You might be interested in older work by philosopher Larry Laudan. He posited mutual positive feedback (or co-evolution) between: a) better data; b) better models or theories; and c) better instruments. Basically agreeing with you.
Likewise, he also noted that constructing better instruments is theory-dependent, as better theory allows a more complete understanding of sources of error (not to mention entirely new technologies).
Laudan called his a "tri-partite" model, as I recall.
Likewise, he also noted that constructing better instruments is theory-dependent, as better theory allows a more complete understanding of sources of error (not to mention entirely new technologies).
Laudan called his a "tri-partite" model, as I recall.