Successful and productive medical informatics research is evidently a combination of luck, creative art, and science, but some researchers focus too much on building computer artefacts and writing anecdotal reports of their experience. They need to adopt a less technology-fixated approach, be willing to evaluate their systems and publish failures as well as successes, and attempt to generalise their results as hypotheses for others to test. It does appear that medical informatics is a distinct discipline, and one based on scientific principles, but it is less clear whether these principles originate within the discipline or elsewhere. If elsewhere, it is usually unclear whether their validity has been tested with the atypical information, decisions and context that medicine represents. This article has presented some criteria for judging such scientific principles, and described a process which would lead to such principles, if they exist, being uncovered more rapidly. If our discipline is to thrive and take root in firm ground, such activities need to be taken seriously by all, otherwise we could end up building edifices on sand.