Slashdot Mirror

← Back to Stories (view on slashdot.org)

Harnessing EVE Online For Science (mmorpg.com)

Posted by Soulskill on from the waste-time-productively dept.

An anonymous reader writes: Scientists and the developers of space MMORPG EVE Online are working on a project to harness the power of the game's huge playerbase to do useful scientific work. The Human Protein Atlas has 13 million images to map, and there's no way a small team of scientists can manage that task alone. CCP Games, the makers of EVE, will try to encourage contribution by creating a mini-game within EVE to train players and get them to do some cataloging. To start, "Project Discovery will feed about a 250,000 images of microscopic cells and tissue that players will then study to identify basic shapes and structures, categorizing the images in a way that will help scientists deduce a given protein's purpose." The developers are confident that the EVE community, which has already come together to support various charity endeavors, will rally behind this noble cause as well. To get players to participate, the devs reward players with loyalty points that have some sort of positive effect within EVE.

49 comments

  1. Science as a Service by Lodlaiden · · Score: 3, Insightful

    If they had an API to request an image and return the results, I'm sure quite a few mobile games would get behind building that in. Way better than click hammering for no purpose.

    --
    Suborbital [spaceflight] is the special olympics of spaceflight. - Rei
    1. Re:Science as a Service by Anonymous Coward · · Score: 0

      That might be what mmos.ch provides.

    2. Re:Science as a Service by mwvdlee · · Score: 1

      Not just a few mobile games, but entire goldfarming datacenters would use the API!
      Or do the results have to be atleast somewhat reliable?

      --
      Slashdot social media options: AIM, ICQ, Yahoo, Jabber and Mobile Text. Why no MySpace?
    3. Re:Science as a Service by Lodlaiden · · Score: 1

      That's where aggregating and ignoring outliers would help stabilize the results. It would be similar to Googles reCaptcha, which crowd sources Google understanding the address numbers on pictures of houses it's taken.

      --
      Suborbital [spaceflight] is the special olympics of spaceflight. - Rei
  2. Great idea. by Anonymous Coward · · Score: 4, Interesting

    I'd love to see more game devs do stuff like this.

    Some sort of opt-in system where you could get some sort of benefit for doing it, even if it is only cosmetic in design.
    Forced on people would likely annoy people enough to maliciously abuse the system, sadly. People can be evil.
    Even Google Image Tagging got abused when large numbers of people started tagging images as nigger.

    1. Re:Great idea. by Anonymous Coward · · Score: 0

      I tried http://galaxyzoo.org/ but found it pretty boring and repetitive. I only classify one or two items each time they email me which according to their statement that if everyone who got emailed classified one thing that would be enough, should be enough. :)
      Anyway, I don't really see how something boring and repetitive integrates well with a game. They'll have to be pretty creative about it.

    2. Re:Great idea. by Anonymous Coward · · Score: 0

      You haven't met the Eve Online community. They are perfect for this sort of thing.

      The kind of gamer that likes Eve is that weird guy who enjoys filing tps reports and gets sweaty with lust when he enters an organized room.

    3. Re: Great idea. by Anonymous Coward · · Score: 0

      I don't really see how something boring and repetitive integrates well with a game.

      Never played a single game in your life, I gather. Level grinding? FPS? Every arcade game ever? Games are filled with repetitions.

  3. some sort of by turkeydance · · Score: 1

    fail

    1. Re:some sort of by Anonymous Coward · · Score: 0

      yupp... like guppies, in my underpants

  4. Why 'scientists' are nothing but arseholes. by Anonymous Coward · · Score: 0, Troll

    Anyone can call themselves a 'scientist' and this is a case where a total waste of money, time, and life is being called science. The game is one where only bastards play and is most probably a front for the darknet and other sick people who only love to inflict pain on others. Besides being so bug ridden that you lose items for no reason other than they disappear.

    1. Re:Why 'scientists' are nothing but arseholes. by Anonymous Coward · · Score: 0

      Tell me again what your raven was equipped with.

    2. Re:Why 'scientists' are nothing but arseholes. by Anonymous Coward · · Score: 0

      I guess the voices convinced you not to take your meds anymore because they were made by scientists.

    3. Re:Why 'scientists' are nothing but arseholes. by GloomE · · Score: 1

      A torp launcher, cruise launcher, a heavy assault launcher, a heavy launcher, a rocket launcher, a light missile launcher, a shield booster, an armor repairer, a hull repairer, a 1MN afterburner, a 10MN afterburner, a 100MN afterburner.....

      Should have been invincible, but the other guy cheated and CCP don't know how to make games.

    4. Re:Why 'scientists' are nothing but arseholes. by sg_oneill · · Score: 1

      A torp launcher, cruise launcher, a heavy assault launcher, a heavy launcher, a rocket launcher, a light missile launcher, a shield booster, an armor repairer, a hull repairer, a 1MN afterburner, a 10MN afterburner, a 100MN afterburner.....

      So you just whacked any old shit on your ship and decided it should be invincible. Good god, read a wiki or something before undocking.

      --
      Excuse the Unicode crap in my posts. That's an apostrophe, and slashdot is busted.
    5. Re:Why 'scientists' are nothing but arseholes. by Anonymous Coward · · Score: 0

      read a wiki or something before undocking.

      Read a wiki? Read several wikis, watch countless hours of battles, read several EVE news sites, decide what you want to do long-term, download helper programs to control your skills acquisition, get some implants, feed the meter for 2 years to get useful skills, undock and die, undock and die, undock and die, undock and die, oh, and undock and die.

  5. Why not computer based evaluation ... by perpenso · · Score: 3, Insightful

    Might make more sense to apply computer vision to the problem, have a software based evaluation of the image and then flag promising images to be viewed by actual medical personnel. This is an approach already widely in use today. Scanning sonogram, x-ray, mri, etc imagery for "anomalies" and highlighting those for medical personnel.

    1. Re:Why not computer based evaluation ... by Fwipp · · Score: 1

      Because they don't have any algorithms that do as good of a job at classifying them as humans do. If you can design one, I'm sure they'll be happy to help you write the paper describing your methodology.

    2. Re:Why not computer based evaluation ... by perpenso · · Score: 2

      Because they don't have any algorithms that do as good of a job at classifying them as humans do. If you can design one, I'm sure they'll be happy to help you write the paper describing your methodology.

      And humans can do a better job evaluating all the other medical imagery I mentioned above, yet computer vision is increasingly used to augment and screen these human evaluations. And we're talking trained medical personnel in these case. Not gamers who received abbreviated and casual training. The fact that a regular person can be quickly trained in these evaluations suggest that machine interpretation is quite feasible.

      And yes, my area of research in grad school was computer vision and this is the sort of project someone would do for a master's or phd thesis. If you look through the computer vision literature you will find that machine analysis of medical imagery is a very common research project and that techniques are quite advanced.

    3. Re:Why not computer based evaluation ... by Anonymous Coward · · Score: 0

      In this case, two trained Human Protein Atlas staff members examine every image. The gamers are there to provide extra eyes to find things that the staff members miss. It makes sense that software based evaluation could also look for things the staff members miss.

    4. Re:Why not computer based evaluation ... by Huitzilo. · · Score: 1

      And yes, my area of research in grad school was computer vision and this is the sort of project someone would do for a master's or phd thesis. If you look through the computer vision literature you will find that machine analysis of medical imagery is a very common research project and that techniques are quite advanced.

      You bet they tried. This problem is a bit harder than some standard image recognition project you'd assign to an undergrad. You also bet that many undergrads have failed at this already.

    5. Re:Why not computer based evaluation ... by Anonymous Coward · · Score: 0

      Computer analysis of diagnostic images are usually looking for specific, known anomolies. This project involves classifying a bunch of unknown labels in diverse tissues with varying image quality and magnification. The label may not even be present. The images they're trying to characterize are a lot more random and featureless than typical clinical images.

    6. Re:Why not computer based evaluation ... by Electricity+Likes+Me · · Score: 2

      They're probably doing both? The recent mystery star discovery was a result of human analysis of transient light curves which were missed by the current side-effects.

    7. Re:Why not computer based evaluation ... by perpenso · · Score: 1

      And yes, my area of research in grad school was computer vision and this is the sort of project someone would do for a master's or phd thesis. If you look through the computer vision literature you will find that machine analysis of medical imagery is a very common research project and that techniques are quite advanced.

      You bet they tried. This problem is a bit harder than some standard image recognition project you'd assign to an undergrad. You also bet that many undergrads have failed at this already.

      As I said, grad students not undergrads. In particular those who are doing their research in the area of computer vision. Who would be writing a master's or PhD thesis as part of such a project. We are *not* talking about people taking a single computer vision class as an undergrad.

    8. Re:Why not computer based evaluation ... by Anonymous Coward · · Score: 0

      Computer analysis of diagnostic images are usually looking for specific, known anomolies. This project involves classifying a bunch of unknown labels in diverse tissues with varying image quality and magnification. The label may not even be present. The images they're trying to characterize are a lot more random and featureless than typical clinical images.

      Note that I am referring to grad students doing research in the field, not some commercially deployed system that has been available for a number of years. The harder sort of problem you are referring to is the sort of stuff researchers are working on today. If lay people, video gamers, can be informally trained to accomplish the task then this might very well be a good thesis project for someone. Degraded imagery, partial imagery, template matching (not overly concerned with magnification or scaling) with missing information, ... people have been overcoming these issues for years.

    9. Re:Why not computer based evaluation ... by Anonymous Coward · · Score: 0

      Yet even in cases where there is machine assistances or even full machine analysis, you need a reference set, either for training and evaluating the performance of the developed system. I've watched enough grad students spending days categorizing their datasets by hand so that they can then test how well their program is doing, or even more so when they need both a training set and a test set. Some now resort to mechanical turk, and something like this would be helpful for development of such software.

    10. Re:Why not computer based evaluation ... by Anonymous Coward · · Score: 0

      This project seems to already have a set of images evaluated by actual scientists. That would be quite useful during development. I would also expect some sort of ongoing sampling of the computer output by actual scientists to confirm results. They are probably go to do so with the gamer generated output so there is no real additional effort required. Plus I expect the software is also going to be recommending further human review of some images as well, that will help with false positives.

    11. Re:Why not computer based evaluation ... by Spinalcold · · Score: 1

      Software just isn't good at pattern recognition and this is far from the first project to seek the public's help in these type of things, but this could potentially harness more participation than previous attempts. Zooniverse started with getting people to classify types of galaxies, now they have a few dozen science projects that use the pattern recognition humans do better than computers. These areas of science span physics, anthropology, biology, and linguistics, even a project that you can sift through LHC data to help find exotic matter (haven't done that one so I don't know how good it is. I did the Planet Hunters one for a while and they managed to locate 7 exo-planets that year using the Kepler data.

    12. Re: Why not computer based evaluation ... by Anonymous Coward · · Score: 0

      Some researchers are always looking for larger sets of training and testing data though. I've seen advisors tell junior grad students to just sit down and spend another week or two doing it again to expand a data set, becauae it is not like it is difficult work, right? At least they splurge to get some cheep undergrad help for the senior grad students when they are starting to feel a time pinch on their thesis or the advisor values the grad student for doing more than grunt work.

    13. Re:Why not computer based evaluation ... by perpenso · · Score: 1

      Software just isn't good at pattern recognition and this is far from the first project to seek the public's help in these type of things, but this could potentially harness more participation than previous attempts. Zooniverse started with getting people to classify types of galaxies, now they have a few dozen science projects that use the pattern recognition humans do better than computers. These areas of science span physics, anthropology, biology, and linguistics, even a project that you can sift through LHC data to help find exotic matter (haven't done that one so I don't know how good it is. I did the Planet Hunters one for a while and they managed to locate 7 exo-planets that year using the Kepler data.

      Software can be very good at pattern recognition. However all software is not created equal. I'm not talking about software that the scientists wrote themselves, or some computer science grad that had an image processing class or even a single computer vision class. I'm talking about post grad computer science folks doing research in computer vision. Seriously, the sort of things you describe *are* / *have* been done as computer vision projects. Go to a university library and sit down in the computer vision journals area and start flipping through the journals. I've done that. I've done research in this field.

      Now, again, consider that we are talking about lay people with no expertise in the field being briefly and informally trained to do these tasks. This suggests the task is one well within the reach machine evaluation.

    14. Re:Why not computer based evaluation ... by Anonymous Coward · · Score: 0

      Yes computers are great at pattern recognition, if you can find/define parameters to match against. The problem here is that the parameters are too broad and undefined for computer matches to be worthwhile.

      Look how long it's taken facial recognition to reach the level it's at today, and that's with some pretty specific parameters to match against. It'll happen someday I'm sure, but for now it's much faster and easier to get a bunch of humans to do it for you.

    15. Re:Why not computer based evaluation ... by Anonymous Coward · · Score: 0

      Yes computers are great at pattern recognition, if you can find/define parameters to match against. The problem here is that the parameters are too broad and undefined for computer matches to be worthwhile. Look how long it's taken facial recognition to reach the level it's at today, and that's with some pretty specific parameters to match against. It'll happen someday I'm sure, but for now it's much faster and easier to get a bunch of humans to do it for you.

      Actually humans are not that great at facial recognition. For example eye witness testimony and eye witness identification of suspects is notoriously bad. Do not confuse a human's ability to recognize someone they know very well, have spent lots of time around, with facial recognition in general.

      Plus human facial recognition is a poor analogy since it is a *very* specific task that we have evolved for over millions of years. And the hard wiring is *very* specific to facial recognition, it rarely transfers over to recognizing something else.

      And again, since video gamers can be trained to identify things with very minimal and informal training your assumption that the parameters are complicated is dubious. Regarding "broad" parameters, that is the level that lots of computer vision software works at. Its not some image processing like process where things are more pixel oriented. The pixels are analyzed to extract high level features and then most of the analysis takes place with respect to these high level features. A set of "broad" and general characteristics and relationships is nothing new in computer vision.

  6. Now that's optimistic by TFAFalcon · · Score: 4, Informative

    Has the writer ever actually played eve? It's a game about causing other players as much frustration as possible. And now they want to give the players a chance to mess up something even more important than internet spaceships?

    I think this is a more balanced take on the issue.

    1. Re:Now that's optimistic by Anonymous Coward · · Score: 1

      Yes, finally, a mini-game for the care-bears while they mine their ore. Perhaps their efforts could be rewarded with tokens redeemable by James315's goons for protection money. But somehow, I, too, think this will only end well for the research scientists. What do the goons think of this plan? What do the Russian botters think of this plan?

      I think what they should work on is a device that can convert asshole rage into electrical power. With EVE, we'd never need fossil fuel again!

    2. Re:Now that's optimistic by Anonymous Coward · · Score: 1

      The devs insist on causing players as much frustration as possible too. Have you ever set up a POS? Have you ever done it BEFORE the anchoring/onlining times were changed? Have you ever fuelled a POS before fuel blocks? Have you ever trained for a capital, only to have that capital ship nerfed a month after you finally finish the almost two years' skill training to fly it properly. This wouldn't surprise me in the least. They should add this mini-game to all gate jumps in and out of Jita, through the Uedama-Sivala bottleneck (but only if you're in a freighter/jump freighter/bowhead or are carrying more than 200M isk in your hold). They should add it to all jumps in nullsec. They should add it so that you could empty your ore hold while mining ore or ice. And especially they should add it in order to enable a point.

      Fuck EVE Online, and fuck CCP.

    3. Re:Now that's optimistic by samspock · · Score: 1

      Bittervet much?

  7. It could work by Anonymous Coward · · Score: 0

    Eve is so boring that doing some "Mechanical Turk" jobs is probably more interesting than "Spreadsheets in Space".

  8. Reality follows science fiction by Chalex · · Score: 3, Informative

    Neal Stephenson had this as a plot point in his book REAMDE. An MMORPG had a mini-game where you had to recognize some objects, but actually they were being fed TSA machine images from the airports, looking for dangerous objects in luggage.

    1. Re: Reality follows science fiction by valdezjuan · · Score: 1

      They were specifically looking for people entering the secured areas via the exit to baggage (essentially looking for people flowing the wrong way on what should be a one-way passage).

    2. Re: Reality follows science fiction by Anonymous Coward · · Score: 0

      And the whole point of the story was that since the program could already detect people going in the other direction (to make the whole mini-game work), it could already do the job without human input.

  9. Harnessing Eve.. by Anonymous Coward · · Score: 0

    Is it just me or did anyone else read the title as Harassing Eve Online for Science. Oh our Freudian minds..

  10. Cynicism by jklovanc · · Score: 1

    Sorry but how will they differentiate between the one who actually do the analysis and the ones that just click buttons to get the reward? I thought about speed but some people are pretty good a pattern recognition and will be fast.

    1. Re:Cynicism by Anonymous Coward · · Score: 0

      that's easy: redundancy

      You give the same test to at least two different people. If the results differ you have a cheater and need to re-analyze the data.

    2. Re:Cynicism by SuiteSisterMary · · Score: 1

      If a thousand people look at a picture of a circle, and 80 percent say 'it's a circle,' and 20 percent say 'it's a square,' you throw it in the 'probably a circle' pile.

      The idea here, I think, isn't to make positive identifications and what not, but to pre-sort or vet the things to allow the real experts to direct their time and attention more efficiently.

      Note that this is how computers in space tend to work, where you have to worry about a stray cosmic particle flipping a bit in a processor; you have at least three processors all do the same calculation, and you take the consensus answer.

      --
      Vintage computer games and RPG books available. Email me if you're interested.
  11. This should stop by Anonymous Coward · · Score: 0

    If you pay for a game, this piggybacking is unexpected. There is nothing to harness, this is not a free-to-play. I pay money for entertainment, I do not expect you go out of your way to derive even more money from me.

    I do love science, I just do not like Darth Vader's deals. In fact, I do not like them at all - and this includes all changeable TOSes.

    1. Re:This should stop by Anonymous Coward · · Score: 0

      Haha, CCP has a history of trying to exploit their players. Personally, I was fed up after the Monocle Debacle.

      Not that the monocle was a problem - a 70$ monocle to troll poor people with is a brilliant idea in a game like EVE. The problem was that there was so many things in eve that were completely broken at the time. That CCP decided to spend developer time on microtransactions for a completely useless Walking in Station environment was just mocking to the people who spent hours each day, trying to deal with broken sovereignty mechanics, unintuitive POS structures, mining which was lol at the time. Not to mention the never-ending drake blobs.

      I know, the visual artists probably are lousy programmers, but this is a matter of principle. Fix the broken things instead of trying to add more broken things. Heck - I'd prefer a redesign of the Raven to that monocle. At least it would have been improving the visual experience for the lvl4 runners in hisec.

  12. Zooniverse? by Toad-san · · Score: 1

    It's odd that Zooniverse https://www.zooniverse.org/ hasn't gotten involved with this. They do all sorts of projects that involve humans categorizing or analyzing images, some quite complicated. I don't know if THIS particular project can be simplified enough for the average Joe to work with, but it might be worth a try.