Deliberate Workflows. Why Explicit Steps Define the SHAPE of Modern Research Data Processing

Research data are rarely published as they were recorded. In between lies processing (cleaning, converting, combining, analysing, plotting), and that is where results are made and most easily lost. Steps live in someone's head, files are handed on without saying what they contain, manual fixes go unrecorded, and figures can no longer be traced back to their data. FAIR describes what shared data should be. ROOT (doi:10.5281/zenodo.19051651) asks which tools can carry them for decades. SHAPE addresses the work in between. SHAPE is a framework of five principles, each a question to ask of any workflow at every step: Structure Stages: can you explain what each part does? Harmonize Handovers: can the next tool or person use this correctly? Automate Actions: can this be repeated without reconstructing manual actions? Propagate Provenance: can you trace every step that led to this result? Evaluate Explicitly: how would you notice that something went wrong? The principles are not a sequence: each applies in every step. The poster demonstrates SHAPE on its own making. It is built from a single Org-mode source with GNU Emacs, Org Babel, LuaLaTeX and GNU Make. Its central figure is generated from tables in that source, and every code excerpt shown is the block the build actually runs. Each build step records who or what ran it, when, and the SHA-256 checksum of every file it used and made, and each entry is linked to the history before it. This history is embedded in the PDF and included in the source bundle. Automated checks stop the build if a principle acts in only one step, if the recorded history is broken, or if text overlaps on the page. This record contains the A0 poster (PDF) and a source bundle that rebuilds it with one make command. The bundle holds the Org source, Makefile, logos, build history, validation results and checksums. Text, figures and code are licensed under CC BY 4.0; the logos are excluded.