation of the structured data on the pageEverything parses, but nodes on 32 of 32 pages are missing properties their declared type needs. NEEDS WORK +1
Everything parses, but nodes on 32 of 32 pages are missing properties their declared type needs.
AIR-4.4 Correct X-Robots-Tag on non-HTML resources0 of 25 non-HTML resources carry an X-Robots-Tag. PROBLEM +1
AIR-4.5 Missing pages return 404 or 4101 of 10 deliberately invalid URLs returned HTTP 200; 7 sampled pages are thin enough to be not-found templates. PROBLEM +1
1 of 10 deliberately invalid URLs returned HTTP 200; 7 sampled pages are thin enough to be not-found templates.
AIR-4.6 Redirects resolve in a single hop PASS 1 pt
0 of 32 sampled URLs redirect.
AIR-4.7 Faceted, calendar, and parameterized URLs are controlled PASS 2 pt
No parameterized URLs appeared in the crawlable sample.
AIR-4.8 Sitemap lastmod reflects real content changes PASS 2 pt
3760 of 4619 sitemap URLs carry a lastmod across 406 distinct dates; 6% share 2025-07-07.
AIR-4.9 Sitemap index split by type, with media sitemaps PASS 1 pt
An index with 2 child sitemaps covering 4619 URLs, with no image or video sitemap.
AIR-5.1 Author entity pages with credentialsNo Person entities appear anywhere in the sample. PROBLEM +2
AIR-5.2 Bylines linked to author entities0 of 24 substantive pages reference an author by @id; 0 carry a bare string; 24 are unsigned. PROBLEM +2
0 of 24 substantive pages reference an author by @id; 0 carry a bare string; 24 are unsigned.
AIR-5.4 datePublished and dateModified are accurate PASS 2 pt
32 of 32 pages carry a publication or modification date across 21 distinct dateModified values.
AIR-6.3 Form fields carry label, name, and autocomplete PASS 2 pt
32 of 33 form fields carry an accessible name, a meaningful name attribute and a valid autocomplete token where one applies.
AIR-6.4 No captchas on browse, search, or filter interactions PASS 2 pt
0 of 32 search or filter surfaces return a challenge.
AIR-7.1 Markdown companion for every canonical page0 of 24 sampled pages have a Markdown companion at {path}.md returning a Markdown content type. PROBLEM +2
0 of 24 sampled pages have a Markdown companion at {path}.md returning a Markdown content type.
AIR-7.3 Accept: text/markdown content negotiation0 of 25 URLs serve Markdown for Accept: text/markdown. PROBLEM +2
AIR-7.4 Alternate representations are edge-cached PASS 2 pt
1 of 1 generated representations show evidence of an edge cache hit on a repeat request.
LI-1 HowTo on procedural content0 of 2 procedural pages carry HowTo with ordered HowToStep. NOT SCORED —
0 of 2 procedural pages carry HowTo with ordered HowToStep.
LI-2 speakable on summaries and ledesNo speakable specification found on any sampled page. NOT SCORED —
LI-3 IndexNow fires on publish and updateNo IndexNow key file was discoverable. The file is named after the key, which is not published, so this is 'not discoverable' rather than proof of absence. NOT SCORED —
No IndexNow key file was discoverable. The file is named after the key, which is not published, so this is 'not discoverable' rather than proof of absence.
LI-4 Editorial policy page referenced via publishingPrinciples0 of 96 Organization nodes reference publishingPrinciples. NOT SCORED —
0 of 96 Organization nodes reference publishingPrinciples.
LI-5 NLWeb endpoint exposed as an MCP server with an ask methodNo NLWeb endpoint was discoverable. NOT SCORED —
LI-6 Domain MCP server over real content APIsNo domain MCP server or tool descriptor was discoverable. NOT SCORED —
BP-3 GA4 channel group for AI assistant referrersWhether a channel group exists for AI assistant referrers is not visible from outside. Analytics presence is; the grouping is not. NOT SCORED —
Whether a channel group exists for AI assistant referrers is not visible from outside. Analytics presence is; the grouping is not.
BP-4 Server-side tagging captures stripped referrersReconciliation between server-side and client-side numbers has to be attested. NOT SCORED —
Reconciliation between server-side and client-side numbers has to be attested.
BP-5 Access log retention with a queryable storeLog retention and queryability are not observable from outside the site. NOT SCORED —
Log retention and queryability are not observable from outside the site.
BP-7 Fixed prompt panel run monthly across modelsAnswer-share tracking happens outside the site and must be attested. NOT SCORED —
Answer-share tracking happens outside the site and must be attested.
BP-8 Extraction-fidelity baseline capturedNo extraction-fidelity baseline is captured by this version of the scanner, which makes no live model calls. NOT SCORED —
No extraction-fidelity baseline is captured by this version of the scanner, which makes no live model calls.
BP-9 Search Console and Bing Webmaster Tools verifiedNeither Search Console nor Bing Webmaster Tools verification was found. NOT SCORED —
Neither Search Console nor Bing Webmaster Tools verification was found.
Not applicable — 22 tests answered with a dash, excluded from the score: AIR-2.2, AIR-2.3, AIR-3.6, AIR-4.1, AIR-4.2, AIR-4.3, AIR-4.10, AIR-5.3, AIR-5.5, AIR-5.6, AIR-6.1, AIR-6.2, AIR-7.2, LI-7, BP-1, BP-2, BP-3, BP-4, BP-5, BP-6, BP-7, BP-8.
By dimension
Where Montefiore Medical Center's 60 came from. Open a line to see its tests, worst first.
1 · Primary Indicators 19 of 30 19 63
AIR-1.4 Have you stated what AI may do with your content?No Content-Signal line in robots.txt. Missing: search, ai-input, ai-train. PROBLEM
No Content-Signal line in robots.txt. Missing: search, ai-input, ai-train.
There is now a machine-readable way to say whether AI may search, quote or train on your material. Having a position — any position — is the point.
AIR-1.8 Is your page structured so a machine can follow it?0% of sampled pages carry one main element, a wrapping article or section, and a nav. 32 exceed 12 divs per semantic element. PROBLEM
0% of sampled pages carry one main element, a wrapping article or section, and a nav. 32 exceed 12 divs per semantic element.
Meaningful sections rather than an undifferentiated wall of containers. It is the difference between a document and a soup of boxes.
AIR-1.13 Can a machine find out who you are?0 of 7 organizational facts are present in the Organization markup: none. PROBLEM
0 of 7 organizational facts are present in the Organization markup: none.
Founding date, leadership, address, contact details, legal identifiers. Thin About pages are the most common weakness we see.
AIR-1.5 Can a person read your AI policy?robots.txt carries a 241-character comment, too thin to read as a policy statement. NEEDS WORK
robots.txt carries a 241-character comment, too thin to read as a policy statement.
A few plain sentences in robots.txt saying what you permit. Journalists and lawyers read this file, and right now most sites say nothing.
AIR-1.10 Do you publish a guide for AI assistants?/llms.txt lists 66 links across 13 sections, covering 1% of the sitemap. NEEDS WORK
/llms.txt lists 66 links across 13 sections, covering 1% of the sitemap.
A file at /llms.txt listing your most important pages. It is early and cheap, and it forces a useful conversation about what actually matters on your site.
AIR-1.11 Can a machine tell what your organization is?142 of 341 entity nodes carry an @id (42%). The WebPage to WebSite to Organization chain is broken at WebSite->Organization. 2 duplicate definitions of WebSite. NEEDS WORK
142 of 341 entity nodes carry an @id (42%). The WebPage to WebSite to Organization chain is broken at WebSite->Organization. 2 duplicate definitions of WebSite.
Structured data that links together — your organization, your site, this page — rather than a scattering of disconnected labels.
AIR-1.12 Can a machine be certain it is you?5 sameAs links: 0 tier-1 authority identifiers, 1 tier-2, 4 social. NEEDS WORK
5 sameAs links: 0 tier-1 authority identifiers, 1 tier-2, 4 social.
Links from your markup to authoritative records like Wikidata or ROR. Without one, an assistant is guessing which organization of your name it has found.
Full marks — 6 pass
AIR-1.1 Can AI crawlers reach your site? PASS
All 10 AI agents may fetch content paths.
Your robots.txt file tells crawlers what they may read. A line left behind by a previous site or a copied template can quietly shut out the assistants people now use to find you.
AIR-1.2 Does your CDN let AI crawlers through? PASS
All 8 AI user-agents receive full content.
Even when robots.txt says yes, a firewall or bot-protection rule can turn assistants away before they reach your pages. We sent a request as each one and recorded what came back.
AIR-1.3 Is your page readable without a security challenge? PASS
No interactive challenges on 32 sampled read-only pages.
A CAPTCHA or bot check on an ordinary page stops an assistant at the door. These belong on forms and logins, not on pages anyone can read.
AIR-1.6 Are you allowing assistants to quote you? PASS
No noarchive, nosnippet or max-snippet:0 directives found.
Some sites carry old settings that tell search engines not to show excerpts. Those same settings stop an assistant quoting your page in an answer.
AIR-1.7 Does this page declare its real address? PASS
97% of sampled pages emit a valid, self-consistent canonical. 1 have problems.
A canonical link tells a machine which URL is the true one. Without it the same content can be treated as several competing pages.
AIR-1.9 Do your headings tell a clear story? PASS
94% of sampled pages have exactly one non-empty h1 and no skipped heading levels. 2 have no h1
One main heading, then sub-headings in order. Assistants use headings to work out what a page is about and which part answers a question.
2 · Rendering and Extraction 8 of 15 8 53
AIR-2.6 Key facts exist as HTML text, not only in images or PDFs19 of 32 sampled pages are thin relative to their images or lead with a document download. PROBLEM
19 of 32 sampled pages are thin relative to their images or lead with a document download.
AIR-2.4 Stable, human-readable anchor IDs on section headings0 of 630 subheadings carry a slug-like id (0%); 105 have generated or unstable ids. NEEDS WORK
0 of 630 subheadings carry a slug-like id (0%); 105 have generated or unstable ids.
AIR-2.5 Alt text on content images, empty alt on decorative351 of 531 images carry appropriate alt text: 2 have no alt attribute, 11 use the filename. NEEDS WORK
351 of 531 images carry appropriate alt text: 2 have no alt attribute, 11 use the filename.
Full marks — 1 pass
AIR-2.1 Primary content is present without JavaScript PASS
Raw HTML contains 100% of rendered text on average across 32 URLs. The worst template is /es: 100% across 1 pages.
3 · Structured Data and the Entity Graph 7 of 15 7 47
AIR-3.4 FAQPage and QAPage on genuine question-and-answer content0 pages carry FAQ or QA markup; 1 pages show visible question-and-answer content. PROBLEM
0 pages carry FAQ or QA markup; 1 pages show visible question-and-answer content.
AIR-3.5 isAccessibleForFree, about, and mentions with entity references0 of 24 substantive pages declare isAccessibleForFree; 0 reference subject entities by identifier. PROBLEM
0 of 24 substantive pages declare isAccessibleForFree; 0 reference subject entities by identifier.
AIR-3.3 Vertical-specific schema types2 of 4 expected healthcare types are present: MedicalOrganization, MedicalCondition. Missing: Physician, MedicalWebPage. NEEDS WORK
2 of 4 expected healthcare types are present: MedicalOrganization, MedicalCondition. Missing: Physician, MedicalWebPage.
AIR-3.7 SearchAction on the WebSite nodeThe SearchAction template returns 20 characters of results for a test term. NEEDS WORK
The SearchAction template returns 20 characters of results for a test term.
AIR-3.8 Schema validation of the structured data on the pageEverything parses, but nodes on 32 of 32 pages are missing properties their declared type needs. NEEDS WORK
Everything parses, but nodes on 32 of 32 pages are missing properties their declared type needs.
Full marks — 2 pass
AIR-3.1 JSON-LD is generated from mapped fields, not hardcoded PASS
Schema values track visible content across 3 templates.
AIR-3.2 BreadcrumbList on every page below the homepage PASS
97% of pages below the homepage carry a BreadcrumbList with at least two steps. 1 have a single-item trail.
4 · Crawl, Index and Licensing Hygiene 6 of 12 6 50
AIR-4.4 Correct X-Robots-Tag on non-HTML resources0 of 25 non-HTML resources carry an X-Robots-Tag. PROBLEM
AIR-4.5 Missing pages return 404 or 4101 of 10 deliberately invalid URLs returned HTTP 200; 7 sampled pages are thin enough to be not-found templates. PROBLEM
1 of 10 deliberately invalid URLs returned HTTP 200; 7 sampled pages are thin enough to be not-found templates.
Full marks — 4 pass
AIR-4.6 Redirects resolve in a single hop PASS
0 of 32 sampled URLs redirect.
AIR-4.7 Faceted, calendar, and parameterized URLs are controlled PASS
No parameterized URLs appeared in the crawlable sample.
AIR-4.8 Sitemap lastmod reflects real content changes PASS
3760 of 4619 sitemap URLs carry a lastmod across 406 distinct dates; 6% share 2025-07-07.
AIR-4.9 Sitemap index split by type, with media sitemaps PASS
An index with 2 child sitemaps covering 4619 URLs, with no image or video sitemap.
5 · Authorship, Provenance, Freshness 2 of 12 2 17
AIR-5.1 Author entity pages with credentialsNo Person entities appear anywhere in the sample. PROBLEM
AIR-5.2 Bylines linked to author entities0 of 24 substantive pages reference an author by @id; 0 carry a bare string; 24 are unsigned. PROBLEM
0 of 24 substantive pages reference an author by @id; 0 carry a bare string; 24 are unsigned.
Full marks — 1 pass
AIR-5.4 datePublished and dateModified are accurate PASS
32 of 32 pages carry a publication or modification date across 21 distinct dateModified values.
6 · Agent Interfaces 4 of 8 4 50
Full marks — 2 pass
AIR-6.3 Form fields carry label, name, and autocomplete PASS
32 of 33 form fields carry an accessible name, a meaningful name attribute and a valid autocomplete token where one applies.
AIR-6.4 No captchas on browse, search, or filter interactions PASS
0 of 32 search or filter surfaces return a challenge.
7 · Alternate Representations 2 of 8 2 25
AIR-7.1 Markdown companion for every canonical page0 of 24 sampled pages have a Markdown companion at {path}.md returning a Markdown content type. PROBLEM
0 of 24 sampled pages have a Markdown companion at {path}.md returning a Markdown content type.
AIR-7.3 Accept: text/markdown content negotiation0 of 25 URLs serve Markdown for Accept: text/markdown. PROBLEM
Full marks — 1 pass
AIR-7.4 Alternate representations are edge-cached PASS
1 of 1 generated representations show evidence of an edge cache hit on a repeat request.
What was read
32 pages, sampled from this site's own sitemap and internal links. The cap is 32 — a ceiling, not a certainty: a site with fewer reachable pages yields fewer.
32 other pages, all 200, by template
The receipt
- Cohort
- health-systems
- Source
- IRS Exempt Organizations Business Master File, filtered to health-care organizations by Wikidata classification (Internal Revenue Service (IRS), with Wikidata, 2026-09-07)
- Index
- 11.0, in effect 2026-09-24
- Scanner
- 0.1.0
- Scanned
- 2026-09-20T00:49:04.038389+00:00
- Result hash
- 06cc8d656becc214d68819e44298ee65d2ce37fb313d25d15d1d87f2c9d981e0
- Basis
- public-interest census: a declared bot with a published policy at https://readinessindex.io/bot, honoring robots.txt addressed to AIRBot, reading public pages only. No per-site permission was sought or given.
The hash is the SHA-256 of the full
result.json, recorded when the scan finished rather than
computed now. The raw evidence behind every line above — every
response, cached — is retained and is a median 1.5 MB per site, so
it is not served from here. Ask
us for it if you want to check a number against the bytes it came
from.
Think this is wrong? It might be. Tell us at hello@readinessindex.io and we will look, correct it if it is, and say so publicly either way.
Is your site ready for AI?
Scan a page, get a number you can verify. Free.
About these numbers. Points are whole numbers everywhere on this page. The arithmetic behind them is not — a test scores its weight times its band divided by four, so a row can earn 1.5 points and print as 2. Scores are computed from the exact figures and rounded once, at the end, never from the rounded rows. That means adding the points column up will not give you the score exactly. The rule is published in the standard.