Hacker News new | past | comments | ask | show | jobs | submit login

Very cool! Signed up for an invite.

Couple questions:

1) Is this a Stripe-like CTF that happens over some caffeine-fueled weekend? Or is it more of an ongoing Project Euler-style drop-in-and-solve type process? I don't think I could handle the former, but the latter sounds quite enjoyable and something I might do in my free time just to learn. I did that with PE for a while since it was fun to earn completion points.

2) You mention "Let's Play" style videos. How do you make it so that the solutions aren't given away? Does each player have a customized CTF somehow?




I loved the Stripe CTF. We're not going to adopt the time-boxed model from them. Starfighter : our CTFs :: Blizzard :: Blizzard games. There's no reason for us to ever turn them off. We'll likely have a particular flagship property but we have the option of dropping a new game on the market any time we want.

You don't have to rush through Starfighter games. Some people will, of course, just like there is a metagame to be the first guild on the server to complete a new raid. Players are gonna play. But if you want to pick up our CTF a year after release and just casually spend a few hours learning a new skill, you'll be able to do that, and we'll fully support you in it.

How do you make it so that the solutions aren't given away?

The same way you do it with math problems. Mastery of the subject material is the easiest cheat code.


>The same way you do it with math problems. Mastery of the subject material is the easiest cheat code.

Elaborate? Math questions are the opposite of a good example, aren't they? Even with randomized values it's fairly easy to write generic solutions. People do this even for Project Euler. How much more so for something even cooler sounding?

(I'm very excited to see this project and it sounds like a lot of fun. I'd just be worried that it would end up being gamed, since the drive for cheating is so strong everywhere.)


I actually hope this happens, and that there's a community of code sharing that forms around it.


Wouldn't that just make cheating easier, then? I'm sure you have this sorted out somehow, it just seems like a rather large unaddressed question.


The complexity of the problem solver generally increases faster than the complexity of the problem generator.

For example, a general solver for "what is #{x} + #{y}?" is relatively easy. A general solver for "what is the derivative of #{random_equasion_with_diffuculty(3)}?" requires a full computer algebra system. Take one more step into advanced mathematics, and things get more dicy -- automated theorem provers, for example, require human guiding and hints.


> The same way you do it with math problems. Mastery of the subject material is the easiest cheat code.

This, at best, sidesteps the question. If a solution is a string of characters, that string of characters will end up on the Internet.

If a solution is a series of tasks---try this string of characters, observe what happens, then try this other string of characters---people will still share the strings and instructions on the Internet. All this still holds if the strings are personalized in some way; then the general patterns will be shared.

I spent a few years at university as a teaching assistant for a programming course using an interactive IDE where this kind of thing was required: Run the provided unit tests, ONLY THEN write some code, ONLY THEN run these other tests, ONLY THEN make the appropriate modifications etc. The idea was to make cheating more expensive than actually thinking about the issues, and to some extent this succeeded in the narrow sense that cheating WAS more expensive, but it did not succeed in the more general sense: people STILL cheated because they really didn't want to think things through.

The higher the stakes are, the more not-so-great people will be attracted to the system and will chug through based on someone else's solutions. This will be a fun game for those who see it as a game, but I don't see how the using-this-for-recruiting part will not implode almost instantly.

Finally, the whole thing about keeping even the general subject areas a secret is a bit silly. The work the best tech companies in the world need ranges from medical visualization to databases to systems programming to pentesting (one of you wrote this will not be a pentest game!) to static analysis and verification to distributed systems to scientific computing to drawing pretty pictures in web browsers to... You will only cover small sliver of this. Everybody knows this, why be an ass about it?


It is long-running and not timeboxed.




Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: