How we assess pledges
This page explains how we choose pledges from the documents governments submit, how we judge each one, and what we cannot measure. It is for journalists and researchers who want to know what each status on the site rests on, and to check it themselves.
Some assessments on the site are still drafts the team has not reviewed; the banner at the top gives the number.
What we track
We track the climate pledges the governments of Iraq, Lebanon, Syria and Yemen made in their Nationally Determined Contributions (NDCs), the documents each party to the Paris Agreement submits to say what it will do to cut emissions and adapt to climate change. In this pilot we track 33 pledges, those in each country’s NDC in force.
We compare each pledge with what open sources show, from public datasets and satellite imagery to official documents and media reports, rather than relying on what a government says about itself.
How we choose a pledge
From each NDC we track the targets and measures that can be tested. Options the document lists as indicative, without making them pledges, are summarised in the country profile, not tracked item by item.
Each pledge is one of four kinds: a target, a headline figure such as an emissions cut; a pathway, a named route with a figure, such as a renewables share; an action, a specific or time-bound measure; and an example, a figure the document gives only as an example, which the page labels as such.
We quote each pledge word for word from the document as submitted, with its page and a link. When the document puts the target in a separate table column, we quote that column too. When the document was filed in Arabic, Arabic readers see the original text; when it exists only in English, as Lebanon’s NDC does, they see an unofficial translation by Pax Memoria with the English text, which remains the reference.
We follow each document’s own definition of conditional and unconditional rather than imposing ours. Lebanon’s NDC counts support through loans as part of its unconditional target and keeps the condition for additional grants; Syria’s NDC makes all its pledges conditional on international support. When a document does not say whether a pledge is conditional, we say so and do not infer it.
Status
Country comparisons use the statuses below. The pilot cards also show what has been implemented and what the evidence tells us about pace.
- Not yet assessed
- Extracted from the pledge text; no evidence checked yet.
- Too early to judge
- We researched the pledge, but the NDC is recent and there are not yet enough data on the years after it to judge it. What we found is shown as a baseline or a first step.
- Achieved
- An independent source or measurement shows the full target reached by or before its deadline. An official announcement alone is not enough.
- Progress seen
- Output has been completed since counting began, at no less than half the required pace where a pace can be computed; the target is not yet met.
- Stalled
- Some output was completed, but below half the pace needed to reach the target on time, or the main projects have stopped.
- No visible progress
- A full year after counting began there is no completed output, or published data for the period show no change or a reversal.
- Cannot be verified
- The pledge’s wording cannot be measured at all, or there is no way to measure it: no open data, no projects that can be tracked and no satellite imagery. This is a finding about the pledge or the means of measuring it, not a breach.
Only facts that came after the NDC was submitted move a status, or since its period began if the NDC sets a period that starts before submission, as Yemen’s does. Earlier figures are shown as a baseline, the starting point we measure from.
We use “Too early to judge” only for NDCs submitted from 2024 on, when we have researched the pledge and no data yet cover the years after it. Apart from “Not yet assessed” and “Too early to judge”, no status is given without evidence beyond the pledge text itself.
We measure time from the NDC’s submission to the pledge’s deadline, which is the end of the target year unless the document says otherwise. When the NDC says a measure starts in a given year, the date is the start of that year; when it sets no date, we use the year the NDC runs to and call it the NDC horizon.
Confidence
Every status other than “Not yet assessed” and “Too early to judge” carries a confidence level saying how strong the evidence behind it is.
- Confirmed
- Two or more independent lines of evidence agree.
- Likely
- Strong evidence with one stated gap.
- Possible
- Consistent with the evidence, not proven.
- Unresolved
- Checked and not settled, or cannot be checked.
- Contradicted
- The evidence contradicts the claim.
When a finding rests on having found nothing, confidence is “Possible” at most, because absence of evidence is not proof of absence.
Combined rating rules
The required pace is a straight line from the baseline to the target at its deadline. A pledge is “Progress seen” when completed output reaches at least half that pace up to the review date. We chose half rather than the full pace because large projects arrive in steps.
It is “Stalled” when some output is completed but below half the required pace, or when the main projects have stopped. It is “No visible progress” when a full year has passed since counting began with no completed output, or when published data for the period show no change or a reversal. Neither is given before a full year after counting begins.
An output is completed when a plant runs, an area is planted and checked on imagery, or a law is issued. A project under construction, or one with only an agreement or a tender, is not a completed output.
When the NDC gives no baseline figure, the baseline is what the NDC reports at the start of counting, and progress is what is completed afterwards. A missing baseline figure alone is not a reason to call a pledge “Cannot be verified”.
A pledge is “Achieved” when a source or measurement shows the full target reached by or before its deadline, with confidence “Confirmed” or “Likely”. An official announcement alone, without an independent source or measurement, is not enough, and neither is a good pace towards a distant deadline.
We keep “Cannot be verified” for pledges whose wording cannot be measured at all, such as “strengthen” or “improve” with no indicator, figure or countable output, and for pledges there is no way to measure: no open data, no projects that can be tracked and no satellite imagery. A missing baseline figure or missing national data is not enough while projects can be tracked, so pledges with this status are being re-examined against this rule.
How we monitor each pledge
Each pledge gets a monitoring plan, a description written in advance of how it will be measured, and no pledge gets a status other than “Not yet assessed” without a complete plan. Plans are being written now, country by country, and each is shown on the pledge page under “How we monitor this pledge”.
The plan gives the indicator and its unit, the baseline, the main route we follow the pledge through and its backup routes, what counts as a completed output for that pledge, and the date of the next review.
We follow pledges through published data, project ledgers, satellite imagery, government announcements, open-source investigations on the Pax Memoria model, and field follow-up. Evidence from an investigation or from the field is saved like any other source.
When the baseline, target and deadline are known, the pledge page shows the required path, the straight line from the baseline to the target, with today’s position on it and the value it requires today. For an absolute target, such as an area in hectares, the yearly requirement runs from the start of counting.
When nothing measures the indicator itself, we accept a route that measures something close to it, a proxy, such as households instead of hectares, or one project’s farmers. The plan and the page say so, and confidence is then “Possible” at most.
The published plan contains no names, no internal review details and no questions we send to the field.
International support
Many pledges depend on international support: finance, technology or capacity building from abroad. We track whether that support arrived separately from the pledge’s status: the status describes what happened, and a support line says whether the condition was met.
We do not assess a conditional pledge before checking whether the support it depends on arrived. If the support did not arrive in full, we call the pledge “Stalled” or “No visible progress” only on measured data, satellite imagery or a dataset, never on an absence of news.
What concerns support in each country is written once in its profile; each conditional pledge gets a short line with one of these states.
- Support not yet checked.
- Too early to judge support.
- The required support did not arrive.
- Part of the support arrived.
- The required support arrived.
We use OECD data on climate-related development finance, Green Climate Fund country pages and the support tables of transparency reports. OECD data record a whole project value once it carries a climate marker, so they show where money was marked, not how much of it went to climate.
The overall level
When the 2035 NDC and the first transparency report are both missing, we set the overall level to “stalled” and explain the reason on the country page. This is an observatory rule for the country level. Each pledge keeps its own evidence-based status.
The gauge at the top of each country page sums up its pledges with one needle. Only four statuses sit on it, worst to best: “No visible progress”, “Stalled”, “Progress seen”, “Achieved”. “Cannot be verified”, “Too early to judge” and “Not yet assessed” are left off the gauge, so they are never read as a middle position.
Otherwise, we show a level based on implementation when at least half of the country’s pledges are assessed, with neither “Not yet assessed” nor “Too early to judge” counted as assessed, and at least two can be placed on the scale. The needle sits at the median of the scores on the scale, rounded down. The median of “progress seen” and “no visible progress” is “stalled”, and the text names both statuses.
The caption under the gauge names the level by status. When the pledges on the gauge differ, it says “mixed results” and names the statuses, with no counts, percentages or score.
Procedural obligations
Beside the gauge we show whether the country met two procedural obligations the Paris Agreement words as “shall”: communicating an NDC every five years, and submitting a transparency report every two years, the first by 31 December 2024. A later decision encourages Parties to communicate in 2025 an NDC running to 2035, so we show whether the country submitted an NDC in 2025 or later. The first transparency report is shown as on time, late or not submitted.
National communications are not counted, because the Convention’s decisions word them as “should”; the latest is shown for information only. We set no deadline for 2035 NDCs, as we have not yet verified the exact date.
The Paris Agreement binds Parties to pursue their NDC’s aims through domestic measures; reaching the number itself is not a legal obligation. The wording of these rules on the site is a draft not yet reviewed by a legal specialist; the links lead to the official texts.
Sources
We keep saved copies of the documents and pages we rely on; web pages are saved as text with their title, link and the date we read them. Every pledge quote, every figure we have tied to its source, and every source in the project ledgers is checked automatically: it must appear word for word on the stated page of the saved copy, or the page is not built. Where a PDF page has no usable text layer, we check the quote against the page image by hand and say so.
What governments announce is a source for what the government says; we do not count it as a result until we have checked it. Datasets we use include:
- Ember: yearly electricity data, for generation shares and solar capacity.
- Global Forest Watch: tree cover and tree-cover loss from satellite imagery.
- NASA FIRMS: satellite fire detections.
- International Renewable Energy Agency (IRENA): electricity capacity and generation statistics.
- World Bank: gas-flaring estimates, power-grid losses and the damage assessment for Lebanon.
- OECD: climate-related development finance.
- UN SDG Global Database: proportion of degraded land.
- Lebanon’s National Council for Scientific Research: fire maps.
Project ledgers
Some pledges are broad: no single decision meets them, but dozens of projects, as with a renewables share or the conversion of power plants. For these we keep a project ledger that asks one question, explains why the pledge needs a ledger, and states its counting rule.
Only what is completed and running after the NDC was submitted, or since its period began if that came first, with a saved source, counts towards the total. What was completed before is shown as the starting point and not counted as progress; announced, agreed, under-construction and halted projects are listed separately and not added. Capacity is recorded in megawatts, area in hectares, and sites and projects by number.
For planting, an area officials say was planted is not counted until we have checked it on satellite imagery, and seedling numbers are not counted as area.
One ledger can serve two pledges that depend on the same projects, as with Lebanon’s two renewables targets, 25% unconditional and 30% with additional grants. Each ledger lists the sources we watch, official, local, secondary and datasets, with the reason for watching each.
Under “Government announcements we follow” we show what official bodies announce, as published and with its date, and do not count it until we check it. When a body announced a completion date that has passed without our confirming completion, we say the date has passed, and give the date we will return to it.
Each announcement falls into one of three groups: “kept” when we have confirmed it; “not kept” when the evidence contradicts it, the body amended or withdrew it, or its date passed without our confirming it; and “unresolved” otherwise. A “Latest updates” section at the top of each country’s pledge list shows new announcements, changes in what we have checked and dates that have passed, newest first.
Field evidence and reconciliation with official reports
Some pledges are followed in the field through a network of local journalists working with Pax Memoria. We do not publish their names: what they collect is attributed to “a local journalist working with Pax Memoria”, with the governorate and date, and a journalist’s name is published only with their written consent.
Field reports are kept in an internal data management system, not in the public repository, with each file’s date and digital fingerprint, and location and device data removed from images. The page shows only a safe summary of what was found.
Each monitoring plan generates specific questions for the field, and the team chooses which go to the network after a safety judgement. No one is asked to approach military sites or conflict areas.
A single journalist’s report does not raise confidence above “Possible”. It reaches “Likely” when two independent journalists agree, or a journalist is matched by satellite imagery or an official document. When a researcher from the Pax Memoria team verifies a fact in person, that alone supports “Likely”. “Confirmed” always needs two independent sources, one of them a measurement: satellite imagery or data.
We list every project we find in the project ledger, then reconcile the ledger with the next official report when it is published. A project in the report that we also found is a strong confirmation. A project we found that is not in the report is raised with the body concerned when we send it the assessment for reply. A project in the report that we did not find is searched for, and not counted if it is not found.
We do not wait for the official report to count what we have verified: a project confirmed on imagery or by two independent sources counts at once, and reconciliation raises confidence later.
Review and corrections
Assessments are drafted first, some with automated tools, then reviewed by the Pax Memoria team together before publication; we record the review date.
Before an assessment is published, we send it to the government body concerned for reply and record the date it was sent. We have not yet set a written procedure for the right of reply.
A status changes when new evidence from the period we count appears: we add the evidence with a saved copy, rewrite the assessment, and it stays a draft until the team reviews it. The foot of every page gives the date of the data it was built from.
Limits
Open data capture off-grid solar panels poorly, and they are the largest source of renewable growth in Lebanon and Syria, so reality may be better than the figures show.
The data may not measure what the pledge measures: Lebanon’s target is a share of electricity demand including private generators, while international data measure a share of grid generation.
Field checks and data collection are hard in areas that have seen armed conflict, such as south Lebanon and parts of Syria, so there we rely on satellite imagery and published sources, and do not ask the journalists working with us to approach conflict areas.
Most datasets come out one to two years late; OECD climate-finance data end in 2024. Many pledges in NDCs submitted in 2025 therefore remain “Too early to judge”.
Implementation progress and pace towards the target
LB-03, LB-05 and SY-01 show two findings. For implementation, we identify an output that can be verified after counting starts. Electricity reaching users documents an operating renewable installation. For land restoration, sapling distribution is a preparatory step. We verify the planted plot and its area before counting planting.
For pace, we compare measurements of the pledge indicator with its baseline, target and deadline. Lebanon measures a share of total electricity demand. Syria measures a share of both on-grid and off-grid generation. Pace remains unknown while comparable observations are missing. For Lebanese land restoration, we can track verified hectares each year, but the NDC specifies no area due by a particular date.
During the pilot, country comparisons use the previous combined ratings. Their half-pace threshold applies to the combined rating. Project implementation can be verified before a new national measurement becomes available.