Voorwerpje Hunting for Beginners

Three months ago I like the rest of you had never heard of a Voorwerpje. Laihro had been working with Bill Keel on the hunting for active galaxies with ionized gas clouds, and had worked out a way to extract active galaxies from SDSS, and had the idea of using my hunting program to search through the list. Two further lists of likely candidates also came from Kevin (from X-ray galaxies) and Bill (known active galaxies). This hunt needed a good tutorial, which Bill wrote, and then the hunt was launched. Voorwerpje hunting was looking for needles in haystacks, most images didn’t have a Voorwerpje, but it included some very pretty and interesting galaxies – enough to keep me in OOTD candidates for a year.

You all agreed on this one (otherwise known as NGC 5256 or Markarian 266).

It started with an announcement in the blog and the forum, and by the second day 37 people had clicked over 20,000 times. All hunts are like this to begin with; usage dropping off after a few days. After ten days, 3 dedicated souls had clicked every single one! By the end 7 had done them all. The clicks per day dropped off, dropping to a few hundred a day when the Supernova hunt was on. I had clicked through a few early on just to check the program and data, but had only clicked 50ish times in the first month. Then one day I just wanted to check something, clicked on to the next image. And there was one! I was hooked, and between then and the end of the hunt clicked my way through nearly 5,000 images (more than I have ever done on Zoo 1 or any hunt since).

You largely agreed on these two (in both cases one person said “possibly”).

Over the hunt (2 months), 139 people contributed 192,000 clicks, another 9 signed up but never clicked on an image. On average they clicked about 1,300 times each (curiously this average clicks per user was true after 10 days hunting and was still true at the end).


Going further down the list, these still had all of you clicking Yes or Possibly.

For every one who took part, I keep a file, which records your result against each object you clicked. As there is no complex database used, it is highly scalable and can handle very high loads (on a server that is 95% idle, it never appeared in the top 100 users). In general, these individual results won’t be used, what is used is an amortized anonymous rating for each object.

There were 18129 different objects listed, 14 had duff SDSS data (an unusually high number), the rest had at least 10 of you click Yes, Possibly or No. Most had 10 clicks, some had 11, a few 12, a handful got to 14 clicks. For each object, counting a yes at 1 and a possible at 0.5 and dividing by the number of clicks, gives a rating between 1 (everybody thinks its a Voorwerpje – 2 instances), to 0 – nobody thought there was anything (most). Everything that scored anything has been passed to Bill for further analysis. They will go through from most probable downwards until the results prove too few and far between. What analysis he will do I don’t know. Though I expect he will blog about it eventually. Reducing your months of clicks and thousands of hours work to this small table took only 5 minutes.

These had equal numbers clicking Yes, Possibly and No.

My hunting program has now been used for Mergers, Peas (3 times), Irregulars, Voorwerpjes and a few other things (not publicised in the forum). What next?

As I write this two days after having sent Bill the results, clicks are still being made… I don’t mind they are pretty galaxies. I now know that a Voorwerpje is a large ionised gas cloud in a galaxy, and they can be rather pretty. As to what they are and what we will learn from them, that will have to wait – as I don’t know either.