Simulations! (late, but how would Scotty have handled it?)
It’s been about five weeks since my last post where I announced we’d start testing follow-ups in the background and also that some improvements to the waterfalls along with adding simulated ET signals to them were imminent. Well, we checked those subtle but important waterfall improvements off the t0-do list two weeks later and yesterday, at about 21:30 UTC, we finally got simulations live. It took five weeks – longer than we had figured, but like most science and engineering tasks, you’re doing something new and don’t always know what you’re getting into until you start making it work. Now, Star Trek’s Chief Engineer Lt. Commander Scott (“Scotty”) might have figured that it should take a week, multiply that by five to get the real time for the task and then multiply it again by five and tell the captain it would take 25 weeks. The Captain would demand it in five and Scotty would say “I’ll do my best, Captain”, self-satisfied and with a barely visible smile. I’ll try to be more Scotty-like in the future.
Trekkies: I know that the Star Trek franchise has alluded to and made references to this process over the years. Any specifics about those references would be appreciated.
Our primary purpose for inserting simulations (“sims”) into the waterfalls is to conduct a controlled experiment which will measure the statistics on how well you, as a group, are able to detect a signal in a waterfall with noise and sometimes RFI signals in the background. The results we get will be an important part of a peer-reviewed scientific paper we plan to eventually publish and we’ll make comparisons to some basic computer algorithms. So, you’ll be a part of these real, important scientific results derived from the SETILive project.
Another benefit of adding sims that users have pointed out is that it’s an occasional “test” that helps us be sure we continue to be careful, looking closely for that weak signal track that’s fading in and out and seems to sort of follow a line. It could be a signal of interest, trigger a follow-up and who knows, even make it to “Wow!” status.
The sims have random positions, angles, “wandering” patterns and randomly fade in an out in brightness to emulate typical signals. The two key parameters that are systematically varied over wide ranges are brightness and how much they randomly move from side to side (“erratic-ness”) . Because we’re trying to find the limits of our users’ ability to pick these out, some will be so dim that no one would be able to see them, so don’t feel badly when we show you where a simulation was that you missed. Currently, you can’t see just how dim a missed simulation was and we’ll look into whether or not we should change that. We have to carefully consider doing anything that might affect how you as a group do the classifying. It might add an unwanted variable into the statistics. There have been some good suggestions on Talk for how to do it if we can – thanks!
Speaking of follow-ups, that’s been an even tougher slog. We’re doing something that hasn’t really been tried before. Yes, SETILive follow-ups are modeled after the automatic ones that are already working at the ATA, but replacing detection algorithms and calls to local computer programs at the ATA with the remote,web-based, human-in-the-loop SETILive detection process is quite different in many details so requires some programming and communication protocols that are not already there. We also can only work at certain times in the ATA schedule and also have to deal with scheduling around the availability of the experts who are making it work. It’s coming along, and we feel it will be working soon. When will it be live? I’m afraid we can’t really say exactly. Where’s Scotty when you need him? He’d have an answer.