I’ve been seeing many people I know talking about how LLMs have "killed" CP. People are getting frustrated seeing leaderboards, and asking what is even the point now.
Lets take a close example: Chess after Deep Blue: In 1997, IBM's Deep Blue beat Garry Kasparov. It was a very shocking to those chess players (first such major AI like event ig), there was also similar discussion if chess was dead then as well. Today, any free chess engine on your phone can obliterate the reigning World Champion. After soo many years did chess really die? No, it became popular than ever.
Why? Because humans don't play chess to prove they are better than Chess engines. We play to beat the other guys, and more importantly, better than we were yesterday. CP is following the exact same trajectory.
The Harsh Truth: Online CP is a Gym, Not a Stadium Let's not sugarcoat it: Online CP, as a perfectly fair competitive arena, is effectively dead. With the current state of AI, online rounds are infested with automated solutions.
We need to accept that online platforms are now strictly for training. They are the gym. The actual "competition" will inevitably shift toward on-site, proctored environments like ICPC or strictly monitored finals. Online rounds are just sparring sessions to build your logic and speed.
Should we stop? Obviously not. If you are the type of person who would have done CP in the pre-AI era someone who actually enjoys the adrenaline of a green Accepted and the process of breaking down complex logic there is zero reason to quit.
If you were only in it for a superficial online rating to slap on a resume, then maybe it's time to move on. But if you are here to build a sharper brain and write highly optimized code, AI changes nothing about your personal growth.
As for dealing with the cheaters online, I keep my motivation simple:
I just tell myself: They literally need a supercomputer to make it a fair fight against me.
Auto comment: topic has been updated by MIRZAPURI (previous revision, new revision, compare).
informative blog, thank you
It's just frustration Bro!
truly insightful, thanks
Very interesting and informative blog. A remark i have is that in my opinion, i do not consider chess and cf a fair comparison. For one, there are many cases in of which one might have a motivation to cheat in cf beyond simply winning, by that i mean as you stated it can be something you can put on a resume whereas in chess one has no exterior motivation for cheating beyond winning. By this, one cheating in chess is for one less likely to happen and also one cheating in chess is likely to very quickly stop. I am or was a fairly avid chess player and i believe i encountered cheaters very rarely and i don't think i ever played a blatant cheater. Your insight at the end was particularly inspiring.
I agree as well, that is the most problematic thing.
But still, saying "CP IS DEAD" is completely wrong, like atmost it may become more offline kind of thing.
Anyways, its just frustrating for me to listen to guys who don't like CP and always say CP is dead without any reason!
I guess CP is just a bad name. CP, or at least its problem-solving subset, can be non-competitive: if it is just about problem-solving. There are tonnes of games on Steam that is easily solved by machines yet is played.
Think Turing Complete, you can do it if you learned CPU arch in college, but it is still fun to play.
I don't have any idea what CPU arch is so I am gonna have to wait ig. Can you tell me like in when I might encounter this?
It's the course that discusses how CPUs are built, their instruction sets, their internal structures, and basically how to use logic gates to trick silicon into thinking
There is a course COA (Computer Organisation and Architecture).
Guess I will learn it in that!
Is it interesting?
Or purely too much theoretical?
Honestly it depends on how you learn it. Most schools would include at least some level of practical training using Verilog to write CPU designs that you run on FPGAs. That would be fun if you like such excursions.
I am kind of excited atp!
Oh, and the game Turing Complete has a Verilog export feature. You know it's a good game when it has one, so play it!
Thanks, I will surely try!
You can try games like it! I played https://nandgame.com once and I loved it
What is it like?
It's quite fun, I recommend you try it, It's hard to describe the experience, everyone differs, but in it you start with literally nothng aside from on/off gates and build up to a whole computer, assembly language and higher level language
Still motivated. Still learning. Nice blog.
You must love CP a lot!!
Yes. I do it for fun.
"They literally need a supercomputer to make it a fair fight against me" hits hard
Indeed, it always hit hard.
well first of all LLMs will not be free forever. as much as it feels discouraging to see everyone be able to download claude and put in a problem and see it solved instantly, every single prompt costs and slowly but surely this won't be as available as it is. also to me, online cp was always about training for the real life competitions but I guess that might not be the case for everyone. but even if it is not the case for you, if you start seeing it that way, CP loses no value due to ai and you can even use it to your advantage if you apply it properly
+1
Honestly, for me CP is more about how you can solve a never seen before problem (kind of), maybe that is also why I don't quite like Leetcode.
More or less, I think the same as well.
I agree!
I think it doesn't matter either tbh. Even if AI gets better than BenQ (maybe?), even if it replaces every programmer on earth (unlikely), CP still won't die, just like how football won't die even if a football playing robot that could beat a team if the best players existed
Problem is we are going to be the ones suffering in this transition phase!
Chess become popular and even better after new IBM invention, because people used it on first place to train, see their mistakes and learn from them. On the other hand, in Codeforces, LLM's are currently being used to gain unfair advantage in contests. This will never change, unless people stop using it as rating gainer tool.
But, it's always been offline things in CP which have most value, no?
$$$AI$$$ title, $$$AI$$$ post, $$$AI$$$ conclusion, your comments are actually how you really talk. why not just write your blog in your own voice, especially given its content? trust me, your own voice & opinions $$$ \gt \gt \gt $$$ whatever nonsense you tell some $$$LLM$$$ to spit out
If you search on the Google for most of the information, it would sound exactly like that, and if I know that these words fit the phrases that well, then why would I want to change that much.
For making a blog, i prefer to search to not cause any misinformation. If the sentence sounds AI like just because its good quality, does not mean I will compromise with the quality itself.
in my opinion, on paper, the sentences are fine, but like when you are reading them, you can tell that there is just no intention behind them. it reads like, well, an $$$AI$$$ wrote it.
Anyway, in my original comment, I could very well be wrong. maybe we are entering a new era where most of the posts on the internet are largely written by $$$AI$$$, and no one really cares anymore.
I don't know man, trying to make blog more human like by degrading it's quality is quite annoying to me, as I really hate when you do your own assignment and we try to make spelling mistakes purposely just to make sure AI checker don't mark you even when it written entirely by you, else you will get that null marks by your professor.
It's just annoying and I don't want to do those silly things in here atleast.
And yes, LLMs are shit, because I know google will give better information always.
Totally agree with this. Honestly wish there was an extension to just hide the rating number altogether — that colour and number staring at you messes with your head sometimes. I'd rather just track solve count in contest and try to beat yesterday's version of myself. Competing with friends is where the real fun is anyway
bros got a british $$$LLM$$$
dude I don't think anyone can outcode the top coders bruh look at tourist's videos he solves the questions in like 2 seconds
Hold up!! His writing is this FIRE??? (It's a meme, if you know, you know)
Do I need to explain it again?
Just read my conversation with 123gjweq2 above.
I am sorry if I have actually misunderstood as I don't know the meme.
Yeah this is pretty much what I've been telling others I know who are a part of the scene. Online sites have always and will always be forms of practice to self-improve.
Sure, it may feel nice in the moment to see a high ranking, high delta, etc, but those are really just eye candy. Stripping those away shouldn't be all it takes to discourage someone from wanting to get better at problem-solving, pattern recognition, math, algorithms, and more generally competitive programming as a whole; if it is, then said person was probably in it for the wrong reasons in the first place.
Plus, in-person contests still serve as a good way to have healthy competition even in the current era. The only thing that I think is lagging behind in the space are multiple in-person competition options for people past college.
well actually chess engine on your phone can't guarantee a win, even top level engines can not guarantee it . there are multiple streams of grand master going against stockfish and drawing the game .
It's actually that, chess engine are not gonna lose, at most you may draw. Like in any competition, you are never going to win against an engine!
Theoretically, you might draw, but never going to win!
A perfect game of chess is always a draw sir , what more do you need from a person then a perfect game . I think there are also cases where people did even win , but i haven't followed chess too much recently so i can't be sure if someone bested the latest model or not .
well actually you can also win if the engine is playing black , since black has a disadvantage and any mistake can cost the engine the game .
Even if White has a slight first-move advantage, modern top engines are strong enough to consistently hold draws as Black against any human. Also, while it's widely believed that perfect play leads to a draw, chess hasn't been solved yet, so we can't say that with certainty.
I got this while searching.
Most chess engine can remove that advantage if you are giving them infinite time , chess engine are extremely slow when they are looking like 100+ moves ahead during the most important times .
That's the thing, Engines like Stockfish or Google's AlphaZero operate at Elo ratings well beyond 3500 which way more than enough to beat any human.
Magnus Carlsen (considered to be best chess player ever) himself said: "I have no chance against my phone". He is not even talking about something like Stockfish (That says it all).
Link
sorry but the article you cited itself says he was able to beat low level engine using a technique he refers to as “anti-computer chess” and it only doesn't work for strong engine.
so idk what you are trying to say , your source is contradicting your statement here .
You are referring to this part:
“Actually, one time I played at a corporate simul, and there was this guy who said, ‘I built a chess program in my university class. Can I let that play against you instead of myself?’ And I was like, ‘Yeah, sure, why not?’”
You are literally referring to the most weak chess engines which literally only calculate to 6 to 8 piles (3 to 4 full moves,), while engines that we generally talk about calculate like atleast 40+ piles.
Read the whole article, him saying that he beat only that university project and feels that normal engines(atleast which mobile games use) are unbeatable for him, not to mention stockfish kinds!
And it been mentioned way too many times by almost every GMs that an engine is an unbeatable thing!
And my point with my source was trying to give the whole picture even someone like magnus has only managed to beat only university student's project chess engine only, which is like on the very weak side of a chess engine.
sorry but i didn't get where you got the phrasing " him saying that he beat only that university project " , the article states his technique “anti-computer chess” only doesn't work for really strong engine . in the line "That doesn’t work against very strong engines, but it can work against weaker ones.”
also even weaker level chess engine can look upto (assuming they didn't use bitmaps and if they did at least double the move count) 15 to 18 moves in future so idk where you got the 6 to 8 number .
the start start of the article joe rogan asks him "whether he could defeat his phone running at the hardest chess level." and to that statement he replied yes his phone can beat him . he never said any phone engine can beat him and assuming every grand master mostly used chess.com and its hardest bot is stockfish at 3500 , so he was referring to the strongest chess engine out there maybe second only to reckless .
so no he didn't just beat university level projects
He is literally saying what he beat what a university student project which is incomparable to real engines and I may be wrong about the exact numbers but my point is that engines like stockfish are unbeatable.
If you think they are beatable then I don't know anymore than this.
After late 2000s, there has never been any instance where a human beat an engine like stockfish in a fair match.
My point is not that every 3rd rate engine or AI is great! It's the really good ones which are great, which happens to be quite accessible to many.
I am not here to say that weaker AI or engine are unbeatable!
This lowkey motivated me, thanks!