The HST Proposal Selection Process
To write a successful HST proposal it is useful to have some insight
into the process by which proposals are reviewed and selected. A
number of recent STScI Newsletter articles may therefore be of
interest. Below we include the full text of the article
by Meg Urry [email@example.com],
Head of the STScI Science Program Selection Office. Previous articles
on the same subject include
How the New Cycle 9 Review Process Worked
There were five key changes in the proposal review process for Cycle
9. With the review now completed, we have received extensive feedback
from the Cycle 9 panels and TAC (Telescope Allocation Committee)
indicating the changes were very successful, hence we plan to repeat
the new process (with minor changes) in Cycle 10. Here we summarize
the process changes, followed by some advice for proposers in future
Goals, Changes, and Outcome
- Goal 1: Increase the fraction of large observing programs, to the
10-30% levels that have been recommended by external advisory
Changes: The role of the TAC was redefined to be review of
Large proposals (100 orbits or more), for which up to 1000 orbits were
available. Medium-sized proposals (15-99 orbits) were encouraged via
orbit subsidies in the review process. To encourage more submissions,
these new opportunities were heavily advertised in the Call for
Proposals, the STScI Newsletter, and elsewhere.
Results: More than 15
times as many Large proposals were submitted as in Cycle 8,
constituting 1/4 of proposed orbits rather than a few percent. The TAC
recommended a total of six Large programs, for ~1/4 of the total
allocated orbits. Requests for medium-sized allocations doubled,
constituting ~60% of the requested orbits and a similar percentage of
the selected orbits. Thus the acceptance rate was largely independent
of proposal size.
- Goal 2: Better determine the balance among
Changes: We greatly broadened the
scientific focus of each review panel and charged them with allocating
100% of the Regular (<99 orbit) proposals, so that science balance
could be achieved within individual panels. In preparation for this
change, proposers were explicitly told to address the overall
importance of their program to astronomy (the "big picture").
All proposals were evaluated via in-depth discussions by experts,
rather than one third (the gray-area proposals) being discussed by a
much broader and overly large TAC. The Cycle 9 panels were highly
supportive of this change. Interestingly, quite a few panelists
admitted to premeeting apprehensions about the breadth of expertise
required, but by the end they embraced the new process. With redundant
panels for each broad area (see next goal), there were effectively two
independent determinations of science priorities per cycle; in a few
cases, mirror panels came up with quite different mixes of science,
making the new process more robust against the vagaries of peer
- Goal 3: Minimize the impact of conflicts of interest while still
allowing experienced HST users to participate in the review.
We assigned two panels to each broad science area (with the exception
of the Solar System panel, for which there are too few proposals), so
that a panelist's proposals could be reviewed by the other panel.
Results: The acceptance rate was the same for PIs who served as
reviewers and those outside the process. There were also dramatically
fewer instances of panelists having to recuse themselves because of
conflicts of interest, leading to a more consistent review involving a
larger fraction of panel expertise.
- Goal 4: Enable exciting science
with fast ToOs and multiwavelength HST+Chandra proposals.
increased the number of fast ToOs allowed, and reduced the minimum
activation time proposers could request (with actual implementation to
be driven by the proposed science).
We instituted the first joint multiwavelength opportunity, for
Chandra and HST, awarding up to 400 ksec of Chandra time and giving
the Chandra project 100 orbits of HST time for their next
Results: Several of the top rated Cycle 9 proposals involved
fast ToOs and Chandra-coordinated science, attesting to the importance
of these opportunities.
- Goal 5: Reduce the burden on the astronomical
Changes:We halved the number of panels, and thus panelists
(and cost). We implemented a triage process to keep the burden on
individual reviewers manageable, after verifying, via tests on the
review databases from previous cycles, that it accurately identified
the bottom 1/3 of proposals.
Results: Only ~75 reviewers served in
Cycle 9, compared to ~150 in Cycle 8. The panels completed their work
just as quickly, in part because of the triage approach, which they
Advice for Future Cycles
The new HST review
process having been highly successful, we plan to continue it in
future cycles. This means proposers should keep in mind several key
- First, don't be shy about submitting Large proposals; they
have the same chance of success as any other proposal (a 1 in 5
chance). The TAC discusses each Large proposal in considerable detail,
so winning proposals are not only of compelling scientific interest,
they are well thought out and involve expert teams.
- Second, make the
effort to address the non-expert. Most successful proposals set the
context with sufficient background information, and clearly described
the importance of the investigation to all astronomy (the "big
picture"). Since this was an explicit criterion for evaluation,
proposals could be downgraded for failing to include it.
- Third, don't
pad your request for time: fewer than 10% of the approved proposals
were cut (and those largely to avoid duplicate observations), so
proposers either got what they asked for or were rejected.
The main advice?
Start from the science, write a compelling story for your fellow
astronomers, ask for the resources you really need, and submit those
proposals whatever their size.
With such a high oversubscription (1/6 by orbits), many truly
excellent proposals had to be rejected.
Remember that rejected PIs are in good company, and that many of these
proposals, possibly revised according to comments from the panel or
TAC, may succeed in future years.
Cycle 10 Announcement Page
Last updated: June 9, 2000