The Users Committee meeting included a presentation by Karen O'Neil and Amy Shelton on a project to implement Dynamic Scheduling (hereafter DS) on the GBT. The stated goal of the project is "to increase the scheduling efficiency of the GBT by allowing the telescope schedule to be optimally matched to both the current weather and the available instrumentation, while retaining observer control of each observation." The basic plan, as we understand it, is to have automated software look at the available project astrid sessions and pick the next project to be scheduled, subject to multiple weighted criteria including sidereal time, scientific ranking of the proposal, weather and equipment and observer availability. Observers currently on-site would be favored in the selection process. The chosen observer would then get a phone call with roughly a 15--30-minute warning, and that observer or his/her delegate would be responsible for ensuring that data acquisition happens correctly and results in acceptable data. An executive summary of the UC reaction to these plans is that, while we recognize that there is anecdotal evidence to indicate that good weather is not being used optimally for high-frequency observing, we have not been presented with the combined observing, data quality and weather statistics that demonstrate the severity of the problem and the expected gains from the implementation of full DS. At the end of this report we list a number of issues that we believe need to be more fully considered before the current DS plans progress beyond the Sept. 2007 test observations. At the moment, a crude form of DS is in place, in which two identical parallel schedules (one high-frequency, one low-frequency) are prepared for days spaced two days apart. Based on the weather conditions on the first day, the high-frequency observer decides whether to use the time on the first day or gamble that the second day will be better, and the low-frequency observer uses whichever day is left over. While slightly inconvenient for the low-frequency observers, this scheme has consistently resulted in perfectly good data, at least at low frequencies. However, high frequency observations are often not performed under acceptable weather conditions. The perceived need for and advantage of full DS rest on a number of arguments: - That the current partial DS scheme results in many good-weather days being used for low-frequency observing, and in many bad-weather days yielding bad data at high frequencies. While we hear quite a bit of anecdotal evidence for this, and recognize that this results in both poor science return and large travel/financial burdens on (especially non-US) high-frequency observers, apparently statistics have not been compiled to support this argument and may be difficult to derive. We believe it is critically important to have these statistics in hand in order to evaluate whether DS will salvage enough high-frequency hours to be worth the investment of many FTEs. - That high-frequency projects are not being completed. Again, this is anecdotally supported and statistics are not available at the moment, although these might be easier to compile than the first set. We note, however, that at most telescopes (both optical and longer-wavelength) the observers assume some weather risk and good-weather time and project completion are not guaranteed. Average annual weather statistics could be a guideline as to how much good-weather time to approve. - That all GBT projects can be run using Astrid. This is probably true. Many also require further interaction with specific backends. - That GB does not have sufficient scientific staff to perform queue-style observing as is done at Gemini or the JCMT. This is certainly true, although it neglects the fact that at JCMT at least, visiting observers do some or most of the queue observing. - That no GBT projects can be reliably run without human intervention to check levels and ensure data quality, or at a maximum with a couple of simple things for the operator or a potential visiting queue observer to check. At least anecdotally, the UC has reason to believe this is not true and that some fraction of projects could be run quasi-automatically. To determine the likely utility of full DS, it makes sense to consider the current scientific usage and return of the telescope, as well as the requirements of typical projects, versus the resources and effort needed to create the full DS system. For telescope usage, we can look at a couple of sample months, 2007 January and 2006 July. According to the observing summary report we were provided with, these days 80% or more of the time (540+ hours/month) is devoted to astronomy. 2007 January: - 634.25 astronomy hours - 400 MHz through C-band: 358.5 hours - 400 MHz through X-band: 431.25 hours - > 10 GHz: 203 hours 2006 July: - 524 astronomy hours - 300 MHz through C-band: 404 hours - 300 MHz through X-band: 429 hours - > 10 GHz: 95 hours In July, there was, not surprisingly, no usage of receivers above the various K-bands, and these comprised less than 20% of the astronomy time. In winter, high-frequency observing accounted for just over 30% of the astronomy time, or 45% if X-band is counted in the "high-frequency" category. We realize, of course, that the desired weather conditions depend strongly on the frequency requested and the type of observation, and recognize that the above classifications are very rough. In terms of scientific return, high-quality publications, including those presented to the UC as highlights of GB output, have resulted from the full range of receivers, and the GBT's unique capabilities have contributed at all frequencies. Therefore there are certainly no grounds to categorically displace either high- or low-frequency observations in the overall scheduling planning. It comes down to trying to optimize both the weather requirements of the high-frequency work with the generally more rigid scheduling requirements of many of the low-frequency observations (eg pulsars and VLBI). Overall we estimate that the goal of DS is to schedule optimally roughly 40% of the telescope time in the winter months, and something less than 20% in the summer. Knowledge of weather statistics in the winter months would be helpful in determining if weather suitable for high-frequency observing actually occurs ~40% or more of the time; does the observatory have these statistics? A question is whether the remaining 60% of the winter observations can effectively be dynamically scheduled to accommodate the high-frequency work. A significant fraction probably can be, and the process would be made a lot more palatable if the simplest projects can be allowed to run independently of the observers (see below). The pulsar observations could pose a problem for DS. Based on the combined projects of Scott Ransom and Ingrid Stairs (probably about ~80--90% of GBT pulsar projects list at least one of these two people as an author) in the period 2007 January through April, it appears that about 100 hours per month are pulsar observations, which is nearly 20% of the astronomy time. The vast majority of these are low-frequency observations. A large fraction, moreover, are timing observations and need to be scheduled within fixed windows (typically monthly ± a couple of days) in order to achieve their science goals. The plan is for DS to allow for such windowed observations (although the planned 1-week allowance may at times not be stringent enough, for example if it results in an ongoing 5-3-5-3-week-spacing pattern). But even with the requirement for an observation to fall within a week, inevitably some good weather will end up being used for pulsar observations. Overall it seems unlikely that it will be possible to use all the good weather for high-frequency observing. A final but extremely important consideration is the planned requirement for the observer to be responsible for the data quality, which translates into a requirement for an observer always to be ready to start up a project with about 15--30 minutes' notice. We are concerned that time will easily and often be lost in trying to track down observers; having the telescope sit idle because of such communication problems would be completely unacceptable. We repeat our opinion that at least some fraction of all projects could probably be run either in automated fashion or with minimal verification by the operators, but a large number of projects will still need to be run by the observers themselves. The need to be almost perpetually on-call, at least around certain LST ranges, will be an enormous burden on all observers, high- and low-frequency alike. Rearranging one's schedule at the last minute once in a while is tolerable, but this becomes a significant annoyance when one is involved with 10 or more observing sessions in a month (which is entirely realistic) and must either be available or arrange for a substitute every few days. It will be very difficult to make this compatible with faculty teaching, student courses and other professional obligations. We note that DS observations can have effects well outside the actual timeframe when the observations occur: for example, unpredictable phone-call requests to start multi-hour observations at 2 am will be detrimental to 9 am classes taught by the observer, particularly if this happens many times in a month. Other concerns include family life and health issues, as well as difficulties for observers who either do not own cell phones on principle or who may be traveling in parts of the world where their cell phones do not work. Some ideas the UC has considered to address this set of practical concerns include reducing the timeframe for windowed observations to 2 or 3 days, or adopting a tiered system in which A-ranked proposals get guaranteed fixed timeslots, B-ranked projects are likely to be windowed or subject to full DS, etc., in a similar manner to what was implemented at the VLA following the 2004 UC report. The implementation of DS should not end up discouraging observers from proposing to use the GBT! In summary, while we recognize that better use could probably be made of the good weather in GB, we are hesitant to recommend "full steam ahead" on the currently proposed DS scheme. We think the following questions should be examined more carefully before proceeding beyond the Sept. 2007 tests: 1) Do the statistics on weather history and project scheduling support a) that much good weather time is going to low-frequency projects, b) that there is enough good-weather time available to hope that all accepted high-frequency projects could benefit, and c) that the recovered time is worth the cost to the observatory and the user community? 2) Can windowing be reduced to a 2- or 3-day interval for the sake of the professional obligations, health and sanity of frequent observers? Will this restriction, plus VLBI, leave a large enough fraction of good-weather time available to make DS on the remaining projects worthwhile? Alternatively, would a VLA-style tiered system be more feasible? 3) Can a list of criteria be devised for projects that could be run either automatically or with only operator checks? Allowing some projects to run in this mode would make life significantly easier for many observers. We note that at Arecibo, many projects are in fact run almost entirely by the operators. This is particularly true for low-frequency projects, which would not benefit from DS anyway. 4) Are the hoped-for gains from DS really worth the very large investment of person-power that has been and is planned to be applied to this project? Is expending several person-years' worth of effort worth an extra 5% of total time per winter (ie about 100 hours) spent at high frequencies in good weather? Or are the projected gains much larger than that, and more able to justify the resources used?