Budgeting for colloquia
From Low Energy Astrophysics
Are colloquia good places to look at when trying to reduce energy consumption? Many of us feel that it is important to be able to "smell" the speaker, and certainly having a colleague come to visit can be an immensely productive activity. However, there are many seminars that either do not provide good value to either the majority of the audience or the speaker, and these can come at great cost (such as flying a busy person in for just one day).
How much does a seminar cost? Below is a toy model, with some adjustable parameters, for seminar organisers to play with - we need only consider the cost per seminar for budgeting purposes:
An institution with N members invites a speaker to travel D miles to give a colloquium: Energy used in travel ~ 1.5 kWh/mile * 2D ~ 3D kWh assuming the speaker either flies or drives on their own (the usual state of affairs, and the energy cost per mile is about the same).
How can we relate this to the host's seminar "budget"? Well, with one colloquium per 7-day period the contribution the colloquium makes to each institute member's daily energy budget is
Energy per host per day = 3D / 7N
where it seems appropriate to divide by all members of the host institute (this is how the financial budget works).
Suppose N = 100 and D = 3000 miles (roughly Princeton to Stanford): then the cost per host would be ~13kWh/day, which is in the same ballpark as a 10-mile round-trip solo daily commute. Its about 10% of the total professional astronomy budget we computed in the white paper. Having 100% distant speakers would make for a crazy seminar schedule though - most groups are financially restricted to spend less on their seminar program than this! A more sensible mean distance for a speaker to travel might be more like 300 miles, so the cost of seminars might be more like 1.3 kWh/day per astronomer, which is about 1% of their overall professional budget.
Note that this cost could be reduced greatly by only having local speakers, and effectively reduced to zero by going completely video:
Suppose we have a program of 20 seminars where 19 are local (D=100miles) and 1 is not (D=3000miles). The average distance travelled is (19*160 + 3000)/20 = 302 miles If the remote speaker gives their talk via videoconference at negligible energy cost, the mean distance becomes 152 miles, a factor of two lower.
Most would agree that an all-video seminar program would provide significantly diminished value though.
Am I right to divide the cost of the seminar between all members of the institute? This is not how we did the accounting in the white paper, which counted up individual astronomers' miles and then divided by the population - but this is how seminar organisation works, financially, and it should lead to the same overall answer in the end anyway. The difference is in who bears the responsibility for the energy cost - and in fact this is the important difference. The seminar speakers that travel long distances tend to be the "hot" speakers who are in demand - they cover a lot of miles, but these should be divided by the number of people in their audience, who have (implicitly) agreed to pay this energy expense by inviting the speaker. Let's hope that pointing this out does not encourage hotshots to travel more though - they are very likely take up a significant fraction of the profession's travel energy budget!
- Local speakers are cheap: ~1kWh/day/astronomer, 1/4 of the mean local computing expense
- Speakers coming from a long distance are expensive (an order of magnitude more, or more), and even in small numbers may dominate the average cost.
- A factor of 2 or more reduction in a seminar program's energy consumption can be gained by having distant speakers give seminars by videolink.
- The benefits of replacing distant speakers with videoconferenced talks lie mainly (as ever) in outreach: if you schedule a video colloquium, make sure it is heavily advertised!