SOSX Research

Research method is a taught discipline and it has nothing to do with us. Declare the position you are arguing from (the ontology and epistemology underneath the question, and whether you are reasoning inductively or deductively) because a method inherits a tradition whether or not you name it. Search systematically: a protocol, seeds, forward and backward snowballing, and a record of what you searched and where you stopped. Verify every reference against the source itself; a citation is earned, not remembered. Make gap claims that a reviewer can test, which means pointing at the search that failed to fill the gap rather than asserting silence. Where the literature genuinely will not answer, elicit expertise in a structured way, the Delphi method Dalkey and Helmer set out in 1963, and record the elicitation as provenance rather than pretending a number was found. Then plan the work with a critical path and honest float, and hand the whole thing to somebody hostile before a panel gets it.

All of this is done by hand, every day, by people who are good at it. The cost is time and it lands on the same people twice: once building the map, and again keeping it current as the field moves under it. The expensive part is not drawing the map: it is building it and keeping it true.

Which is the honest frame for what follows: the discipline above is the product of centuries of scholarship, and what the platform gives you is a way to hold all of it at once. The research stays yours to direct and yours to defend, and SOSX does the reading, the mapping and the evidence chasing underneath it, so the weeks of searching stop being the price of the work.

Every claim cites a catalogued source by stable ID; analyses are unit-aware and carry uncertainty markers; judgement, formal proof and computation are kept visibly separate. SOSX Research points that same discipline at a literature: papers and concepts as a connected, typed network, contradictions surfaced rather than smoothed over, and gap claims that can point to the search that failed to fill them.

Related work: a paper on the SOSX design has been submitted to a peer-reviewed journal.

Across the two hundred papers this programme sits on, where do the findings actually contradict each other, and which of our assumptions is load-bearing? Weeks of close reading go into answering that by hand, and the answer decays as the field moves. With a mapped field it is a standing property: inspectable, cited, re-checkable as new work lands.

We hold open questions of our own, and we would welcome collaboration on them: higher-dimensional network modelling for emergent properties, fluid-dynamics analogies for parameter flows, and graph-optimisation methods beyond our current Bayesian and Monte-Carlo modes.