In competitive programming games like LiveCTF, robust infrastructure with automated re-queue systems ensures smooth competition despite technical issues, while strategic bot design evolves from simple survival tactics to complex multi-phase approaches, with teams leveraging local testing and phase-based updates to develop increasingly sophisticated competitive strategies.
深度探索
先修知识
- 暂无数据。
后续步骤
- 暂无数据。
深度探索
Phase 1 Recap - LiveCTF @ DEF CON CTF Quals 2026本站添加:
Hello everyone and welcome back to live CTF phase one recap. We hope you all had a good night of hacking and bird training.
And good morning. We're back. It's me, Josh and Carl.
Hello.
And I think >> good afternoon.
Phase one seemed interesting.
Um I was asleep for most of it, so do you want to give me kind of your your takes?
Yes. So, uh well, you know, maybe we will Well, we can we can come back to kind of the the bigger picture later, but yes, so we had some minor issues. We started up off completely fine.
Then, I think on the seventh game, we started getting crashes.
Uh and then looking at the error log, I saw that uh the driver was trying to allocate like an infinite amount of memory, basically.
Uh and we tracked this down to uh a bug. We can we can come back to that one uh later. It's a It was a little bit funny. I posted in the announcement channel as well. Um but then that was quickly fixed, and we could just kind of rerun those games.
So, like overall, the the infrastructure is fairly robust in that sense in that we can like if something happens with the game, we can just kind of re-queue it, and the appropriate bots will still uh play in it. So, that's fine.
Um also, like you know, more than three quarters through the phase, we got a ticket opened by one of the teams saying that uh you know, the the syscalls in the documentation uh they are misnumbered, uh which they were.
Uh and there was some other small details that are actually incorrect. So, we fixed that. Uh so, I just assumed that all the other teams just like discovered this independently as well, and this wasn't really a problem because it took like over 3 hours for anyone to tell us.
>> if you can argue well enough, the code is self-documenting.
Yeah. Uh uh Yeah, so that's that's fine. I think that was basically it in terms of like, you know, minor issues. So, we completed the first uh phase. Yeah, overall, the eight games have been smooth.
>> Yeah.
Um and uh we haven't had too much feedback from the teams yet. The The feedback we've had was positive, but I think, you know, also no news is kind of good news. Kind of tells us that they're busy actually playing the game instead of talking about the game. So, you know, good thing.
Yeah, honestly, that's a trade-off I'm happy to make. If at the end of the CTF, they're just like, "By the way, it was cool." I'm like, "Good, you were busy.
You enjoyed it while it was running."
Uh but let's take a look at the phase one scoreboard, I guess, the final state.
>> So, >> Yes. we have 4B52 taking first place in phase one, 300 points over Orange Fan Club. And actually, I didn't notice this, but there's a one-point difference between [laughter] second and third place.
Oh, yeah. They were like So, the uh the I wish we could do kind of like an animated timeline thing here.
Unfortunately, we, you know, we don't really have that technology, but I have been keeping track a little bit throughout the phase.
And I will say that the teams in second to fourth, even like fifth place, like Orange Fan Club, Re: Pokémon Collection, Soul Sauna Shogunate, Super Dice Code Lovers, uh they were like going up and down. Uh there was at some point I actually took a screenshot and posted in our internal chat here.
Uh so, at one point Re: Pokémon Collection actually Oh, and Obscurity in sixth place. At one point, they were uh first and second place with the same amount of points. That was after 50 games.
>> [laughter] >> Uh so, they were on the exact same number of points after 50 games.
Um I >> we should go into game design as a profession. Live CTF on.
>> [laughter] >> Let's Let's become uh contractors for designing games.
Yeah. Yeah, I think that's uh it's a good idea.
Uh and I will say that uh 4X, sorry, uh 4B52, um they kind of uh shot up a bit at the end. I think if you if you click on their name, uh you can see here that they have like in the last like 20 or so games, they have a couple of really high scoring ones.
You see there is a 367, 258, 265, 220. So some like really good performances towards the end. So they they they shot up uh towards the end. I didn't really see them early on in the scoreboard.
>> looks like already in phase two, they have some really good Now we don't know what these are relative to other players in the game, to be fair. But it's generally a good sign that getting more points in the game will put you higher up.
Yeah. There's a finite number of players in the game.
As you might have noticed, um the players are kind of padded with NPCs. So uh we have this big beautifully drawn map. We have professionally designed, very in-depth, a lot of back and forth, months of work in it. Um definitely not Glenn sitting there with a tablet being like, "I got a new tablet. What can I draw?"
>> [laughter] >> So I have a big map. It's beautiful.
It It is. I mean, it's one of those things like, "I don't think this will make This won't make it to prod." And then it's in prod and you're like, "Well, it's already here."
You know, it's the classic case of using a placeholder graphics and then like it sticks all the way to release. Yeah.
Um so we have this big map that we need to fill with a lot of players. And yeah, I'd love if we had, you know, 10,000 people, 10,000 uh bird trainers competing.
But >> Yeah. you know, that's not realistic and we can't always guarantee in phase one that'll be the case. People might submit in phase two. So we pad out with NPCs and those are handed out uh to y'all so you can kind of base your bot training program on something.
And there's like a thousand, I think, is the minimum number in the map.
Uh >> Yeah. so if you see a score like 367, where it's likely you got somewhere around like maybe 30 kills, 20 kills, or something. That's a good chunk. Uh because if if everyone did that, you know, birds would go extinct. Um so it's is good sign that they have around that point. Now, you also see some cases where they have zero points, though.
Uh and that's part of the reason why we run so many games. And we're actually going to look at one. This is the uh winning team from the first match.
Have to turn on sound. It's uncomfortable being around. They're like All they were jumping around. Waiting for something to happen. It looks like they're scanning around, maybe. And then they jump around and they're just like owning this section of the map. And as we go through, you see they're jumping around, going back to their safe spot. And then they just wait.
No one entered their safe spot, but they got all of these skills.
And they ended up like first place above an NPC, actually.
Um And then if we go to, I assume, like a different bot that they submitted.
Uh and we play this replay.
Um it was, let me see, tick four of 3,600 >> [laughter] >> that they died.
So, it's really up to a little bit of chance in each individual game, but with enough games deployed, you know, we try to eliminate that. And you can see it's actually the sole Sole Sauna Shogun it, which I love when teams lean into the like PPP BBB MMM like Yeah, and we have I've seen a few bird related things on the scoreboard as well, so I really like that. I wonder, by the way, if this I don't know if teams are doing this yet. Well, we don't really know exactly what the strategies are yet. We haven't really started digging into the code. We'll we'll get back to this, but um you know, maybe there would be something like you know, maybe your bot should start with kind of like some dumb, you know, less computationally heavy strategy initially to kind of clear the surrounding area, like go a little bit rampage initially to just stay alive, and then start doing some more intelligent thing. Like maybe that would be a strategy. Yeah, maybe. I mean, it's up to all of these teams to figure out what strategy works best. They have the driver locally, so they can test maybe their new bot against their old bot.
Throw them in together and see what happens.
And I think that we're going to see a lot of strategies kind of evolving. I think we're going to lead into towards one like main strategy that we're going to see a lot of teams taking with the release of bots after 4 hours, which actually now bots from phase one would have been released for the first couple of matches, right?
Yep, and I have checked that that part of our code is working. I mean, it was tested before, >> [laughter] >> but I'm also like I've seen the download links in the in the system now, so basically, uh like 4 hours after a bot was used in a game or was like assigned to a game technically, you anyone can download that bot. Um so, if a bot if you upload a new version of the bot before it ever gets used in a game, uh it is never revealed to the other players.
Uh so, a bot that hasn't played will not be uh revealed to other other players.
Uh this is, you know, this could be like to uh prevent you from making mistakes when uploading things. It's also kind of to like not make it a viable strategy to just upload tons of different bots and make it confusing for for your oppo- opponents to like figure out which ones they actually want to uh look at. So, it's kind of for everyone's benefit.
Um And I think that's going to lead to a lot of like not catch-up mechanics, but it's going to increase the floor, I guess, as we go through the phases in the games. You're going to see like the simplest bots that people submit in the beginning, the walk forward and shoot directly in front of you.
Um I think you're going to see those replaced with maybe the winning bot from phase one or phase two.
Yeah.
Uh I I wonder if we're going to see some like countermeasures employed by teams.
Uh because this custom architecture that you're working with and and and the way it works is the whole memory space, you know, there's no memory protections.
Everything is like read write execute, right? So, in theory, you could do like, you know, self-modifying codes, the stuff like this. Uh but uh we haven't really started digging into the to the bots yet. Uh but we'll see what what happens there.
Um Yeah. And phase two handout is out.
People have already been already been competing with it.
Um and let's just compare real fast. We see at the beginning or at the top of the scoreboard, rather, we have 4v52, Orange Fan Club, aren't we all in that?
Uh re Pokemon collections. But interestingly, all of these people have uh 80 games played, and then number seven is 74.
21 is 66. So, some people didn't get their bot in in the very beginning.
And we see some of those same people at the top or near the top of the scoreboard here. We see some familiar faces, Orange Fan Club, Soul Soul Sauna Shogunate. What they couldn't have come up with a harder uh name to pronounce.
Other than maybe at I think finals last year, there was one team that their name was like I forget.
It was like five words or or just it was a sentence that we needed to say every time.
Uh >> Oh, yeah. Yeah, Super Nice Code Lovers currently on top of the leaderboard by a decent margin. Um but I mean, 25 games in, we still have what, 55 games left to go.
So, there's a lot of time.
>> I do think it's interesting though to see that like uh we have some um like null for example uh they were you know kind of in the you know lower half of the scoreboard in the first phase and they suddenly are you know in second place here and there a couple of others there this uh water cake thing it's one of >> [laughter] >> it's like it's a I guess it's like water paddler and whatever >> water or No. Well, maybe it's it's I I don't know. It's one of those combination teams again. We've seen them in the in previous DEF CONs. Uh anyway, also organizers kind of catching up a little bit uh here. Although they were doing fairly well in the phase one as well. But it's it's you know there is a little bit of a shakeup a little bit of a uh different uh order. Oh, we actually saw some teams jump up with one more game being played.
Uh wow, Super Dice Code Lovers is really ahead now. I think their score jumped up a decent amount.
Yeah, they got 115 in that last game.
So, they're doing pretty well.
I think definitely Digital wants to look at their bots when they're released.
Um and maybe just kind of like to recap a little bit for people who haven't followed you know we've been talking about all of these bots and and and so on. So so uh basically the the idea of this game is we have a custom architecture that uh um and you create a bot in this uh architecture to play this shooter game.
You upload it. It's uh you know put in a game against the other bots and then try to um kill them and score points in various ways.
Uh so that's what we're doing here.
>> Be friend them. This is a a third training program. Nothing malicious happening. The lasers are you know uh vestigial make a connection.
Uh yes.
Uh and so we've also talked about this like second phase, right? So uh the game progresses in six phases and there's slight changes to the game. So there was a the phase itself started now almost like 1 hour 1 hour 15 ago, but half an hour before that there was the new files released. Uh so they could kind of like look at the differences between the phases and uh think we have some like light notes on that. So basically in the base game there wasn't much to it apart from I said like you fly around uh you shoot or you know, whatever euphemism we want to use here. Uh other You can say shoot. It's similar to birds.
Yes.
Uh and you get points.
In the second phase uh there are kind of like two major changes. One is that there are now multiple weapons.
You know, I won't go into detail how that works or that's that's for something for the teams to figure out themselves, but there are now multiple weapons.
Uh and we also fixed a uh bug in the game uh where you it was kind of the classic like you move faster diagonally uh thing. Pythagoras cries.
Yeah.
Um so those are kind of like the the the tweaks from phase one to phase two. So nothing, you know, nothing major yet.
>> Yeah, like we're game devs here.
We make mistakes. Game code is obviously the most secure [laughter] code, but you know, we'll we'll patch we'll patch some of our bugs across the phases.
Don't worry about it. The game will only get >> Yeah.
There there might be you know, there might be some bug fixes. There might be some new features released.
Uh that's kind of Maybe. Just make sure that you're always working against the latest phase handout is what I'll say and make sure you check back half an hour before the next phase for the handout.
for it. Yeah.
Yeah.
Um.
I Is there anything else we want to cover at the moment? So, we will do one of these streams 1 hour after the end of a phase. Also like 1 hour after the start of of the next phase where we talk a little bit about what happened, what we saw, and so on.
Um, you know, if you have any questions or anything, feel free to put them in the chat.
Feel free to join the the Def Con CTF Qualifiers Discord as well if you want to ask us stuff like between the between the streams. We can maybe bring something up on the stream later as well. Yeah, if anyone has issues with live CTF, open a ticket. We're able to see those as well and respond to those.
Um, but yeah, other than that, I think Yeah, would love to hear if you know all the feedback and stuff from the players as well, but we will, you know, we'll come back to to that when we get towards the end of the end of the game. We still have the majority of this left, right? So, a lot of stuff can happen.
>> Yeah, we've covered the the scoreboard, but it's going to get shaken up a lot, I imagine.
Yeah.
Um, cool. Yeah, so hope everyone has fun with phase two.
Um.
Enjoy. Let us know if you have any issues or feedback with tickets and just in the Discord in general.
And other than that, I think we're we're good to end the wrap up of phase one.
Yeah, great. Enjoy everyone and see you later. Yeah, see you all later.
>> Mhm.
相关推荐
resume fixed instantly 😭 Comment “app”andI’ll sendyou the link #parakeetaipartnership #resumetips
Ritcareer
686 views•2026-05-31
Re: 🗣️📍theprophedu📍2026 GST 103 CLASS (E-EXAM REVISION)
theprophedu
636 views•2026-06-04
WEB TECHNOLOGIES UNIT-2 | Degree 4th sem BCOM Computers web technologies unit-2 full explanation💯✅
LearnwithSahera
1K views•2026-05-29
More tests are always better? How to use AI to identify tests that bring little value
Alliance4Qualification
335 views•2026-05-29
Search Algorithms Explained in 60 Seconds! 🤖💨
samarthtuliofficial
218 views•2026-06-01
Making Minecraft Clone with C++ & Raylib
PecaCSLive
686 views•2026-06-04
People of Game of Thrones using JavaScript DOM
AltCampus
296 views•2026-05-30
Instagram accounts got PWNed
EricParker
13K views•2026-06-03
热门趋势
Why Batman Lets The Joker Live 🤨
zackdfilms
9222K views•2026-05-30
They're Complete Trash
penguinz0
558K views•2026-06-04
The Murder of Deputy Caleb Conley
MidwestSafety
810K views•2026-06-04
I Bought FAKE HopeScope Merch (and paid a subscriber to give it a makeover) | Hopeful Hauls
HangWithHopescope
158K views•2026-06-04











