Learn how Intel Mapper assigns confidence scores to conflict events and why these ratings matter for understanding Ukraine and Middle East coverage.
Every event plotted on Intel Mapper—from a reported artillery strike to a claimed drone attack—carries a confidence score. This rating reflects how reliably the event is documented based on available evidence. Rather than treating all reports equally, confidence scores create a hierarchy of certainty, allowing readers to distinguish between well-verified incidents and preliminary reports that may change as new information emerges.
In fast-moving conflicts, the fog of war is real. Combat unfolds across vast territories, often in remote or inaccessible areas. Multiple parties have incentives to claim victories, deny losses, or exaggerate impact. Journalists, militaries, and analysts are working with incomplete information in real time. Confidence scores acknowledge this reality and give readers a transparent framework for evaluating what they're seeing on the map.
The purpose of confidence scoring is not to filter out uncertain information entirely—sometimes preliminary reports are the first available indication that something significant has happened. Rather, it is to label information honestly, allowing each reader to decide what weight to give different events based on their own needs and risk tolerance.
Intel Mapper assigns confidence scores based on the types and consistency of sources reporting an event. A high-confidence event typically has corroboration from multiple independent sources, such as official military statements from both sides, geolocated video evidence, humanitarian organizations on the ground, or respected independent open-source intelligence networks. When a Ukrainian official announces a strike and Russian sources subsequently acknowledge damage in the same location, confidence rises.
Medium-confidence events rest on fewer or less diverse sources. A single military claim, a social media post from someone with a credible track record, or reporting from established news organizations investigating the incident may all justify a medium rating. These events are plausibly documented but lack the redundancy or independent verification of high-confidence reports.
Low-confidence events include unverified claims, early reports that lack secondary sourcing, or incidents described only by parties with direct interest in the outcome. A combatant's initial claim of a successful strike, before any independent verification, would typically start at low confidence. As corroborating evidence accumulates—satellite imagery, eyewitness accounts, damage assessments—the score can be upgraded.
Consider a hypothetical Ukrainian claim that a strike damaged Russian air defences in occupied territory. If only the Ukrainian military has reported this, and no independent observers have yet visited the site or posted verification, the event might appear on Intel Mapper with a low or medium confidence score. This tells readers: the claim exists, it may be true, but treat it as preliminary until further evidence arrives.
Hours or days later, satellite imagery might emerge showing damage consistent with the claim. Independent open-source analysts might geolocate the site and confirm the structure type. International news organizations might obtain interviews with witnesses. As this evidence accumulates, the confidence score updates upward. A reader checking Intel Mapper days after an event may see the same incident with a notably higher confidence score than when it first appeared, reflecting the evolution of available information.
Confidence scores also vary by event type. A specific territorial claim—'Unit X now controls Village Y'—can be verified through geolocated images, satellite data, or official statements. In contrast, casualty figures or equipment loss counts are inherently harder to verify independently; multiple estimates may exist, and confidence scores reflect this structural uncertainty. A report of a destroyed vehicle might reach high confidence quickly; a claim about total losses in a week-long operation may remain medium confidence for much longer.
It is essential to understand the limits of confidence scoring. A high-confidence score means an event is well-documented by available evidence; it does not mean the event is accurately understood in all its details. A strike might be confirmed to have occurred, but its precise impact, the weapons used, or the strategic significance may still be debated among analysts. Confidence scores rate source agreement and evidentiary quality, not the completeness of understanding.
Confidence scores also reflect the state of available evidence at a particular moment. They are not permanent verdicts. As new information emerges—including information that contradicts earlier reports—scores can be adjusted downward. A widely reported incident might later be found to be a misidentification, a recycled image from an earlier event, or propaganda. Responsible maps update confidence ratings when evidence warrants revision, even if this means acknowledging earlier errors.
Furthermore, confidence scoring cannot account for deliberate disinformation at scale. If multiple reliable-seeming sources all report the same false claim, confidence scores will rise based on source agreement. Over time, independent verification typically uncovers such problems, but in the immediate aftermath of an event, coordinated false reporting can temporarily inflate confidence ratings. This is why readers should always ask not just 'how confident are sources?' but also 'what do the sources actually know?'
Some readers might prefer maps that show only 'verified' events, filtering out anything preliminary or uncertain. Others might want to see all claims, regardless of evidence, to monitor what each side is saying. Intel Mapper's confidence score approach splits the difference: it shows the full landscape of reporting while labeling the reliability of each element. This serves readers better than either extreme.
Confidence scoring also reflects professional practice in intelligence analysis and journalism. Experienced analysts routinely distinguish between confirmed reports and working hypotheses. When a defense ministry claims an outcome, that claim is newsworthy and worth reporting—but responsible outlets note that it has not yet been independently verified. Confidence scores bring this discipline to live maps, making the reasoning transparent rather than hidden in editorial judgment.
For ongoing conflicts, this honesty about uncertainty is particularly valuable. Readers checking Intel Mapper over weeks or months see not just what happened, but how the evidence base evolved. Early low-confidence reports sometimes prove prescient; preliminary claims sometimes turn out to be false. Watching these scores change teaches readers how information becomes reliable in real time.
When you consult Intel Mapper, start by checking the confidence score of events relevant to your question. If you are writing a news story or making a decision, high-confidence events provide the most reliable foundation. If you are tracking the broader conflict, medium-confidence events offer valuable context. Low-confidence events deserve attention too—they often indicate where events are unfolding fastest or where information remains most contested.
Pay attention to patterns, not just individual scores. If multiple low-confidence reports converge on the same claim from different sources, that convergence itself becomes evidence, even if individual sources remain uncertain. Conversely, be cautious if a high-confidence score rests on a single type of source, such as geolocated images from one side only. Diversity of sources matters as much as the number of sources.
Remember that confidence scores are a tool for informed reading, not a substitute for critical thinking. Always ask: What does this source actually know? What incentives does this actor have? What would prove this claim wrong? A confidence score tells you how much agreement and evidence exists; your own judgment must decide what weight to give that evidence for your specific purpose.
Confidence scores on Intel Mapper are a transparent way of saying: 'Here is what we know, here is how we know it, and here is how sure we should be.' High-confidence events rest on multiple independent sources and corroboration; medium-confidence events have some verification but less redundancy; low-confidence events are preliminary or contested. When reading the map, use confidence scores as a filter—prioritize high-confidence events for critical decisions, but do not ignore lower scores if they seem relevant. Most importantly, remember that scores reflect available evidence at a moment in time. As conflicts evolve and information accumulates, scores change. Watching that change teaches you how reliability emerges in real time.