We need an automated follow-up observation marketplace. A system where an observation request is submitted and automatically matched to an available resource based on the properties of a request and the "price" (in some form of credits) the requestor is willing to pay and the…
@mjuric
What is the cumulative cost of 1 astronomical image including the full cost accounting of the major telescope from conception to design and construction, through operations, energy expenditure, and science extraction, to decommissioning and environmental restoration? Is it known?
@SzabiMarka
It's very telescope-dependent. For Rubin it's ~$500/image (for design through operations). For smaller telescopes could be 10x less.
This is the cost, not the value. Value is tougher to estimate; probably correlates better with (say) citations and use of the telescope's data.
@megschwamb
I got the impression AEON right now allows to more easily request observations from a particular participating telescope, rather than any telescope capable of taking it. I may be very wrong (can't get to the paper right now, it's paywalled). Will ask Rachel tomorrow.
@astrokiwi
Interesting -- it's the talk about in-kinds at LSST
@Europe
that made me think about this as I couldn't quite make out how the contributed resources would be matched to end-users (in a potentially very oversubscribed situation).
@mjuric
The primary challenge is not technological but sociological. In a functioning marketplace you need incentives and without $ exchange (star bucks?) there is no universal value (ie. authorship, reciprocity on a different aperture). Disincentives to bad actors also needs…
@profjsb
Agree on all points (esp. the sociology issue), but there are already moves in this direction in other areas.
E.g. the NSF switched smaller HPC resource allocations (, the successor to XSEDE) to a "credit" system. Users are not allocated cycles, but…
@mjuric
This is something we are attempting to achieve with the Recommender Engine For Intelligent Transient Tracking (REFITT). But instead of human decisions, an automated AI system makes time-critical parameter inferences of transients, judges priority, makes recommendations, and…
@mjuric
The problem is complex. There were various attempts in the past, most of which failed so far. But that certainly doesn't mean it can be achieved.