Passage
Some instruments cannot be built twice. A detector built to study very rare events costs more than any single laboratory can commit, takes a decade to design and lasts perhaps twenty years, so the laboratories that want one form a consortium: they pool money, staff and, above all, time on the finished apparatus. What each member acquires is not a share of the instrument but a claim on its schedule.
Consortia of this kind are rarely funded in one grant. Money is released in tranches, and each tranche is conditional on the previous stage having produced what the proposal said it would. The arrangement is usually described as a device for limiting a funder's exposure, and as far as it goes that is accurate. It is also incomplete, because the staging is partly forced by the apparatus. A subsystem cannot be tested until the subsystem it depends on is running, so a programme of this kind would have to proceed in stages whether or not anyone were paying for it in stages.
The consequences for the work itself are less often noticed. A group that must report at the end of each stage will prefer, among questions of roughly equal interest, the one that yields a reportable intermediate result; and the order in which questions are taken up comes to follow the release schedule rather than the order in which they arose. Members learn to describe partial findings in a form that satisfies a review without foreclosing the analysis, which is a skill, and not a trivial one, but it is a skill in reporting rather than in measurement.
It would be easy to draw a gloomy conclusion from this, and too quick. The same arrangement disciplines a programme that might otherwise pursue a question for a decade without producing anything a reviewer could weigh, and consortium members themselves often defend it on that ground. What the arrangement does not do is what its critics and its defenders both tend to assume: it does not settle which questions are worth asking. It settles the order in which they are asked and the form in which answers are first presented. Whether a sequence eventually changes the questions is a further matter, on which nobody yet has evidence.
