This map tries to help answer one question: "where do I fly?" — whether you are asking it after a filament dropped you into an unfamiliar system and you want to scan, or you are heading out to hunt solo in the next region.
From statistics gathered off ESI and the killboards, the map tries to predict what is going on in a system "now" (the natural lag of that data is an hour), over a week, over a month, and even within the stretch of the day you care about — which matters in EVE, given how its activity rolls through the time zones. And through presets for activities, it tries to show how well the situation in a system suits one line of work or another.
I have tried to leave you as many knobs as I could, but the presets alone will do to start with. The coefficients are still being tuned and the scores will still change: treat a score as a reason to look closer at a system with the statistics in hand, not as a promise of El Dorado.
Three steps, and the first one decides almost everything.
A number from 0 to 100, and it is absolute: a system scores the same whichever point you ask from. That is why a result list crossing a region boundary is directly comparable — 55 in Domain and 55 in Delve mean the same thing.
It adds up like this: each criterion turns its raw value into a share between zero and one along a calibrated curve, the shares are summed by weight, and the sum is multiplied by the multipliers. A criterion with no data for that system is dropped together with its weight rather than counted as zero: missing data is not the same as a bad reading. That is why highsec has no "occupancy vs sec norm" — people rat everywhere there, and about equally, so there is nothing to tell apart.
This is the main thing to understand about the left panel.
WEIGHT — the criterion's share of the score. Counted against the sum of all enabled weights, not against a hundred: 30 out of a total 100 gives a third of the score. Zero keeps it out of the sum.
MULTIPLIER — a penalty applied to the whole score, not to a share of it. 0 — no penalty. 1 — a bad criterion zeroes the score entirely, however good everything else is. Values in between cut partially: the presets set 0.7 on incursions and 0.5 on faction warfare, so a system under an incursion loses 70% of its score but does not vanish from the list.
The difference is simple: weight answers "how much does this matter", multiplier answers "is it worth flying there at all if this one is bad".
And a third thing the numbers hide: which way a criterion counts as better. "Ship kills" in the Scan preset means "the fewer the better", and in Gang it means exactly the opposite, "someone to fight". So every row carries a mark: ↓ less is better, ↑ more is better, ∧ the middle is best. On five criteria the mark sits in a box and can be clicked, flipping that criterion onto its second, equally measured curve: ship kills, gate jumps, occupancy against the band norm, neighbour occupancy and gate count. A dim mark with no box only reports: no reverse curve was ever measured for that one, and the map will not invent one.
What the numbers do not tell you is which end of the scale is the bad one. The curve knows, and for Truesec you can see it: under that criterion's row sits a choice of "highsec / lowsec / null-sec is better". Take the Scan preset: Truesec there is 0 and 0.5 — it scores nothing at all, yet it cuts the total, by 40% in highsec and hardly at all in null, because the tier of relic and data sites follows the security band. "As in the preset", in that same list, means the criterion carries a curve of its own, not a class one, and nobody asked to swap it.
Scan — where to hack data and relic sites in peace: few jumps and kills, nobody ratting, a dead end beats a crossroads. Null-sec depth does not matter here: relics are the same in all of null, and no NPCs guard them.
Abyss — a staging point for filaments: quiet highsec with a station and weak through traffic. Pod kills carry the most weight — those are the gankers waiting at the exit of a trace.
Green — a leg of a route for running anomalies for escalations and officers: deep truesec, occupancy below the norm for its band, quiet neighbours.
Solo — a lone ratter you can still catch: occupancy above the norm for its band, quiet neighbours, no hunters here before you. Highsec goes dark entirely — CONCORD answers aggression there.
Gang — a leg for a small fleet: targets close to each other, live neighbours, traffic and crossroads instead of dead ends. Ship kills count in your favour here: it means there is someone to fight.
The "+" button asks for a name and saves the current set of weights as your own preset — it stays in your browser and appears in the same row.
Activity in EVE depends on the time of day more than on anything else. The period is what to average over: "now" is the last hour, then 7 and 30 days. The UTC hour window is your prime time: a system that is empty during your evenings but packed during someone else's suits you, and an average over the whole day will not show that.
The score breakdown shows, per criterion, the raw value and how many points it contributed out of its weight. Below it is an hour of killmail intel from EVE-KILL, and this is the most useful part of the page:
The bubbled-camp flag is worth a glance on its own: it means you will not jump in and warp away.
Colouring: by security, by score, or both halves of a node at once. The score scale offers five palettes, which is a matter of taste. Node size can be tied to a metric (jumps, kills), so traffic reads at a glance without reading numbers. "Dim below score" clears everything under a threshold off the map.
A zero kill count is missing data, not safety. A covert ops may be sitting there having killed nobody and been killed by nobody: no killmails exist for it.
Wormholes are not on the map at all — known space only. Metrics are hourly and arrive from ESI with a delay; the age of the data the page states itself, in the footer. And once more: the coefficients are being tuned, and the scores will change.