Jump to content

Ryan Marcus

  • Content Count

  • Joined

  • Last visited

  • Days Won


Ryan Marcus last won the day on November 16 2010

Ryan Marcus had the most liked content!

Community Reputation

71 Excellent

About Ryan Marcus

  • Rank
    Registered User
  • Birthday 11/30/1991

Contact Methods

  • Website URL
  • Skype
  • AIM
  • MSN
  • Yahoo

Profile Information

  • Name
    Ryan Marcus
  • School
    Los Alamos High School
  • Location
    Los Alamos, NM
  • Occupation
    Freelance Programming: It's FTW
  1. For sure -- computers aren't able (yet!) to read a judges facial expression when making an argument, or come close to comprehending spreading. But their damn good at the logos part. And I don't know if I'd look towards Watson as an example of language comprehension -- there are actually better examples. Watson's biggest restriction was that it had to be able to listen to a specific question and give a specific answer. Language analysis in general doesn't require the same level of precision. Ex: Watson hears "The first color in the color spectrum" and must come up with "What is red?" A language analyzer can here "the first color of the color spectrum," and can come up with "red, orange, yellow, green, blue..." The difference is that while Watson was required to give one answer, a analyzer can come up with "fuzzy" answers, preferably with an associated probability. Relating this back to debate, a computer would hear: "South Korea free trade agreement will pass, your plan kills South Korea free trade agreement, South Korea free trade agreement is good." A debate computer, unlike Watson, doesn't need to come up with the hyper-specific, one-word-is-one-chance answer. it does, however, need to assign arguments to premises (which is easy, given punctuation / taglines: (A) South Korea free trade agreement will pass ( Your plan kills South Korea free trade agreement © South Korea free trade agreement is good (D) You lose (we assume this premise) From parsing these premises, we also gain (read: assume) the logical structure: A -> B -> C -> D From here, we can permute all the possible responses very easily: ~ is not, -> is implication, => is therefore ~A: ~B -> ~C -> ~D ~A AND ~B: (~B -> ~C -> ~D) AND (~C -> ~D) =>~D ~A AND ~C: (~B -> ~C -> ~D) AND (~D) => ~D ~B AND ~C: (~C -> ~D) AND (~D) => ~D ~A AND ~B and ~C: (~B -> ~C -> D) AND (~C -> ~D) AND (~D) => D ~C: ~D => ~D etc. (notice how the logic makes the non-unique double turn OK but a unique double turn not ok? Cool how logic works...) We can also analyze each of these logical setups to determine the "strength" of each argument. That's where the fuzzy logic comes in -- it is possible to "rate" a card (however imperfectly) from 0 to 1, and then determine which strategy uses the "best" cards. So it is certainly possible... not that it would be good, and it certainly wouldn't be able to beat a team who was paying attention, but it could still be pretty awesome. No, I definitely haven't claimed to have created a language parser capable of determining logical arguments in terms of theory, Ks, Ts, or basically anything but a very, very cut-and-dry politics shell. I'm still working with symbolic tag lines (i.e., the tags I'm feeding it are things like "plan -> unpopular -> no political capital"). I thought so too! Thanks Arik.
  2. That's just the thing though -- code I write can generate an opinion about a DA, but your wallpaper can't think about "jane eyre." I'm not saying that some computer program can evaluate if a DA is good (although it might be able to see if one is bad), I'm just saying it is really interesting that computational linguistics and software are getting to the point where computers can "think" about something like a DA at all! Actually, it's pretty terrible at it at the moment... but this isn't an experiment in "brute force" debate, this is an experiment in computer AI. I just think policy debate is a really cool platform for it. To each his own, I suppose.
  3. Ah! Well, that's actually pretty simple. Look at this argument: (1) Men are mortal (2) Socrates is a man ------ © Socrates is moral This argument could be presented like this: Since your tag for this would likely contain "Socrates is mortal," my program using grammatical parsing to determine what sentences are relevant. In order to do this, it breaks each sentence down into a more basic form that is kinda like this: Fragment 1: man -> mortal Fragment 2: socrates -> man Fragment 3: socrates -> mortal Since your tag contains the relevant objects (mortal and socrates), the program selects the fragments that are relevant. Consider this form of the argument: We get: Fragment 1: man -> mortal Fragment 2: socrates -> man Fragment 3: football -> awesome sport Fragment 4: socrates -> mortal Since you've given the program no indication that you care about football or awesome sports, and since there aren't any relevant fragments related to football or awesome sports, fragment 3 doesn't get cut. @ TLF: Because I secretly want the entire debate community to slowly decay into technocratic babble over who has the better card cutting algorithm so that I can root out all those smart bastards who were able to beat me in high school! More likely, because I'm a computer science major, pretty damn nerdy, and come on, you've got to admit, it'd be pretty cool to debate a computer. You could throw a DA at it and see what cutting-age AI thinks would be a good response. Logic is what we do, and computers happen to be really good at logic. Think of a (fairly bad) winner's win argument: Winner's win because getting bills passed creates political momentum which allows for more bills to be passed.Therefore, Obama passing the plan will allow Obama to pass KORUS> A computer can interpret that as: X(passing bills) => Y passing through momentum Obama(passing the plan) => Obama passing KORUS And with a few near-millisecond transactions, come up with: Obama(passing the plan) => momentum => KORUS passing This may seem mundane to us, but the power in combining (permuting) those two statements is truly phenomenal. Now we can use a standard cost/benefit algorithm, and a computer can instantly have questions like: Is KORUS passing good? Is momentum good? Is Obama passing the plan good? A computer can then, using my cutting and Google News stuff, find and cut cards relevant to each of these questions, pick the best ones, and stick them in a file. You've got to admit that's kinda cool. People seem to think that there's an art behind logos -- there isn't. There can be creative arguments, but that's nothing you couldn't come up with through brute force. Computers are pretty good at brute force. What a computer will not be able to do (at least in our lifetimes) is grasp the concept of pathos. That's where the "art" part of debate, in my opinion really is. Thanks, Ryan Marcus
  4. Not quite sure what you're intending to get across here. Is there a specific style of argument you think my program is unable to handle? As long as the words are logically connected using standard and grammatically correct English, it should be no problem. More than likely, you're using a sub-optimal tag. Feel free to email me the card/tag combination that isn't working for you and I'll do my best to either help you out or improve the code. I'm actually working on a "deep blue" of debate using my auto cutter and a language called Prolog -- sadly, it won't be able to hear spreading, but I might set it loose on a virtual debate at some point!
  5. There is a Windows version now. There's a link in the first post. A few of the features I've discussed in the video are not yet in the release version. Done.
  6. Shameless bump because I updated the first post with a new feature and video.
  7. for (Offense o : partner2AC) { if (o.wasDropped() || o.badlyAnswered()) { Speech.extend(o) } } if (Speech.timeLeft()) { for (Defense d : partner2AC) { if (d.wasDropped || d.badlyAnswered()) { Speech.extend(d); } } } if (Speech.timeLeft()) { Speech.laughAtNegBlock(); } Probably not the best 1ARs you could give. Impact calculus, dropped theory, and other such essentials in the next version.
  8. I'm not entirely sure what you want or what you mean by this request... although yes, it does sound much simpler then what I'm doing. Process the first and last phrase of what? For what purpose? It seems like this could be interesting. I'm sorry if you take offense to my desire to charge a dollar -- I'd like to point out that my program has half as many flaws as Windows Vista, which sold for a bunch more than a dollar. I'll admit, at its current point, it is more of a tech demo or a toy than a super-useful project. However, as I think the more recent video shows (linked on the webpage), this is going to turn into something much much more than a simple card cutter. Who knows... maybe I'll adopt the Minecraft theory... charge a dollar for the first version, then two for the next version... probably not though. That seems a little evil. Why would a camp pay a few hundred dollars for posting a link on a wiki? I don't understand what you are saying. I'd like to thank everyone who has donated, some in very generous amounts, and some more in the amount of a single dollar. From contributions, we've raised enough money to cover our transportation costs to the California State University Northridge tournament. I'd personally like to thank everyone who has helped make this happen.
  9. It would be SUPER helpful if you could send me the article you were cutting as well as the tag you were using. There's an art to writing tag lines that properly make the cutter work -- one of the biggest problems with supervised machine learning is the difficult in collecting data. As always, it would be a huge help if you could send me the articles/cards you were trying to cut. As a reminder, this is mostly designed to cut politics. Thanks, Ryan Marcus
  10. Ya... admittedly, I thought of that pricing scheme at about 1AM. I'm normally a proponent of free software, and while I can tolerate commericalware, I really don't want to charge people for updates. I'll change it to buy once, get all the updates. The website will reflect these changes soon. ALSO: Check out the original post. New video with google news feature.
  11. I actually went questing for that earlier today. The algorithm (although your spelling is far cooler) is not public, but there are some public ways of doing it... but for now I think that's a bit out of reach for me.
  12. Good news everyone! Check out the original post.
  13. Well, good suggestion -- if you've got anything more specific floating around in your head, let me know. Right now I'm thinking of letting the user enter a Google News search, showing a list of article headers, letting you decide which ones should be cut and which ones shouldn't, and then asking you to assign a tagline to each article. Hopefully from there I'll spit out a file of some kind. Thanks, Ryan Marcus
  14. I'm not quite sure what you are talking about -- do you mean cutting everything within a certain google news search? For example, google news searching "republicans will win the house", then automatically cutting every article that shows up in that search? I think there would be some issues with that -- first, the results of the Google search probably aren't all going to be the cards you want, second, each card should probably have a different tag, and "warrant detection" is a long way off. Am I missing what you are saying?
  15. Actually, in terms of style, persuasion, etc., modern linguists are really starting to break it down into a science. Aristotle put it all in motion with ethos, logos, and pathos, and modern researchers have only made it better. Realistically, this is more of a case study in computer comprehension for me, and far less of an attempt to automate debate. But I think you miss the point -- the "artistic", creative, and truly analytic parts of debate don't show themselves with cutting cards. Tagging cards, perhaps. The truly awesome parts of debate is the reactionary strategy, the catch-all blocks, and the intelligent applications of logos (face it, sheer pathos won't fly in front of a good judge) are a long way away from being computerized. I realize you are making a slippery slope argument here, much in the same way that an amish person would probably argue that labor saving devices make us lazy, but it is important to remember that technology is inevitable. It will change the game, but it'll probably never conquer the game. By the way, I'm hoping to have a public release of the card cutter in about two or three weeks -- stay tuned to this thread for the D/L link. If you are especially interested in the cutter and would be willing to do some pretty serious beta testing, drop me a PM and I'll (probably arbitrarily) decide if I want you on the beta list. Edit: also, yes, this is mostly going to be used for politics cards. You should probably still handcut the main cards in your 1AC/shells/blocks and other things you're going to read more than once. Edit2: tl;dr: PM me if you are willing to beta test. Release coming in 2 or 3 weeks. It'll probably not be too high quality.
  • Create New...