Varying alpha, blinding, and bias in existing measurements

The high resolution spectrograph ESPRESSO on the VLT allows measurements of fundamental constants at unprecedented precision and hence enables tests for spacetime variations predicted by some theories. In a series recent papers, we developed optimal analysis procedures that both exposes eliminates subjectivity bias in previous quasar absorption system measurements. this paper analyse spectrum z_{abs}=1.15 towards HE0515-4414. Our goal here is not to provide new unbiased measurement fine structure constant, alpha, (that will be done separately). Rather, it carefully examine impact blinding applied same data Murphy (2022) prior that, several other analyses. To do use supercomputer Monte Carlo AI calculations generate large number independently constructed models complex. Each model obtained using AI-VPFIT, with alpha fixed until "final" obtained, which point then released as free parameter one final optimisation. results show "measured" value systematically biased initially-fixed i.e. process produces meaningless implication straightforward: avoid bias, all future must include from beginning modelling process.