FROM THE BIC TOOLKITSmall useful things built in response to a slightly broken internet.

ASSESS / ARTEFACT 005 / EXPERIMENT A

Good
Internet
Score

How much nonsense stands between a person and the thing they came here to do?

Ten services. Ten tests. One hundred points.

See the first ten scores

WHY BIC BUILT THIS

The internet has plenty of ratings for speed, SEO and customer satisfaction. It has fewer for being decent to use.

The Good Internet Score rates how humane, useful, independent and user-respecting a website or digital service is.

It is not a score for visual taste, business success or whether BIC likes the brand. It rewards experiences that work, respect people and let them retain some control.

EXPERIMENT A / LAST TESTED 9 AUGUST 2026

Ten deliberately different services.

An independent shop, supermarket, parking service, streaming platform, bank, airline, smart-home product, public service, major platform and small independent website. These remain provisional editorial assessments: the experiment tests the rubric as much as the services.

Parking serviceCONFIDENCE VARIES BY TEST
51/100

RingGo

Mild Hostage Situation

The app is optional, which saves RingGo from an App for No Reason offence. The web route still turns paying for an ordinary parking space into an account relationship, while the cookie banner offered acceptance without an equally visible refusal. Functional, but distinctly more complicated than putting money in a machine.

Read the full profile
Independent ecommerceCONFIDENCE VARIES BY TEST
86/100

Hodmedod's

Mostly Civilised

Mostly Good Internet. The shop lets people browse detailed product information, add an item and reach payment without creating an account. It also provides the rare modern luxury of an email address, telephone number and postal address, although the pre-ticked marketing box needs putting back in its box.

Read the full profile
Large subscription platformCONFIDENCE VARIES BY TEST
88/100

Netflix

Mostly Civilised

A surprisingly civilised result for a very large platform. Prices, browser access, contact routes and two-click cancellation are unusually explicit, and an account is justified by the paid personalised service. The score also reveals a calibration problem: the current rubric may not fully express the wider dependency and data power of a platform this large.

Read the full profile
Major supermarketCONFIDENCE VARIES BY TEST
63/100

Tesco Groceries

Getting a Bit Platformy

Useful, broad and perfectly capable on the web, but online groceries begin with an account and now require a linked Clubcard. The app is optional and human help is unusually visible; the compulsory relationship starts before the first banana reaches the basket.

Read the full profile
BankCONFIDENCE VARIES BY TEST
80/100

Barclays

Mostly Civilised

A complicated institution with a surprisingly wide set of exits from the digital maze: website, Online Banking, telephone, branch and accessible contact routes. The account is justified; the score is for this support task, not a declaration that every banking journey is delightful.

Read the full profile
Airline and travelCONFIDENCE VARIES BY TEST
75/100

British Airways

Mostly Civilised

The useful bits are all here and an account is not required to research or manage a booking. The experience carries the usual travel-site thicket of fares, extras and scripts, but BA also publishes unusually concrete phone, address and accessibility routes.

Read the full profile
Smart-home productCONFIDENCE VARIES BY TEST
51/100

Hive

Mild Hostage Situation

The thermostat still has physical controls and subscription cancellation does not switch the heating off. But much of the 'smart' value depends on an account, app, hub and remote service: your radiator has acquired a small technology stack.

Read the full profile
Public serviceCONFIDENCE VARIES BY TEST
96/100

GOV.UK

Good Internet

Good Internet. The task begins immediately, asks only for what it needs and ends with an answer rather than a new relationship. There are telephone, post, relay and BSL alternatives too: public infrastructure behaving like public infrastructure.

Read the full profile
Major platformCONFIDENCE VARIES BY TEST
70/100

YouTube

Getting a Bit Platformy

Enormously useful internet living inside an enormously powerful platform. Watching remains possible without an account or app, but recommendation, advertising and creator dependency make the front door far less revealing than the building behind it.

Read the full profile
Small independent websiteCONFIDENCE VARIES BY TEST
95/100

Marginalia Search

Good Internet

Good Internet in workshop clothes. Search works without an account, the index tries to surface the web beyond commercial rankings, the code is open and the privacy explanation is unusually specific. It is less polished than a giant search engine and considerably more honest about what it is doing.

Read the full profile

WHAT EXPERIMENT A EXPOSED

The awkward results are the useful results.

01

Necessary accounts need clearer boundaries.

Netflix and Barclays legitimately need accounts for paid or private services. That exemption can still hide the dependency and data power created after sign-up.

02

Privacy is hard to score from the pavement.

Observable consent design and published policy are evidence. The complete data picture rarely is. Privacy therefore carries more uncertainty than a visible login gate.

03

Different tasks produce different truths.

Barclays scores well for finding urgent help; that does not mean every authenticated banking task deserves 80. Each result belongs to its stated test.

04

Small is not automatically virtuous.

Marginalia scores highly because it is open, direct and specific about privacy—not because independent projects receive a complimentary ten points and a tote bag.

ASSESSMENT 001 / PARKING SERVICE

RingGo

Test task: Pay for two hours of parking using the web, without installing an app.

51/100
BEST BIT

A real web alternative exists alongside the app, with downloadable receipts and support information.

MAIN OFFENCE

Starting a basic web parking session redirects to login or registration, and the visible cookie banner only offered Accept.

ACCOUNT HOSTAGECOOKIE OBSTACLE COURSENO APP REQUIRED
01 / Just Let Me Use It3/10

Can I accomplish the core task without being forced through unnecessary barriers?

Observation: The prominent Park Now route redirected a signed-out visitor to login. Registration is the next available route before a web parking session can begin.

CONFIDENCE: HIGHRingGo login
02 / No-App Test8/10

Does this genuinely need an app?

Observation: RingGo strongly promotes its app, but a personal web portal and automated telephone route exist. The app is not the only interface.

CONFIDENCE: HIGHHow RingGo works
03 / No-Account Test2/10

Do I genuinely need an account for this?

Observation: An account is required to start the ordinary online parking flow. The task is simple, but registration is the gate.

CONFIDENCE: HIGHRingGo home
04 / Exit Is a Feature6/10

Can I leave?

Observation: Receipts can be downloaded and older history can be requested, but only three years is visible directly and some retrieval depends on Customer Care.

CONFIDENCE: MEDIUMParking history policy
05 / Dark Pattern Test4/10

Is the interface trying to manipulate me?

Observation: The cookie message said continued use meant agreement and exposed an Accept button, with no equivalent reject control in the visible banner.

CONFIDENCE: HIGHRingGo cookie policy
06 / Privacy & Proportionality4/10

Does the service collect roughly what it needs, rather than everything it can?

Observation: Parking requires account, contact, vehicle, location and payment data. Some collection is necessary, but the consent presentation and breadth of the service reduce proportionality.

CONFIDENCE: MEDIUMRingGo privacy notice
07 / Access for Humans6/10

Can a reasonably broad range of people actually use it?

Observation: The tested pages used headings, labels and a published accessibility route, but account dependence and the one-sided cookie control remain meaningful barriers. This was not a WCAG audit.

CONFIDENCE: MEDIUMRingGo accessibility
08 / Does the Bloody Thing Work?7/10

Is it fast, reliable and technically competent enough to use?

Observation: Public pages loaded and the route to parking behaved consistently. The final paid session was not completed because that would require an account, location and payment.

CONFIDENCE: MEDIUMRingGo parking portal
09 / Can I Find a Human?6/10

Can I understand who is responsible and contact somebody when things go wrong?

Observation: Support articles, live chat references and request forms provide a route to Customer Care, although escalation remains mediated through the support system.

CONFIDENCE: MEDIUMRingGo support
10 / Worth Visiting5/10

Does this actually contribute something useful to the web?

Observation: The portal performs a useful transaction and provides receipts and help, but offers limited value beyond the parking system itself.

CONFIDENCE: HIGHRingGo services

ASSESSMENT 002 / INDEPENDENT ECOMMERCE

Hodmedod's

Test task: Buy a 500g bag of British chickpeas without creating an account or joining a mailing list.

86/100
BEST BIT

Guest checkout, clear human contact and genuinely useful product, farming, cooking and recipe information.

MAIN OFFENCE

The checkout pre-selected 'Email me with news and offers' rather than waiting to be invited.

JUST WORKSHUMAN FOUNDUSEFUL INTERNETNO APP REQUIRED
01 / Just Let Me Use It9/10

Can I accomplish the core task without being forced through unnecessary barriers?

Observation: A product could be found, added to the basket and taken to payment directly. The only notable interruption was optional marketing consent.

CONFIDENCE: HIGHBritish chickpeas
02 / No-App Test10/10

Does this genuinely need an app?

Observation: The complete shopping journey works on the website. No app is requested or required.

CONFIDENCE: HIGHHodmedod's shop
03 / No-Account Test10/10

Do I genuinely need an account for this?

Observation: Checkout displayed Sign in as an option while leaving the email, delivery and payment fields available to a guest.

CONFIDENCE: HIGHShopping basket
04 / Exit Is a Feature8/10

Can I leave?

Observation: Returns are documented, customers can contact the company directly and an account is not needed for a one-off purchase. Some portability criteria are not relevant to a bag of chickpeas.

CONFIDENCE: MEDIUMReturns policy
05 / Dark Pattern Test7/10

Is the interface trying to manipulate me?

Observation: The main choices were clear, but the checkout arrived with marketing email consent already checked. Express payment options were optional rather than obstructive.

CONFIDENCE: HIGHPrivacy policy
06 / Privacy & Proportionality7/10

Does the service collect roughly what it needs, rather than everything it can?

Observation: The store explains its privacy practices and asks for information expected for delivery, but the pre-selected marketing consent and ecommerce tracking prevent a higher score.

CONFIDENCE: MEDIUMPrivacy policy
07 / Access for Humans7/10

Can a reasonably broad range of people actually use it?

Observation: The checkout provided a skip link, labelled form controls and a clear heading structure. Several shop controls used icon fonts or weak accessible names. This was not a WCAG audit.

CONFIDENCE: MEDIUMHodmedod's home
08 / Does the Bloody Thing Work?8/10

Is it fast, reliable and technically competent enough to use?

Observation: Product, basket and hosted checkout pages loaded reliably and retained the selected item without errors during the test.

CONFIDENCE: HIGHBritish chickpeas
09 / Can I Find a Human?10/10

Can I understand who is responsible and contact somebody when things go wrong?

Observation: The site publishes a named company, company number, postal address, email address, telephone number and contact page.

CONFIDENCE: HIGHContact Hodmedod
10 / Worth Visiting10/10

Does this actually contribute something useful to the web?

Observation: Detailed provenance, cooking instructions, recipes and farming stories make the website useful beyond extracting a transaction.

CONFIDENCE: HIGHHodmedod's recipes

ASSESSMENT 003 / LARGE SUBSCRIPTION PLATFORM

Netflix

Test task: Find the UK monthly price, confirm where the service works and understand how to cancel.

88/100
BEST BIT

Cancellation is explained plainly, completed online and carries no cancellation fee.

MAIN OFFENCE

Personalisation, advertising and cross-device account data make the privacy bargain much larger than the tidy front door suggests.

EXIT IS A FEATUREHUMAN FOUNDNO APP REQUIRED
01 / Just Let Me Use It8/10

Can I accomplish the core task without being forced through unnecessary barriers?

Observation: The UK landing page states the starting price, plan range, device coverage and cancellation terms before signup. Joining still requires a multi-step paid account flow.

CONFIDENCE: HIGHNetflix UK
02 / No-App Test10/10

Does this genuinely need an app?

Observation: Netflix documents full web-browser viewing on supported computers and tablets. Mobile and TV apps exist but are not the only way to use the service.

CONFIDENCE: HIGHSupported browsers
03 / No-Account Test10/10

Do I genuinely need an account for this?

Observation: An account is necessary for a paid, personalised streaming subscription. The rubric explicitly says not to punish legitimate saved subscriptions.

CONFIDENCE: HIGHGetting started
04 / Exit Is a Feature10/10

Can I leave?

Observation: Netflix says cancellation takes two clicks online, has no cancellation fee and can be done at any time. Viewing history can be downloaded in a standard CSV format.

05 / Dark Pattern Test9/10

Is the interface trying to manipulate me?

Observation: Pricing and cancellation claims were prominent. The cookie dialog gave Accept and Reject equal first-level buttons, with a separate option to personalise choices.

CONFIDENCE: HIGHNetflix UK
06 / Privacy & Proportionality6/10

Does the service collect roughly what it needs, rather than everything it can?

Observation: Netflix explains controls and offers a first-level cookie refusal, but its service relies on extensive account, viewing, device, personalisation and advertising data.

CONFIDENCE: MEDIUMPrivacy choices
07 / Access for Humans9/10

Can a reasonably broad range of people actually use it?

Observation: Netflix publishes support for captions, audio description and mobility, sight and hearing needs, plus a dedicated accessibility feedback route. This was not a WCAG audit.

CONFIDENCE: MEDIUMNetflix accessibility
08 / Does the Bloody Thing Work?8/10

Is it fast, reliable and technically competent enough to use?

Observation: The public web experience was stable and Netflix supports many browsers and devices, although minimum versions and platform-specific resolution limits apply.

CONFIDENCE: MEDIUMBrowser requirements
09 / Can I Find a Human?9/10

Can I understand who is responsible and contact somebody when things go wrong?

Observation: The public site links to Contact Us and Netflix documents call and chat routes for members and non-members.

10 / Worth Visiting9/10

Does this actually contribute something useful to the web?

Observation: The core service is substantial and intentionally revisited, while the public catalogue and help centre provide useful information outside the signed-in product.

CONFIDENCE: HIGHWhat is Netflix?

ASSESSMENT 004 / MAJOR SUPERMARKET

Tesco Groceries

Test task: Build a basic online grocery basket and understand how to check out without installing the app.

63/100
BEST BIT

The grocery service remains available through an ordinary website, backed by long-hours telephone and WhatsApp support.

MAIN OFFENCE

A Tesco account and linked Clubcard are required for an online grocery transaction.

ACCOUNT HOSTAGENO APP REQUIREDHUMAN FOUND
01 / Just Let Me Use It5/10

Can I accomplish the core task without being forced through unnecessary barriers?

Observation: Browsing is open, but Tesco's own shopping guide begins with sign-in and slot booking before checkout.

02 / No-App Test10/10

Does this genuinely need an app?

Observation: The full grocery service remains available on the website; the app is not required for the defined shopping task.

03 / No-Account Test2/10

Do I genuinely need an account for this?

Observation: Tesco's guide requires sign-in and its terms say online grocery transactions require a Tesco.com account linked to a valid Clubcard.

CONFIDENCE: HIGHTesco grocery terms
04 / Exit Is a Feature6/10

Can I leave?

Observation: Orders and personal details can be managed online, but opting out of Clubcard requires contacting the Clubcard team.

CONFIDENCE: MEDIUMOpting out of Clubcard
05 / Dark Pattern Test6/10

Is the interface trying to manipulate me?

Observation: Cookie preferences are offered and the basket is conventional, but mandatory Clubcard linkage turns a loyalty identity into part of ordinary checkout.

06 / Privacy & Proportionality5/10

Does the service collect roughly what it needs, rather than everything it can?

Observation: Delivery needs address and payment data, but the linked retail and loyalty account produces a broader data relationship than the transaction alone.

07 / Access for Humans7/10

Can a reasonably broad range of people actually use it?

Observation: The public pages use clear structure and publish accessibility links, but the full basket and checkout were not audited with assistive technology.

08 / Does the Bloody Thing Work?7/10

Is it fast, reliable and technically competent enough to use?

Observation: Browsing, product pages and help loaded reliably during the review. No paid order was placed, so fulfilment and final payment were outside the test.

CONFIDENCE: MEDIUMTesco groceries
09 / Can I Find a Human?9/10

Can I understand who is responsible and contact somebody when things go wrong?

Observation: Tesco publishes telephone and WhatsApp routes with extended opening hours and explicitly says when a person is available.

CONFIDENCE: HIGHContact Tesco
10 / Worth Visiting6/10

Does this actually contribute something useful to the web?

Observation: The catalogue and service are useful, although most of the public experience is designed around a transaction and loyalty system.

CONFIDENCE: HIGHTesco groceries

ASSESSMENT 005 / BANK

Barclays

Test task: Find how an existing personal customer can report a lost card and reach human help without relying on the mobile app.

80/100
BEST BIT

Important help remains available through Online Banking, telephone and branches rather than being app-only.

MAIN OFFENCE

Some support journeys steer customers towards secure app messaging before revealing other routes.

NO APP REQUIREDHUMAN FOUND
01 / Just Let Me Use It8/10

Can I accomplish the core task without being forced through unnecessary barriers?

Observation: Help content exposes direct routes for urgent and routine customer problems, although navigation spans many product-specific pages.

CONFIDENCE: MEDIUMBarclays help
02 / No-App Test9/10

Does this genuinely need an app?

Observation: The app is promoted, but Barclays also provides Online Banking, telephone banking and branches for core account support.

CONFIDENCE: HIGHWays to bank
03 / No-Account Test10/10

Do I genuinely need an account for this?

Observation: Identity and authentication are legitimate requirements for access to private bank records; public emergency contact information is available before sign-in.

04 / Exit Is a Feature6/10

Can I leave?

Observation: Customers can move money and switch accounts through regulated processes, but full exit, records and product closure vary by product and were not completed.

CONFIDENCE: MEDIUMSwitching bank account
05 / Dark Pattern Test8/10

Is the interface trying to manipulate me?

Observation: The tested help pages presented routes plainly. App-first wording adds persuasion, but no material deception was observed in the defined task.

06 / Privacy & Proportionality6/10

Does the service collect roughly what it needs, rather than everything it can?

Observation: Banking necessarily involves extensive identity and financial data. Barclays publishes controls and explanations, but this was not a privacy audit.

CONFIDENCE: MEDIUMControl your data
07 / Access for Humans9/10

Can a reasonably broad range of people actually use it?

Observation: Barclays states that its app and Online Banking target WCAG 2.2 AA and publishes tailored support for different access needs.

08 / Does the Bloody Thing Work?8/10

Is it fast, reliable and technically competent enough to use?

Observation: Public help and contact pages worked in the review; authenticated banking reliability was not independently tested.

09 / Can I Find a Human?9/10

Can I understand who is responsible and contact somebody when things go wrong?

Observation: Telephone numbers, branches and accessibility routes are published, including alternatives to app chat.

10 / Worth Visiting7/10

Does this actually contribute something useful to the web?

Observation: The help centre provides substantial practical information, although much of the wider public site is product marketing.

CONFIDENCE: HIGHBarclays help

ASSESSMENT 006 / AIRLINE AND TRAVEL

British Airways

Test task: Find a return flight, understand the fare and locate cancellation and accessibility help without creating an account or buying a ticket.

75/100
BEST BIT

Manage My Booking works with a booking reference and surname, without requiring loyalty membership.

MAIN OFFENCE

Comparing the true final cost still means navigating branded fare families, seat choices and optional extras.

NO APP REQUIREDHUMAN FOUND
01 / Just Let Me Use It7/10

Can I accomplish the core task without being forced through unnecessary barriers?

Observation: Flight search is available immediately, but a realistic fare comparison involves several fare and extras stages.

CONFIDENCE: MEDIUMBook flights
02 / No-App Test10/10

Does this genuinely need an app?

Observation: Booking, check-in, boarding-pass printing and booking management are available on the web; the mobile app is optional.

CONFIDENCE: HIGHManage My Booking
03 / No-Account Test10/10

Do I genuinely need an account for this?

Observation: Research and booking do not require a British Airways Club account; an existing booking can be opened with its reference and passenger surname.

04 / Exit Is a Feature6/10

Can I leave?

Observation: Manage My Booking exposes changes and refund routes, but fare conditions determine cost and some agent bookings must return to the agent.

05 / Dark Pattern Test6/10

Is the interface trying to manipulate me?

Observation: Core prices and fare choices are presented, but the layered extras and branded bundles make comparison more work than the headline fare suggests.

CONFIDENCE: MEDIUMFare types
06 / Privacy & Proportionality5/10

Does the service collect roughly what it needs, rather than everything it can?

Observation: Travel requires substantial identity, passport and itinerary data. Cookie and privacy controls exist, but no full tracker audit was performed.

CONFIDENCE: LOWPrivacy policy
07 / Access for Humans8/10

Can a reasonably broad range of people actually use it?

Observation: BA publishes mobility, hearing and visual-assistance routes, including minicom and BSL relay support. The booking flow was not WCAG-audited.

CONFIDENCE: MEDIUMDisability assistance
08 / Does the Bloody Thing Work?6/10

Is it fast, reliable and technically competent enough to use?

Observation: Public information and search entry points loaded, but the large, script-heavy booking environment is denser and less resilient than a simple information service.

CONFIDENCE: MEDIUMBA home
09 / Can I Find a Human?9/10

Can I understand who is responsible and contact somebody when things go wrong?

Observation: BA publishes telephone numbers, opening hours, a postal address and specialist accessibility contacts alongside chat.

10 / Worth Visiting8/10

Does this actually contribute something useful to the web?

Observation: Flight status, baggage, destination, assistance and disruption information provide substantial value beyond sales copy.

CONFIDENCE: HIGHHelp and contacts

ASSESSMENT 007 / SMART-HOME PRODUCT

Hive

Test task: Understand whether a purchased Hive thermostat can perform its main functions without the app, account or an ongoing subscription.

51/100
BEST BIT

Basic heating control survives on the physical thermostat, and cancelling Hive Plus removes extras rather than the core device.

MAIN OFFENCE

Remote control and smart features depend on the Hive app, account and service infrastructure.

APP FOR NO REASONACCOUNT HOSTAGEYOUR PURCHASE NOW DEPENDS ON OUR SERVERS
01 / Just Let Me Use It5/10

Can I accomplish the core task without being forced through unnecessary barriers?

Observation: Basic temperature control is available on the thermostat, while setup and remote smart functions introduce account, hub and app steps.

02 / No-App Test2/10

Does this genuinely need an app?

Observation: Hive markets the mobile app as the control surface for schedules, remote changes and smart features; only basic local control survives without it.

CONFIDENCE: HIGHHive smart heating
03 / No-Account Test2/10

Do I genuinely need an account for this?

Observation: An account is required to connect and operate the smart service, even though the physical thermostat can still make local temperature changes.

CONFIDENCE: HIGHHive login
04 / Exit Is a Feature5/10

Can I leave?

Observation: Subscriptions can be cancelled and account closure is supported; cancellation removes paid history or features while devices retain more limited operation.

05 / Dark Pattern Test7/10

Is the interface trying to manipulate me?

Observation: The public product and support pages describe subscriptions and hardware roles, with no strong dark pattern observed. Signed-in cancellation was not tested.

CONFIDENCE: LOWHive support
06 / Privacy & Proportionality5/10

Does the service collect roughly what it needs, rather than everything it can?

Observation: A connected home service necessarily processes account and device data. Public notices exist, but device telemetry and app permissions were not independently audited.

07 / Access for Humans6/10

Can a reasonably broad range of people actually use it?

Observation: Physical controls provide a fallback and support documentation is available. The app and hardware were not tested with assistive technology.

CONFIDENCE: LOWHive user guides
08 / Does the Bloody Thing Work?6/10

Is it fast, reliable and technically competent enough to use?

Observation: Basic heating can work locally, but remote features depend on several components and connectivity. No live household system was stress-tested.

CONFIDENCE: LOWHive status
09 / Can I Find a Human?7/10

Can I understand who is responsible and contact somebody when things go wrong?

Observation: Hive publishes webchat and a customer service telephone number, plus specific routes for payment, cancellation and account closure.

CONFIDENCE: HIGHHive support
10 / Worth Visiting6/10

Does this actually contribute something useful to the web?

Observation: The site contains useful guides and energy information, but much of its value supports a proprietary device ecosystem.

CONFIDENCE: HIGHHive blog

ASSESSMENT 008 / PUBLIC SERVICE

GOV.UK

Test task: Check a person's State Pension age and find a non-digital route for pension help.

96/100
BEST BIT

The checker needs no account and the service publishes several genuinely usable non-digital support routes.

MAIN OFFENCE

The broader pension journey becomes more complex when moving from a general age check to a personal forecast.

JUST WORKSNO APP REQUIREDHUMAN FOUNDLOW-NONSENSE WEBSITEUSEFUL INTERNET
01 / Just Let Me Use It10/10

Can I accomplish the core task without being forced through unnecessary barriers?

Observation: The page explains the result, starts the checker directly and avoids marketing, onboarding and unrelated interruptions.

02 / No-App Test10/10

Does this genuinely need an app?

Observation: The complete age-checking task works in a web browser and has no app requirement.

03 / No-Account Test10/10

Do I genuinely need an account for this?

Observation: The general age checker does not require an account. A personal pension forecast appropriately requires identity verification and also offers phone and post routes.

04 / Exit Is a Feature10/10

Can I leave?

Observation: The task creates no account, stores no user-owned content and leaves nothing to cancel or retrieve.

05 / Dark Pattern Test10/10

Is the interface trying to manipulate me?

Observation: Choices are plain, neutral and free from upsells, false urgency or unequal commercial prompts.

06 / Privacy & Proportionality9/10

Does the service collect roughly what it needs, rather than everything it can?

Observation: The checker requests only the information needed for the calculation and GOV.UK exposes cookie controls. This was not a formal privacy audit.

CONFIDENCE: MEDIUMGOV.UK cookies
07 / Access for Humans9/10

Can a reasonably broad range of people actually use it?

Observation: The design is clear, keyboard-friendly and backed by an accessibility statement; the end-to-end service was not independently WCAG-audited.

CONFIDENCE: MEDIUMGOV.UK accessibility
08 / Does the Bloody Thing Work?9/10

Is it fast, reliable and technically competent enough to use?

Observation: The lightweight service loaded quickly and completed the defined task without visible errors.

09 / Can I Find a Human?9/10

Can I understand who is responsible and contact somebody when things go wrong?

Observation: The Pension Service publishes telephone, Relay UK, BSL video relay and Welsh-language routes with opening hours.

10 / Worth Visiting10/10

Does this actually contribute something useful to the web?

Observation: The service provides a concrete public answer, related guidance and openly licensed content worth linking to directly.

ASSESSMENT 009 / MAJOR PLATFORM

YouTube

Test task: Search for and watch one public video while signed out, then identify the available privacy and data controls.

70/100
BEST BIT

Public videos can still be searched, linked and watched on the web without creating an account.

MAIN OFFENCE

Behavioural data and recommendation systems are central to the experience, while meaningful human support is difficult for an ordinary signed-out user to reach.

NO APP REQUIREDUSEFUL INTERNETHUMAN SUPPORT MISSING
01 / Just Let Me Use It8/10

Can I accomplish the core task without being forced through unnecessary barriers?

Observation: Searching and watching a public video works signed out. Consent prompts and promotions add friction but do not block the core task.

CONFIDENCE: HIGHYouTube
02 / No-App Test10/10

Does this genuinely need an app?

Observation: The core viewing experience works in a browser on desktop and mobile; the app is optional for the defined task.

CONFIDENCE: HIGHYouTube mobile web
03 / No-Account Test10/10

Do I genuinely need an account for this?

Observation: An account is not required to search for or watch ordinary public videos, though personalisation, commenting and subscriptions require one.

CONFIDENCE: HIGHYouTube
04 / Exit Is a Feature6/10

Can I leave?

Observation: Google provides activity deletion and data export, but downloaded Premium videos expire when membership ends and creator/audience relationships remain platform-bound.

05 / Dark Pattern Test5/10

Is the interface trying to manipulate me?

Observation: Prominent recommendations, autoplay, Premium promotion and consent surfaces shape continued viewing. No single deceptive act was treated as universal.

CONFIDENCE: MEDIUMYouTube settings help
06 / Privacy & Proportionality3/10

Does the service collect roughly what it needs, rather than everything it can?

Observation: YouTube says it uses search, watch, location and account activity for recommendations and advertising, while also providing controls and deletion tools.

CONFIDENCE: HIGHPrivacy on YouTube
07 / Access for Humans8/10

Can a reasonably broad range of people actually use it?

Observation: Captions, transcripts, keyboard controls and screen-reader guidance exist, but creator-supplied caption quality and individual videos vary.

08 / Does the Bloody Thing Work?8/10

Is it fast, reliable and technically competent enough to use?

Observation: Playback and search are mature and broadly compatible, although the page is script-heavy and performance varies with video, ads and device.

CONFIDENCE: MEDIUMSupported browsers
09 / Can I Find a Human?2/10

Can I understand who is responsible and contact somebody when things go wrong?

Observation: Help articles and community support are extensive, but an ordinary signed-out viewer has no clear general route to a responsible human.

CONFIDENCE: HIGHYouTube Help
10 / Worth Visiting10/10

Does this actually contribute something useful to the web?

Observation: YouTube contains an extraordinary range of useful, original and culturally important material that people intentionally seek and revisit.

CONFIDENCE: HIGHYouTube

ASSESSMENT 0010 / SMALL INDEPENDENT WEBSITE

Marginalia Search

Test task: Search for an obscure topic, explore a result and understand who operates the search engine and what it records.

95/100
BEST BIT

An independent, open-source search index built to send people towards overlooked websites rather than keep them inside a platform.

MAIN OFFENCE

A small independent service cannot promise the universal reliability, support coverage or accessibility assurance of public infrastructure.

JUST WORKSNO APP REQUIREDLOW-NONSENSE WEBSITEGOOD OLD-FASHIONED LINKUSEFUL INTERNETHUMAN FOUND
01 / Just Let Me Use It10/10

Can I accomplish the core task without being forced through unnecessary barriers?

Observation: A search box and discovery routes are available immediately, without onboarding, popups or marketing gates.

CONFIDENCE: HIGHMarginalia Search
02 / No-App Test10/10

Does this genuinely need an app?

Observation: The complete service is an ordinary website. No app exists or is required.

CONFIDENCE: HIGHMarginalia Search
03 / No-Account Test10/10

Do I genuinely need an account for this?

Observation: Search and discovery require no account or identity capture.

CONFIDENCE: HIGHAbout Marginalia
04 / Exit Is a Feature10/10

Can I leave?

Observation: There is no account to leave; the AGPL source can be inspected and self-hosted, reducing service lock-in.

CONFIDENCE: HIGHMarginalia source
05 / Dark Pattern Test10/10

Is the interface trying to manipulate me?

Observation: The reviewed interface contains no advertising, false urgency, unequal purchase choice or manipulative retention flow.

CONFIDENCE: HIGHMarginalia Search
06 / Privacy & Proportionality10/10

Does the service collect roughly what it needs, rather than everything it can?

Observation: Marginalia says it uses no non-functional cookies, keeps IP-bearing logs for up to 24 hours and minimises query logging.

07 / Access for Humans7/10

Can a reasonably broad range of people actually use it?

Observation: The text-first interface is comparatively simple, but no current formal accessibility statement or independent audit was found.

CONFIDENCE: LOWMarginalia Search
08 / Does the Bloody Thing Work?8/10

Is it fast, reliable and technically competent enough to use?

Observation: The service returned results during review and uses deliberately modest infrastructure; a small project cannot offer the same redundancy as a major platform.

CONFIDENCE: MEDIUMAbout Marginalia
09 / Can I Find a Human?10/10

Can I understand who is responsible and contact somebody when things go wrong?

Observation: The operator is named and publishes a direct email address alongside project and source information.

CONFIDENCE: HIGHAbout Marginalia
10 / Worth Visiting10/10

Does this actually contribute something useful to the web?

Observation: It provides an original index, discovery tools and direct links specifically designed to reveal overlooked parts of the web.

CONFIDENCE: HIGHMarginalia source

THE V1 RUBRIC

Ten categories. Zero secret sauce.

Each category is scored from 0 to 10 using the same rubric. Every published score records an observation, evidence, score and confidence. Missing evidence creates uncertainty, not a convenient invention.

01

Just Let Me Use It

Can I accomplish the core task without being forced through unnecessary barriers?

10Simple, direct and unusually user-respecting.

7-9Generally good, with minor friction or limitations.

4-6Noticeable barriers, dependency or manipulation.

1-3The task is significantly obstructed or user control is poor.

0The reasonable core task is blocked, broken or fundamentally hostile.

02

No-App Test

Does this genuinely need an app?

10Simple, direct and unusually user-respecting.

7-9Generally good, with minor friction or limitations.

4-6Noticeable barriers, dependency or manipulation.

1-3The task is significantly obstructed or user control is poor.

0The reasonable core task is blocked, broken or fundamentally hostile.

03

No-Account Test

Do I genuinely need an account for this?

10Simple, direct and unusually user-respecting.

7-9Generally good, with minor friction or limitations.

4-6Noticeable barriers, dependency or manipulation.

1-3The task is significantly obstructed or user control is poor.

0The reasonable core task is blocked, broken or fundamentally hostile.

04

Exit Is a Feature

Can I leave?

10Simple, direct and unusually user-respecting.

7-9Generally good, with minor friction or limitations.

4-6Noticeable barriers, dependency or manipulation.

1-3The task is significantly obstructed or user control is poor.

0The reasonable core task is blocked, broken or fundamentally hostile.

05

Dark Pattern Test

Is the interface trying to manipulate me?

10Simple, direct and unusually user-respecting.

7-9Generally good, with minor friction or limitations.

4-6Noticeable barriers, dependency or manipulation.

1-3The task is significantly obstructed or user control is poor.

0The reasonable core task is blocked, broken or fundamentally hostile.

06

Privacy & Proportionality

Does the service collect roughly what it needs, rather than everything it can?

10Simple, direct and unusually user-respecting.

7-9Generally good, with minor friction or limitations.

4-6Noticeable barriers, dependency or manipulation.

1-3The task is significantly obstructed or user control is poor.

0The reasonable core task is blocked, broken or fundamentally hostile.

07

Access for Humans

Can a reasonably broad range of people actually use it?

10Simple, direct and unusually user-respecting.

7-9Generally good, with minor friction or limitations.

4-6Noticeable barriers, dependency or manipulation.

1-3The task is significantly obstructed or user control is poor.

0The reasonable core task is blocked, broken or fundamentally hostile.

08

Does the Bloody Thing Work?

Is it fast, reliable and technically competent enough to use?

10Simple, direct and unusually user-respecting.

7-9Generally good, with minor friction or limitations.

4-6Noticeable barriers, dependency or manipulation.

1-3The task is significantly obstructed or user control is poor.

0The reasonable core task is blocked, broken or fundamentally hostile.

09

Can I Find a Human?

Can I understand who is responsible and contact somebody when things go wrong?

10Simple, direct and unusually user-respecting.

7-9Generally good, with minor friction or limitations.

4-6Noticeable barriers, dependency or manipulation.

1-3The task is significantly obstructed or user control is poor.

0The reasonable core task is blocked, broken or fundamentally hostile.

10

Worth Visiting

Does this actually contribute something useful to the web?

10Simple, direct and unusually user-respecting.

7-9Generally good, with minor friction or limitations.

4-6Noticeable barriers, dependency or manipulation.

1-3The task is significantly obstructed or user control is poor.

0The reasonable core task is blocked, broken or fundamentally hostile.

HOW A SCORE IS MADE

A task first. Evidence second. Jokes last.

  1. Define one simple, realistic user task.
  2. Test it on desktop or mobile as relevant.
  3. Record what happened and the page or action supporting it.
  4. Score all ten dimensions without changing the rules for a liked or disliked brand.
  5. Name the best bit and main offence.
  6. Use labels only when the evidence clearly earns them.
  7. Add the ten scores and write a short, fair BIC verdict.

Serious fails

Used sparingly. A label is not an extra numerical penalty.

APP FOR NO REASONACCOUNT HOSTAGECANCEL QUESTCOOKIE OBSTACLE COURSEHUMAN SUPPORT MISSINGLOCK-IN SPECIALDARK PATTERN SPECIALYOUR PURCHASE NOW DEPENDS ON OUR SERVERSPOPUP PINBALL

Positive labels

For unusually direct, portable or useful behaviour.

JUST WORKSEXIT IS A FEATURENO APP REQUIREDHUMAN FOUNDUSEFUL INTERNETLOW-NONSENSE WEBSITEGOOD OLD-FASHIONED LINK

THE SCORE BANDS

A hundred points. Five possible diagnoses.

90-100

Good Internet

Unusually good at letting people accomplish what they came to do.

75-89

Mostly Civilised

Generally good, with a few unnecessary annoyances or dependencies.

60-74

Getting a Bit Platformy

Usable, but meaningful friction or dependency is appearing.

40-59

Mild Hostage Situation

Significant friction, lock-in, app/account dependency or poor control.

0-39

Your Fridge Now Needs an Account

A deeply user-hostile or unnecessarily complicated experience.

FAIRNESS, CORRECTIONS & CONFLICTS

The joke sits on top of evidence.

Scores can change. A company can point out a factual error, explain something missed or request reassessment. Improvements should improve the score; this is not permanent shaming.

A company can never pay for a better result. If BIC later sells assessments, consulting or sponsorship, relevant relationships will be disclosed and the scoring method will remain independent.

This is editorial research, not certification. Experiment A is complete; the next step is to review overlap, subjectivity, evidence gaps and weighting before publishing rubric v1.1. No badge is available.

Send a correction or reassessment note

FOLLOW THE REPEATING PATTERN

Ten profiles become one evidence board.

When everything wants an account or an app

Compare forced registration, justified accounts, guest access, app dependency and open-web alternatives across all ten assessed services.

Explore the first BIC pattern

NEXT EXPERIMENT

Score your own corner first.

How much of your internet presence do you actually own?

Try the Internet Health Check