Editorial Methodology Current standard
How We Evaluate Displays, Docks & Multi-Screen Setups
Before trusting a screen-extender or display review, you should be able to verify five things: access, identity, evidence, compatibility and accountability.
Run the 60-second review audit
Every verdict is traceable along this chain
- Claim
- Evidence
- Gate
- Verdict
- Record
01 Reader audit
Before you trust a display review, check these five signals
This works on any display review on any site, including ours. Each check takes about ten seconds.
-
01Access
Does the page say whether the product was physically accessed, partly accessed, or not accessed at all?
A plain statement near the top: hands-on with a named unit, research-based with no hardware access, or mixed — in the writer's own words.
Photographs and confident sensory language with no access statement anywhere on the page.
-
02Identity
Is the exact model, revision, configuration or listing variant clear enough that you could buy the same thing?
Full model string, panel size, the specific variant where several exist, and a note when one listing sells more than one configuration.
A category name doing the work of a model name — "this 15.6-inch portable monitor" — in a market where near-identical shells carry different panels and ports.
-
03Evidence
Can you tell measurements and direct observations apart from documentation, third-party reports and the writer's own inference?
Attribution at claim level: this figure is the manufacturer's, that behavior was observed, this conclusion is inference and is labeled as inference.
Specification-sheet numbers written in the grammar of experience — "we found the panel hits 300 nits" when 300 nits is simply the published figure.
-
04Compatibility
Are host limits, operating-system behavior, video path, power and physical fit named before the recommendation?
The page tells you which machines this fails on, above the purchase link rather than in a closing caveat.
"Universal compatibility" or "works with any USB-C laptop" — a USB-C connector alone does not establish that a port carries video.
-
05Accountability
Are the author, verification date, evidence confidence, commercial disclosure and correction route visible on the page?
A named human, a date the facts were last checked, a stated confidence level, a findable affiliate disclosure, and a route to report an error.
A staff byline with no person behind it, or an "updated" date that moves while nothing on the page changes.
How to weigh the result
These five are not equally weighted, and there is no threshold at which a page automatically becomes worthless. Read the result like this:
- Some failures are critical on their own. An unresolved product identity, an undisclosed evidence basis, or no support at all for the claim your decision rests on — any one of these is enough to stop acting on the recommendation.
- Other omissions reduce confidence without invalidating the page. A missing verification date on an otherwise well-sourced explainer is a gap, not a disqualification.
- The more signals a page fails, the less weight its recommendation can bear. That is a judgment you make, not a score the page computes.
This is a rapid reader audit, not a validated scientific instrument. It is designed to tell you how much to trust what you are reading, and where to look next.
02 Evidence model
What our evidence labels actually mean
A score is the end of the process, not the evidence. Three questions have to be answerable on every page we publish — how close we got to the product, where each individual claim came from, and how much weight that evidence will bear. They are separate questions, so we answer them separately.
One evidence basis per page
Every review or recommendation governed by this standard declares exactly one of these, near the top rather than in a footnote.
Hands-on
We had direct physical access to the named product and configuration. The page identifies the access conditions and keeps measured results separate from qualitative observations.
It does not mean every published specification was independently verified. Access is not a warranty over the spec sheet.
Research-based
No direct physical access was used for this evaluation. The assessment is built from identified primary documentation, standards and manufacturer sources, dated commercial listings, prior verified records and carefully qualified analysis.
It may not use "tested", "measured", "observed" or "lab" for the current evaluation. Research-based work says so plainly.
Mixed
Some findings come from direct access and others from documentation, reports or inference. The page identifies which is which at claim level.
A hands-on component may not lend direct-evidence authority to claims that were never checked directly.
Seven claim labels
Page-level honesty is not enough, because a single page mixes claim types constantly. Any material claim should be able to carry one of these.
Measured
A quantitative result recorded during direct access. Carries the tool or method, the condition, the exact unit and the date.
Observed
A qualitative behavior seen directly during product access. Carries the condition, unit and date — and is never converted into a number that was not measured.
Primary-documented
A claim taken from a standards body, platform owner, manufacturer manual, support page or specification. Carries the named source, a link where one exists, the exact model or variant, and the date we checked it.
Independently sourced
A claim from an identifiable, credible third party. Carries who the source is, why we are relying on it, and any conflict with primary documentation.
Corroborated report
A recurring field pattern across multiple identifiable reports. Establishes that a failure pattern exists — never how often it occurs.
Editorial inference
A reasoned conclusion drawn from declared inputs, with the reasoning boundary stated and hedged language: suggests, likely, we infer.
Unknown or conflicted
Evidence is missing, stale, variant-dependent or contradictory. The page says what is unknown, why it matters to your decision, and whether it blocks a score.
One sentence, three claims
Take a sentence of the kind that appears in display reviews everywhere: "A bright 300-nit panel that stays readable near a window, though it dims noticeably on battery." That is three claims of very different strength.
- "300-nit panel" is Primary-documented when it comes from the manufacturer's specification. It becomes Measured only if an instrument was put against that exact unit.
- "stays readable near a window" is Observed, and only where we had the unit in that light. It cannot be derived from the nit figure.
- "dims noticeably on battery" is a Corroborated report or an Editorial inference from the power budget. A reader who needs full brightness untethered is deciding on that clause alone, so the page has to say which.
A generic example, on purpose. It is not a finding about any real product.
Confidence is graded separately from quality
Each governed review carries an evidence-confidence grade describing the strength of the evidence, not the merit of the product. Confidence never adds points to a score. A good product can carry low-confidence evidence, and a poor one can be documented with high confidence.
- High confidence
Current direct evidence or consistent primary documentation supports the material claims, and the exact model is unambiguous.
- Moderate confidence
Credible but partly indirect, incomplete, or dependent on conditions we have stated.
- Low confidence
Material claims rest on limited, stale, conflicted or variant-uncertain evidence. The page says which claims are affected.
Source authority depends on the claim
There is no single ranking of sources that holds for every question. A manufacturer specification is authoritative for what a product is designed to do and weak evidence for how it performs; a careful measurement is the opposite. So authority is assigned per claim type.
Protocol or interface capability
Standards body, platform owner, and exact-model manufacturer documentation. What a port is specified to carry is defined by the specification, not by a reviewer's experience of one unit.
Quantitative performance
Documented direct measurement under stated conditions. Here a properly documented measurement outranks the published figure, because the specification describes a target and the measurement describes an outcome.
Qualitative behavior
Direct observation under stated conditions. Wake-from-sleep behavior, stand flex and coating glare are seen, not specified.
Failure patterns
Credible independent technical evidence and corroborated reports. These establish that a pattern exists and is worth checking; they never establish incidence.
Price, availability, box contents
Timestamped retailer or manufacturer listing. Evidence of what was advertised on a date, and nothing more — which is why these claims carry the date they were checked.
Unknown or conflicting
Left explicitly unresolved. We record the conflict, identify the model, region and date each source refers to, and check whether the disagreement is really about two variants, two firmware states or two OS versions — which it may be. We do not average incompatible facts, and an unresolved claim can block the score.
The terms used here and across our reviews are defined in the glossary, and the questions readers ask about compatibility are worked through in the compatibility guide.
03 Process
The six-stage evaluation record
This is a decision record, not a checklist ticked after the writing is done. Each stage produces an output that has to exist before the next one runs, and the last stage is what you see on the page.
-
Define the decision and the exact identity
Establish whose decision this is — the workflow, the constraint that actually hurts, the outcome that counts as success — then pin the product down: model, revision, listing variant, included accessories, target market. A review that cannot name what it evaluated cannot be checked.
Output → named use case + unambiguous product identity -
Declare access and build the source ledger
State the evidence basis before gathering anything, so the writing cannot drift into implied access later. Then collect sources in the order that suits the claim type, recording what each one actually supports.
Output → declared basis + dated source ledger -
Run the compatibility gates
Determine whether the system can work at all before asking whether it is desirable. The five gates below are run against the named use case, not against an abstract "typical laptop".
Output → compatibility state + the conditions attached to it -
Evaluate performance, failure modes and trade-offs
Assess what matters for the named use, in the order it would bite. Every product trades something for something else; the job is to name the trade, not celebrate the feature. This stage also records what could not be checked — often the most useful sentence on the page.
Output → findings, failure modes, explicit list of gaps -
Score and issue a suitability verdict
Apply the published category rubric, attach confidence and compatibility state as separate outputs, and apply the veto rules. Some evaluations stop here without a number, and say why.
Output → score or reasoned refusal + who should skip this -
Publish with accountability
Ship with a named author, the method note, the verification date, the commercial disclosure and a route for readers to challenge it. A recommendation nobody can be held to is not a recommendation.
Output → a page the reader can audit
What this standard governs, and what it does not
Methodology v2.0 governs content first published or substantively re-audited on or after its effective date. Earlier pages may remain labeled Legacy until their identity, evidence basis, compatibility claims and score have been re-verified. A page-specific evidence note takes precedence over any general description here.
We are not claiming the existing archive has been migrated. Re-auditing is page-by-page work, and a page carries the current label only once that work is done and dated. Not every page is a product test, either: comparisons declare inclusion criteria and item-by-item evidence, explainers publish a source ledger and say when no hardware was involved, troubleshooting guides state their diagnostic assumptions and their escalation point, and decision tools name their inputs, their limits and what happens when an answer is unknown.
04 Compatibility
Compatibility is checked before score
A weighted score cannot rescue a product that fails a requirement your machine imposes. Many avoidable disappointments in this category are not quality failures — they are compatibility failures that were predictable before purchase. So compatibility runs first, and it can veto everything after it.
-
Gate 1
Host and display ceiling
What the machine itself can drive: the GPU or SoC, how many external displays it supports, which resolution and refresh combinations are reachable, and whether an arrangement is natively supported or depends on a driver path. Ceilings differ by chip tier and OS version, so this is checked against the specific host, not the brand.
-
Gate 2
Signal path
A USB-C connector is a shape, not a capability — and the symbols printed beside a port only sometimes tell you the difference. This gate identifies what the port and cable actually carry: DisplayPort Alt Mode, Thunderbolt, USB4, HDMI, DisplayPort, DisplayLink, MST or another architecture.
-
Gate 3
Power path and budget
Where power comes from, and how many budgets are actually in play. On a shared supply or pass-through path, one budget may be divided between the host, the dock and the displays. Separately powered components each negotiate their own supply. A host-powered display is limited by what that port can output. Under-powered setups can fail unclearly: depending on the architecture the symptom may be flicker, a dropout under load, a failed power negotiation, slow or absent charging, or battery drain while plugged in.
-
Gate 4
Physical fit and load
Laptop size range, attachment geometry, stand or clamp behavior, where the weight sits, desk footprint, and the hinge and ergonomic consequences of hanging mass off a lid. Fit is assessed against documented outer dimensions and declared fit constraints — measured directly where we have access — and never inferred from a screen diagonal.
-
Gate 5
OS, driver and workflow fit
Driver availability and permissions, sleep and wake behavior, protected-content or latency limits where they matter, travel versus fixed-desk conditions — and the question that closes every review: who should skip this product entirely?
The two paths a multi-screen setup depends on
The same figure, written out
Video path. Host GPU or SoC (gate 1) → port and cable (gate 2) → either a native path over DisplayPort Alt Mode, Thunderbolt, USB4, HDMI or DisplayPort, or a DisplayLink path that sends compressed display data over USB and renders it through a driver → the displays.
Power path. Supply, whether a wall adapter or the host port itself → watts actually delivered after cable and adapter losses (gate 3) → whatever budgets that architecture creates. A shared or pass-through supply divides one budget between host, dock and panels; separately powered components negotiate independently; a host-powered display draws on the port's own output. When a budget is short, the symptom can surface on either side — a display that flickers or drops, or a host that charges slowly, fails to negotiate, or drains while connected.
Native is not the same as DisplayLink. A native path is constrained by what the host is architecturally capable of driving. On supported systems a DisplayLink path may add displays beyond a native external-display limit — whether it does depends on the host, the operating system and the driver — and it introduces a software dependency, a compression stage and a permissions requirement on some operating systems. Neither is universally better; they fail differently, and a review should say which one it is describing. When one of these fails in practice it usually presents as a symptom rather than a cause, which is why our dock detection checks start from the gates rather than the symptom.
One of four compatibility states
Every relevant evaluation ends with exactly one, stated as text rather than signaled by color:
✓ Compatible for the named use ▲ Conditional ✗ Not compatible for the named use □ Unverified
A conditional answer is useful only when its conditions are named: which chip tier, which port, which OS version, which power arrangement. "Conditional" on its own is a shrug.
Check your inbox for the audit.
Your Pre-Purchase Compatibility Audit is on its way. Use it before buying a laptop screen extender, portable monitor, docking station, or desktop setup so you can avoid the most common compatibility mistakes.
05 Judgment
Hands-on boundaries, scoring and verdicts
"Hands-on" has been worn smooth by overuse. Here it is a narrow, checkable claim about a specific unit at a specific time, and it belongs on the review that earned it — not as a standing promise on a policy page.
A hands-on review names the unit it actually had: model and variant, where it came from, the conditions it was assessed under, the methods or instruments behind anything reported as a number, and the date. Those details live on the review because that is where they can be checked. A blanket claim here that "everything is tested" would be unverifiable by design, which is exactly why we do not make one.
Measurement and observation are different things. A measurement is a number produced by a method against a named unit. An observation is something seen — a stand that flexes, a wake-from-sleep sequence that drops a display, a coating that scatters window light. Observations are frequently more decision-relevant than numbers, and they never become numbers afterwards. Holding a product also does not confirm its published claims: a brightness figure taken from the manufacturer stays Primary-documented even on a hands-on page until someone measures it.
Research-based work is not a lesser form of review. For compatibility questions it is often the more rigorous option, because the answer lives in documentation and geometry rather than on one writer's desk. Our small-desk dual-monitor analysis is a current example: named method, cited sources, a dated verification line, an explicit statement that no hardware was assembled or measured, and "Not proven" returned instead of a guess when a required input is missing.
How the number is built
The public score is a ScreenExtendersHub editorial score on a 0–10 scale, always written as /10. It is not a star rating and not an average of customer sentiment. Each category uses a published, page-specific weighted rubric whose weights total 100%, and the weighted result is shown to one decimal place.
final score = sum of (axis score × axis weight)
| Axis | Weight | Score | Contribution |
|---|---|---|---|
| Axis A | 0.30 | 8.0 | 8.0 × 0.30 = 2.40 |
| Axis B | 0.25 | 9.0 | 9.0 × 0.25 = 2.25 |
| Axis C | 0.20 | 7.0 | 7.0 × 0.20 = 1.40 |
| Axis D | 0.15 | 8.5 | 8.5 × 0.15 = 1.28 |
| Axis E | 0.10 | 6.5 | 6.5 × 0.10 = 0.65 |
| Total | 1.00 | — | 7.98 → 8.0/10 |
Axis names and weights belong on the review itself, because they differ by category; pages carry them as they are written or re-audited under this standard, and a page that does not yet show its axes has not yet been through that work. Scores are comparable only where the same axes and weighting were used — an 8.0 built from a laptop-extender rubric is not equivalent to an 8.0 built from a docking-station rubric. When products need to be set against each other, that work belongs in a head-to-head comparison on shared axes, not in a score subtraction.
Three outputs, never merged
- Compatibility stateConditional
Can this work on the reader's system, and under what named condition?
- Evidence confidenceModerate
How much weight will the evidence behind those claims bear?
- Editorial score8.0/10
How good is the product for the named use, once it clears the gates?
Illustrative values. Collapsing these three into one number is a common way review scores can mislead.
When arithmetic is overruled
A compatibility veto, unresolved product identity or insufficient evidence can override the calculation and produce an unscored verdict. A dock that cannot drive the arrangement a reader came for does not get to average its way back to a recommendation on build quality. Unknown facts are not guessed, and an unknown is not silently scored as zero — that would punish a product for our gap rather than its own.
Worked case — two units, one name
At least two Kado 15.6-inch listings carry near-identical titles, photographs and price band, and are not the same product. We know because we evaluated them separately, in April 2025 and again in February 2026. One unit runs 75 Hz with a 100% sRGB gamut claim; the other charges a laptop at 45 W. Kado publishes no brightness or weight figure for either.
The consequences run in two directions. The 75 Hz unit scored 7.6/10 — below an Acer at 9.1 and a KYY at 8.5 that it beats on paper for gamut and refresh — because those two publish brightness and weight you can check and Kado does not. The 45 W unit carries no number at all: the evidence available could support a qualitative read and not a scored one, so a qualitative read is what shipped. Neither outcome came from the arithmetic. Both came from the gap between what is claimed and what is verifiable.
The full working, including which unit suits which buyer and who should avoid both, is in the Kado 15.6-inch analysis.
Sometimes no score is the more honest output. A number implies a precision the evidence may not support. Where a product's identity is ambiguous across listing variants, or the decisive claim is unverifiable, publishing "not scored" with the reason attached tells you more than a confident 8.4 would.
There is no minimum score for publication. Negative, limited and unscored evaluations may be published whenever that protects a reader from a bad or incompatible decision. This is a change from the previous standard, which stated that products scoring below 8.0 were not published: a publication floor produces a library in which every product looks good, which is the pattern this methodology exists to argue against. Removing the floor changes what we may publish going forward; it does not retroactively re-score anything already published.
Topics are selected by the decision they represent — a recurring compatibility question, a category where buyers are routinely caught out, a listing whose claim is worth checking. We do not claim market data we do not have, and we do not treat popularity as evidence of suitability. Published pages are then re-verified when the facts still hold, amended when a claim was wrong or overtaken by a driver or platform change, re-scored when the evaluation changes materially, demoted to legacy when they predate the current standard, or retired when a product is gone in a way that makes the page misleading. The error classification and correction timescales are set out in the Corrections & Updates Policy, which is authoritative on that subject.
06 Accountability
Independence, tools and the record we publish
Affiliate revenue funds this site, and pretending otherwise would undermine everything above. Some links are affiliate links; if you buy through one we may earn a commission at no extra cost to you. What matters is where the link sits relative to the reasoning — compatibility constraints and the "who should skip this" section belong above the purchase link, not below it. The complete policy is on our disclosure page.
Where a product was supplied free, loaned, discounted or otherwise provided by a manufacturer, retailer or intermediary, that relationship is disclosed on the page it affects, in the evidence record, where it is relevant to the claim being made. Where the best answer is "don't buy this", or "the category you are looking at is the wrong one", the page says so. A recommendation that could not survive the removal of its affiliate links was never a recommendation.
Software and AI-assisted tools may help organize research, compare records, check structure or draft presentation. They are not accepted as evidence. A named human author remains responsible for verifying material claims against the cited sources before publication. That boundary is narrow and deliberate: a tool can find a contradiction between two documents but cannot establish which is true; it can restructure a page but cannot observe a stand flexing.
Editorial responsibility sits with a named person — Boniface Musembi, Independent Research Analyst, Portable Productivity Systems. The editorial principles behind the site are in About Us, the limits of what any published guidance can promise are in the disclaimer, and the editor can be reached through contact. Substantive disagreement backed by your own hardware details is treated as evidence, not as complaint.
ScreenExtendersHub is a participant in the Amazon Services LLC Associates Program, an affiliate advertising program designed to provide a means for sites to earn advertising fees by advertising and linking to Amazon.com.
The record a governed review carries
This is the evidence passport you should be able to find on a page evaluated under this standard. Everything below is generic and fictional — it demonstrates the fields, not a verdict.
- Content type
- Product review
- Exact product
- Generic 14-inch dual-panel laptop screen extender, revision B, single-cable variant
- Evidence basis
- Mixed
- Hardware access
- Partial — attachment and stand behavior assessed directly; panel and power figures not independently measured
- Claim labels used
- Observed, Primary-documented, Corroborated report, Editorial inference, Unknown or conflicted
- Compatibility state
- Conditional — requires a host port that carries display signal; not usable on data-only USB-C
- Evidence confidence
- Moderate — brightness and power claims rest on manufacturer documentation, not measurement
- Editorial score
- Not scored — the listing sells two panel variants under one model name and the unit could not be matched to a purchasable configuration
- Sources last checked
- A dated line naming when each primary source was last opened
- Analyst of record
- Boniface Musembi, Independent Research Analyst, Portable Productivity Systems
- Commercial relationship
- Retail purchase, no manufacturer relationship; page carries affiliate links
- Re-verification trigger
- A driver change, a firmware revision, or a listing that changes what ships in the box
- Methodology version
- v2.0
Note what the illustrative record does in its two most important rows: it declines to score, and it says exactly why.
The same record, from a page we published
A demonstration proves nothing on its own, so here is the record a live page actually carries. Every value below is on the small-desk dual-monitor analysis and can be checked against it.
- Content type
- Decision tool and geometric analysis — not a product review
- Evidence basis
- Research-based, declared on the page as documentary and geometric analysis
- Hardware access
- None. The page states in terms that no dual-monitor desk was assembled, measured or bench-tested for that edition
- Claim labels used
- Primary-documented (manufacturer dimensions and workplace guidance), Editorial inference (geometry derived from those figures)
- Compatibility state
- Returned per input, not per product: a documented conflict returns Does not fit as drawn
- Evidence confidence
- Bounded by what the reader supplies — the instrument sees nothing and says so
- Editorial score
- None. A decision tool is not scored, and the page does not manufacture a number to look like a review
- Unknown handling
- Not proven is returned wherever a required measurement is missing, instead of a guess
- Sources
- Eleven cited, each attached to a visible claim, each with the date it was opened and the condition that would trigger a recheck
- Analyst of record
- Boniface Musembi, Independent Research Analyst, Portable Productivity Systems
- Verification line
- Carried on the page, dated, and naming what was verified rather than when the file last changed
- Methodology version
- v2.0
The two records differ in one instructive way. The illustrative one refuses a score because a product could not be identified; the live one carries no score because the page is an instrument rather than a verdict. Both are legitimate outcomes, and neither is a number withheld to seem cautious.
Standards this methodology answers to
Technical and product authority
Interface and platform behavior is settled against the standards bodies, platform owners and manufacturer documentation that define it, per the claim-type table in chapter 02. No third party certifies our evaluations, and we do not claim otherwise.
Editorial and disclosure
Material connections are disclosed in line with the US Federal Trade Commission's endorsement and review guidance (opens in a new tab). We describe our practice; we do not claim certification by any regulator. Search-platform guidance on review quality is relevant to how these pages are published rather than to how products are judged, so it sits in the implementation register below rather than here.
Accessibility
Components are designed and tested toward the W3C's WCAG 2.2 Level AA (opens in a new tab) criteria. Conformance can be determined only on the complete assembled page across its responsive variations — theme header, breadcrumb, page content, embedded form and site footer together, not any component in isolation. Our current position and known gaps are therefore published in the accessibility statement rather than asserted here.
Search, metadata and distribution
Implementation guidance for structured data, performance and social metadata governs how pages are published, not how products are evaluated. It is listed separately in the register below so it is never mistaken for part of the evaluation method.
Implementation register and methodology history
Publishing and distribution references
- Review quality: Google Search Central on writing high-quality reviews (opens in a new tab), which treats evidence of first-hand knowledge and original analysis as what distinguishes a useful review. Listed here, not among the standards above, because it describes publishing practice rather than evaluation method.
- Google Search Central — helpful, reliable, people-first content (opens in a new tab) and the reviews system (opens in a new tab).
- Structured data: general guidelines (opens in a new tab) and Article documentation (opens in a new tab). Markup describes content visible on the page and is not added to manufacture a rich result.
- Performance: Core Web Vitals (opens in a new tab) — 2.5 s Largest Contentful Paint, 200 ms Interaction to Next Paint, 0.1 Cumulative Layout Shift, each at the 75th percentile of real visits.
- Social metadata: Pinterest Article Rich Pins (opens in a new tab), which defines the headline, description and author metadata an article destination exposes.
- Endorsement detail: the FTC's Endorsement Guides Q&A (opens in a new tab).
Primary technical authorities
Material claims about interface capability, video paths and power negotiation are settled against the bodies that define them, not against secondary summaries.
- Power negotiation and budgets. USB-IF, USB Power Delivery (opens in a new tab) — the specification that defines what a source and a sink may negotiate, and therefore what a pass-through budget can and cannot promise.
- USB-C names the connector, not the video mechanism. The same oval port may carry display signal by more than one architecture, or by none at all. The four mechanisms below are different from each other, and which one a given port supports is a property of the device, its implementation, its operating system and its revision — never of the connector shape.
- Native DisplayPort over USB-C — Alt Mode. VESA, DisplayPort Alt Mode 2.0 (opens in a new tab), and the original DisplayPort Alt Mode on USB Type-C (opens in a new tab) announcement — the mechanism that assigns the connector's high-speed lanes to native DisplayPort signalling. Alt Mode is one of the ways display reaches a USB-C port; it is not the only one, and its absence does not establish that a port cannot drive a display.
- DisplayPort tunnelled through USB4. USB-IF, USB4 (opens in a new tab) — USB-IF describes USB4 as an architecture that shares one high-speed link dynamically between multiple simultaneous data and display protocols, with the host allocating bandwidth for display data flow. Display signal therefore travels inside the USB4 fabric rather than over lanes handed to DisplayPort. What any individual USB4 port can drive depends on what that host, that device and that revision implement: USB4 ports are not uniform, and one port's display behavior cannot be read off another's.
- DisplayPort tunnelled through Thunderbolt. Intel, Thunderbolt technology (opens in a new tab) — Intel states that Thunderbolt is more than a USB-C connector, that certification mandates capabilities the connector alone does not guarantee, and that current generations are built on the USB4 and DisplayPort specifications, with Thunderbolt 4 requiring dual-display support and Thunderbolt 5 raising that ceiling. Generation and host implementation both decide the answer.
- DisplayLink — not native DisplayPort at all. Synaptics, How Does DisplayLink Work? (opens in a new tab) — host software presents itself to the operating system as a display, the operating system writes into a frame buffer, the DisplayLink driver encodes those pixels with a codec, the encoded data crosses the link as USB data, and DisplayLink hardware in the dock or adapter decodes it and drives a standard video output. It is a software-plus-silicon path, not DisplayPort Alt Mode, and it carries driver, permission and compression dependencies that the native paths do not.
- Revision dependence. Where a claim turns on a specific revision — a USB4 version, a Thunderbolt generation, a DisplayPort revision, a DisplayLink chipset — the publisher's own document for that revision is consulted and cited on the page that relies on it, rather than summarized here. Nothing above should be read as asserting that every port carrying one of these names behaves alike.
Methodology history
Current standard (v2.0), effective 11 August 2026. Introduced the three evidence bases, seven claim labels and a separate confidence grade; added the five compatibility gates and the compatibility veto; separated compatibility state, confidence and score into three outputs; replaced the fixed source ranking with claim-specific source authority; removed the publication floor; published the legacy boundary; withdrew the static equipment inventory and blanket testing language; withdrew the fixed review-count and rating-range statements; and removed page-level structured data that duplicated the site graph.
Previous standard (v1.1), February 2026 — archived. A five-step framework: workflow context, hardware constraints, compatibility risk, trade-off analysis, disclosure. All five ideas survive inside the six-stage process above. Its specific claims about universal testing, a fixed number of published reviews, a rating range and a minimum publishable score are withdrawn and are not current policy anywhere they may still be quoted.
Next
Where to apply this standard next
-
The review library
The review library contains both current and legacy work. Use the five-signal audit on every page, and treat each page's own evidence note and verification date as authoritative.
-
Best laptop screen extenders
A ranked guide built compatibility-first: it works from host and port constraints toward the shortlist rather than the other way round.
-
The Pre-Purchase Compatibility Audit
The workbook applies the same compatibility-first logic to your own hardware, port by port. Start here if you skipped the form above.
Editorial Methodology · current standard, last reviewed 12 August 2026 · maintained by Boniface Musembi, Independent Research Analyst, Portable Productivity Systems.
This page describes our operating standard. Where a specific page carries its own evidence note, that note governs that page.