Single pane of reality: why the map, not the lake, is the interoperability layer

Twenty years of big data gave every organisation more data than it had ever held, and no better decisions. The industry answer has been the single pane of glass: one more dashboard above the dashboards, fed by one more integration. It fails for a structural reason. A pane of glass shows what the systems say. It cannot show what is actually there.

The alternative is to put reality in the middle instead of the systems. A big data map is not a warehouse of records. It is a spatial and semantic model of the thing itself, and every system attaches to it as an attribute rather than claiming to be the source of truth. The building. The vessel. The furnace. The tunnel. The person. Once that map exists, interoperability stops being a point-to-point integration problem and becomes a question of what attaches where, and who is permitted to see it.

That reframing collapses three conversations into one. Semantic interoperability, data spaces and data sovereignty are the same question asked three times: what is the shared object, what attaches to it, and who decides who sees what. Answer it once against a map and all three have an answer. Answer it three times against a lake and you get three integration projects.

The pattern does not change when the domain does. In steel production, a twin of the plant where process, energy and maintenance read the same model. Underground, in rock, where an installation that cannot be surveyed from the air still has to be understood as one object. In real estate, an edge and gateway layer serving a portfolio of around five hundred buildings, built open and modular so any component could be replaced. In a Swedish municipality, an EU-funded innovation partnership with more than three hundred wireless sensors and over a hundred actuators, where the control loop stayed local in the building and operations, finance and leadership saw the same picture of the same reality for the first time. In energy, resilience twins for reconstruction work and for one of Europe's largest renewable producers. In manufacturing, master data work where the question was never the data but the ownership of it: which system owns the record, who may change it, who approves the change. In health, a platform where the individual owns their own data and the experts go to the data rather than the other way around.

Different industries, different regulators, one structure. The shared structure is the argument, and it is why this belongs in a cross-sector conversation rather than a vertical one.

The method behind it is published openly with a DOI. SMILE is built on the extreme collaboration practice that NASA JPL used to take early mission design from nine months to about three weeks, with the engineering itself fitting into roughly nine hours of facilitated sessions. The gain came from proximity and a shared live model, not from automation. The ordering that follows is the argument in one line: goal, benefit and impact first, then decisions, then insights, then information. Data last.

Most data strategies are written in exactly the opposite order. That is why they produce lakes instead of maps.

Written for the European Big Data Value Forum 2026 Speakers Corner in Galway, run by BDVA and the European Commission. The three-minute slot on the exhibition-hall stage is open to anyone already in the room, and the pitch was submitted after the 23 August deadline had closed. The argument stands on its own either way.