In the wake of current contests, observation has become very crucial in every problem, thinking from the problem statement and trying to reach the solution is no longer easy, because the crucial observation is no longer easy to arrive from the problem statement alone, I think we have to start from random points and see if we can reach solution from there.
Let's take today's C (Edu round 99), the crucial observation here is Bob can't win more than y matches, and there is a way for him to always win y matches, this was not easy for me, I saw the high number of AC's lowered my thinking scope, saw the sample, observed the pattern then tried to justify it. If there had been no sample, and I had not seen the submissions, I probably would have taken more time.
And unpredictability factor has increased, sometimes you get the observation in a few seconds, and at times you never get it, this definitely has impacted the ratings.
Any suggestions or comments on this are welcome.
EDIT : Some have misinterpreted my post, I am not complaining nor am I urging the problem setters to change the format, my post is more about how to tackle these kinds of problems, before screaming practice, add me as your friend have a look at the number of problems I have solved in the standings. And regarding unpredictability, I think it makes sense, even experienced coders at times have a blind spot, and relatively young ones may get that observation immediately, that's why I said it's more about the thinking style.
Would you rather interesting observations with "OMG i should have seen this" moments, or codeforces just being typeracer or history class? I think the elegance of the problems (and its unpredictability) is what makes CF and CP in general fun for people
.
Could you elaborate further on how adding an extra layer of DSA on the initial problems would help stabilise ratings? And why that would be desirable?
And regarding "Even if it's observation based": it isn't really possible to solve a non-trivial problem without observations, unless you have seen that problem and its solution before. Ergo, the 'history class'.
.
Why are you artificially valuing a simple implementation of a DS/A more than observation? Imo observation is not as random and luck-based as you are talking about, it gets better with practice and experience just like DSA. Besides, its not that every contest nowadays has A-C all heavily observation-based. People are way too salty nowadays
.
Yeah I was referring to the dislike bombing and general sentiment after today's contest, not you specifically. Also the author's post says 'nowadays', although I assume today's contest had much more to do with the creation of this post.
Simple advice, make yourself in div1. There you'd see problems that also need good DSA skills after getting the crucial observations.
You won't expect a problem which requires both clever observation and heavy coding in div2. Cause div2 is for inexperienced guys
Interesting how an 1100 gets to judge what it means to have a red rating.
You shouldn't say it like that. This kind of thinking demoralizes people from trying to improve and makes them feel lesser of themselves. Despite the fact that this person has a rating of 1100, the will to compete is still there and his graph shows improvement as well. If you want to know how strong an 1100 can get, see the graph rating of pikMike, the one who set the problems for today. I very harshly admonish such behavior of belittling people on the basis of their rating. I mean you no hate, just a simple advice is all I meant to give.
That's not the point I'm trying to make. I'm saying it's ridiculous for someone with little competitive programming experience to blindly claim that it's a crapshoot.
That's okay. I was a bit harsh and I am sorry for that. I did not mean you any hate. Sorry if I caused any anguish.
.
Going through your logic only tourist should be allowed to comment on anything.
He is referring to your comment on assuming your red color would be of no use. Only tourist should get to comment on what it is like to be a red coder? There are many other red coders who can comment on whether red color is meaningless or doesn't define some sort of skill, but you are not of red color, nor close...
Interesting how an unrated gets to judge what it means to have 1100 rating.
I never told it's not fun, my blog is about how to adapt to this style(which is evident from the title), as it's somewhat difficult for a few people.
lol, im not hating. I'm just explaining why this is the case.
Once i had a contest where my score was 63-30-100 and my friend score 63-100-47. I was so annoyed because i spent much much more effort solving problem 3 than he did on problem 2. But that's life, there will always be unpredictability.
I disagree as long as it's possible to guess the solution and get AC having no idea about the proof. And that was me in the last div1 contest, where I guessed the solutions to B/C without proving key ideas, those problems were "OMG i have seen this kind of geniosity pattern before but I can't prove it", not "OMG i should have seen this".
This is so true. You just enter the competition read the problem and think "waiiiit I think this is gonna work" and you either get absolutely unproved AC or WA on test 2.
Yeah, it can be quite annoying (especially as a problemsetter) if some guy manages to get lucky and solve a problem without proof.
But i guess there isn't much that really can be done? hahaha
I'd prefer maybe or maybe not running out of time on the last problem I was on and all following problems be like "OMG this is so op I would have never thought of this I need to practice more."
"OMG i should have seen this" problems are almost always unsatisfying if I solved it or annoying if I didn't.
thats exactly i how solved C in the contest ..I saw more submissions in C than B.plus .i saw one of my friend solve C in 7 mins..so i guessed it has to be something like a one liner..
Include B too. The observation if we want to decrease this sum $$$(1+2+...+ i +...)$$$ by $$$i$$$ then put $$$-1$$$ at the $$$i-1^{th}$$$ place. Anyway It's my weakness if I an unable to observe something and they are named as $$$Problems$$$ so the are supposed to create some problems!
IMO, these are really good types of problems for code forces. They require us to think, make observations (maybe shouldn't be easily deductible from samples, like in today's C, ans = a — 1, b), and then construct a solution and not just throw the inputs into a black box algo/ds.
That being said, maybe this question wasn't suited for an "educational" round.
Hot take: an entire contest filled with problems with "interesting" observations is not fun.
Not just today's Edu rd, but several of the past CF rounds have not been enjoyable for me. It doesn't even matter that later problems are interesting, because the early ones are always "do out some test cases, find the pattern/observation, and write five lines of code," and they leave a shitty first impression.
Also I don't understand why everyone labels problems not of this nature as "boring" or "typeracer." Maybe I'm just a noob for thinking that figuring out the parameters for DP or reducing problems to graphs is interesting...
... oh wait, the primary audience for Div 2s are people around my rating or below. Hmm...
Yes, it seems their line of thinking is like this "let's avoid graph, dp, binary search, data structures and the like, these are all uninteresting and too standard for div1 and become mere typeraces for them, instead let's make whole rounds of adhoc/observation-forces and give them to div2, afterall these are more interesting to div1".
Literally all of those topics you just mentioned except for graph appeared in today's edu.
You looked at the 3 easiest problems in 7 problems contest and decided that you know everything about the contest. I don't even know what kind of graph problem are you expecting from div2 ABC level. If you want to face those problems in contest, stop getting stuck on one liner problems and get good enough to reach those problems.
Yeah, I only looked at the 3 easiest problems, I was stuck on them for 3/4s of contest time. After that, I was like "meh, I don't care about this". What should I have done? Go to div2 E when I'm stuck in div2 B?
I don't enjoy solving these adhoc/one-liner problems and I don't want a barrier made of them stopping me from getting to the interesting ones. Maybe I should just quit CF then...
Whether you enjoy CF or not is not related to the argument. I was just arguing against your point that div2 doesn't have those topics.
If you want to solve them in contest, what's stopping you from jumping straight to div2EFG? If you care about ratings, just solve them outside the contest. Algorithms/DS problems are hard. That's why they're placed on the back. If you can't solve the easier div2 problems fast, you won't be able to solve the latter in contest time anyway unless you improve.
Do they need to be though? I have never noticed this fact when solving problems in gym. Look at most ICPC contests, they have all kinds of problems with all kinds of difficulties.
Why do we need ABC to be completely different from EFG? What's the point? Why not have separate contests, one with ABC type of problems and others with DEFG type?
Of course it is, this whole blog post is about people who have not been enjoying the last CF rounds...
ICPC and CF have completely different format. I'm not sure why are you expecting to see an ICPC problem in a CF contest.
ABCs are not different from DEFG. The only difference is that you've solved ABC and have not solved DEFG. The thought process needed to solve those merely gets more complicated on latter problems.
Why are we even throwing one liner problems in the front as preliminary barriers anyways? Do high-rated people enjoy them? Do low-rated people enjoy them? Who enjoys them?
Well I don't really know for sure but I presume high-rated people don't really think about whether they enjoy a problem which they took 5 minutes to solve.
Also, who told you they're the "barriers"? Are there any rules that you shouldn't solve latter problems without solving previous problems? They're placed back because in general, people who can't solve easy problems in front can't solve the latter problems during contest time.
Obviously no explicit rule, but it's undeniable that early problems are visited by most participants first and make up the majority of people's impressions of the contests.
Ultimately, a lot of people seem to hold the opinion that contestants' frustrations with early problems don't matter, and I genuinely cannot understand why. I've seen so many comments on CP philosophy and what the attitude of contestants towards early problems should be, but very few genuinely defending ABCs of recent contests as good problems for beginners. Imagine if game designers just threw shitty levels at the beginning of their game and told the player they have to get good to reach interesting levels. Yes, high-rated people will probably never find ABCs enjoyable, but if low-rated people don't either, then who are these problems here to serve?
Yes. So there is no such barrier in the first place.
I don't believe this is true. It's just that earlier problems tend to have frustration from unexperienced users.
They dont defend it cuz it's not true. CF contests aren't made for people completely new to CP.
Some, not all. And some people will always get frustrated no matter which problems are in ABC
I enjoy them. I find it charming when something hard may be simplified very much. But I usually suck at this kind of problems, but it's only me who I can blame.
Why would you hate a probelm you yourself authored :/ Oftenly a problem from edu A, B, C belongs to you :/
It might come as a surprise but creating problems is in itself an incredibly hard problem. There always will be frustrated contestants: some dislike tricky implentations, some dislike one-liners, yada yada yada. You should also note an average participant nowadays is much more knowledgeable than before. What are the chances that the cool DP/DS task you came up with appeared somewhere before? Low? If that's so, I can certainly say your task is too hard for div2B/C.
All things considered "ah ha!" tasks are in a pretty good spot. If you don't solve you will be frustrated (in a good way) otherwise you spend few minutes and move on to the next task. Not every task ought to be mind-blowing.
Well, I agree that today's C is not a good problem. One can just look at samples/standings and guess the formula. This should probably be avoided in contests with open standings.
Wait, isn't Div 2 meant to be for up to 1900 rating primary?
And also yeah, I totally agree with what you are saying. But this is CF. It is a competition for speed, fast thinking and intuition. And still, altough again heavily observation-based, there are problems which require graphs, dp, data structures, common techniques and so on.
Oh and right, the last few rounds were really dry and I am kind of happy I was busy and couldn't participate.
Yeah that's fair. I guess my comment is more of a reaction to recent contests and might be a bit too pessimistic concerning CF as a whole. I have nothing against any one particular problem, just wished there was more diversity in the entire overall package.
Well, I get a lot fun in "interesting" observation problems, the more problems, the more fun. Yes, I'm a div2 participant and there is a lot of them like me.
Yeah, I'm not really good at algorithmic problems that I'd tag it as boring, but yeah, there's something more funny and something less funny. I don't get the feeling that "Oh that's insane" in most algorithmic problems.
Anyway I'd get a lot of fun in this type of algorithmic problem where you need crucial observations. 1363E - Tree Shuffling
Anyway, there are adhoc problems which would easily beat even this algorithmic problem in terms of being interesting, like 1417D - Make Them Equal
I absolutely agree there are good ad-hoc problems. I just prefer things in moderation, not too much of one type.
To be fair, I'll give an example of a problem I'm not really a fan of and why I'm not.
Today's Edu Rd, Problem B:
You need at least $$$x$$$ moves such that $$$x(x + 1) / 2 \geq n$$$, and $$$x$$$ moves is always sufficient except for when $$$x(x + 1) / 2 = n + 1$$$, in which case $$$x + 1$$$ moves is needed. One way to reason this out is to think of all $$$x$$$ moves initially dedicated to moving forward, then we choose some subset of our moves and change them to $$$-1$$$. If we change the $$$i$$$ th move to $$$-1$$$, then we subtract $$$i + 1$$$ from our total. If $$$x(x + 1) / 2 = n + 1$$$, then there is no way to pick a subset of moves that subtracts just $$$1$$$, so we need one more move.
I mean it's alright. I found this solution just by poking around with some test cases, coming up with a claim, and then verifying that the claim is true.
After sweeping through announcement comments and hearing from others, here's the general consensus I seem to have gathered:
High-rated people: solved it quickly, erased it from memory and moved on.
People who didn't solve instantly but still solved it: maybe solved it after doing out some cases/kinda proved it.
People who didn't solve it: varying degrees of annoyed.
Importantly, I couldn't find a single comment in favor of B or stating that B was enlightening. It wouldn't be such a big deal if problems like B happened occasionally, but Educational Rounds A-C are similar style problems almost every single time. After seeing the 699th ad-hoc/greedy problem, it's not "wow another exciting observation!" but rather "here we go again." And I feel it's more important than others that A-C is varied because these are the problems that almost every single contestant comes in contact with.
Well, what's the problem, this is a nice idea. This is an easy div2B idea. We get tougher B most of the time.
For me, I logged in after 10 min of starting of contest, so there was no point of doing that round where you'd surely get killed, I was just looking through B and C, after getting B and C, I was raging for being late.
Yeah, I liked it. I wasn't quick, it took 5 min, then I said, "Alright. It's very cool, how you get such good simple ideas?"
Although not high-rated I wanted to erase it from my memory as fast as I can and now I am still wondering if my solution is legit.
I feel like this is the style of Codeforces — light, dynamic, observation-based problems which do not require hard algorithms (at least not up to mid to high Div1 I guess, never been there). Yes, the observations are really hard to get but that's the point of it. Altough it's really frustrating when you finish the competition, read the editorial and think "Oh, I should've seen that!", it is all worth it in the end when you upsolve and get that AC.
I almost took 1.5 hours and was not able to solve it . Reason for this was that i was thinking more complicated than required . I am going to get around -100 but i think contest was fine . Instead of blaming and down-voting the contest we can learn to think simple. One thing we can improve is that Educational contests should have some algorithmic problems , but it's not always easy to prepare tricky problems out of well known algorithms .
Also it's never easy to make problems and problem-setters don't have any motivation to make bad problems.
In last contest also , problem-D required very simple observation : Either three consecutive number will have same highest bit or size of array will be around 60 . I always find that most of the time i am not able to solve the problems due to thinking more complicated than required.
I might get a lot of downvotes here but imo these problems are not suitable for a programming contest. Like i dont get the fact that how is printing a single line of answer is of educational purpose. If we talk of div 2, how is a dp, graph or any othr dsa problem less interesting than these adhoc problem. Like if I am reading a problem C of a contest, I do have a mindset that this problem might test my actual algorithmic skills, I know I suck at these pattern problems but its been quite while now for me doing cp and all I have learned out of these type of probleme is keeping hitimg random things until you get the solution or be ready for a -120 coz its a very trivial problem for 6000 other participants
When were contestants ever able to predict their delta? I don't see any reasoning in your claim that unpredictability has increased.
Are you seriously expecting some data structure/algorithms on div2 C? Were they prevalent on div2 C level problems on past day?
I also find it funny that you made this post off of a single div2 C problem.
No he didn't make it because of the single div2 C problem, it was just the culmination of rounds and rounds of this happening.
It's not bad, the problems aren't bad at all, they are extremely enjoyable and interesting but it starts to get annoying. At least for me nearly everything I solve on div2 A, B, C problems is by intuition and not much thinking. If you haven't seen the something similar to the given task, you have absolutely no idea where to start from.
Again I want to say that these problems aren't bad, they are amazing but the div2 contests are flooded with them right now and you can see that some people are getting frustrated.
I see people getting frustrated on cf all the time after they lose rating, I feel like this time, they're blaming on "contest trend" whatever they think the trend is. Not like authors have a "setters' network" where they agree on specific trends on problems.
I still don't recall when was div2 ABC not like this.
Yes if you don't practice CP, you can't solve CP problems. What a surprise.
All the algorithm/DS problems are placed later in Div2 because they're hard. Most of the contestants don't see them because they aren't good enough to reach them. Even if they've reached, the majority of them won't be able to solve them. That's why they're placed on the back.
Nobody said anything about bad performance. People are talking about the contest problems and how they feel about them. Is it forbidden now?
I haven't been here long enough to talk about deep past div2 ABC but from what I've seen, the type of the problems was always pretty much this but lately it's been getting more like "do or die" or "do the observation or you lose all the rating you've gotten from the last 3 rounds".
The algorithm/DS problems are the core of CP. When you get to that point the real grind begins. But when I start a contest I don't want to be stressed about the observations on tasks ABC. And now this stress comes from the fact that they are unpredictable. Maybe you don't feel it because you do only div1s right now but believe me — it's really annoying.
I don't recall forbidding anything. I was just saying that people get frustrated over contest all the time.
Still waiting to see a reasoning on why this is a "new trend".
That's very controversial.
Your rating unpredictable by nature. If your rating is 1900, you have about 50% to solve a 1900 rated problem. All that you can do is reduce the probability of failing to solve a problem by getting better. This "unpredictability" has been there ever seen the cf was created, and is not related to the "new trend of problems" that people are claiming to feel.
Yes I don't feel it because I have relatively higher chance of solving those problems, not because I don't participate in those contests, I do participate most of them in fact.
Yeah sorry about the last one, I forgot that you can still participate.
I won't dispute with you anymore but just think about it. A mass of people all around the world suddenly start to feel a change in the problems. There are more and more blogs about it and people are getting frustrated, although they solve the problems.
I don't care how they call it "a trend" or something else. But it is a thing and it is here.
I don't see any reasoning in your claim that unpredictability has increased.
While I don't think there is any solid reasoning in the op's post, for me personally I am much more worried in div2 rounds about being able to solve A, B, or C than D or E on average, as I am more likely to spend a longer than normal time on them and mess up the rest of my round's pacing.
I think dumb one-liners are easier to overthink and waste time on, especially the higher they are in the pset since ur not expecting such a simple solution. Since there is very little to the solution I also find it harder to "work towards" the solution because it is more about having a single "aha!" (or for me proof by ac) moment that playing around with small examples doesn't help as much if you don't see it immediately. This is in contrast to a harder problem, dp for example, where i first find some brute force transitions, think of some states, realize i can drop one out, realize a transition can be made faster with some range query, and finally can put together the full sol. So at least for dumber noobs like me it at least feels more unpredictable.
Sedmoklasnikut's sentence above accurately describes what I mean: At least for me nearly everything I solve on div2 A, B, C problems is by intuition and not much thinking. For me personally this isn't very satisfying and doesn't feel like an accurate show of skill, though I know some people like it.
IMO, I would rather prefer to try find an observation for 40 mins and then code it in 5 mins, than knowing how to solve immediately and then spending the whole round trying to write a bug free code.. and if problem C is a very straight forward algo/DS problem, it will become too easy for the participants.. and if the writers try to make it hard by adding something not very intuitive or something tough to code (like Round #659 or Round #684(div2)), participants complain about the difficulty.
P.S. it might be an unpopular opinion, but i like the rounds which require more to think and less to code, and I like contests like today's (C and D were great problems).
I think people downvote contests if it does not feel fair. In this sense the unpredictability is a problem. A boring segment tree with a lame twist might not be very sexy, but it causes less downvotes than todays C.
Today's contest felt like an experiment on the setters behalf. It seemed as if they wanted to see what would happen if we give a contest with just observation problem. The D was kinda misleading as the constraints hinted at DP whereas it was easily solvable by n^2 approach. During the contest when I got WA with my n^2 approach, I did not go back to it cause I reasoned that it couldn't be this easy. So I started to struggle with states of DP and couldn't get AC. So according to me, this round was weird, as a few of the last few CF rounds been. I miss the contests where I had to figure out graph problems or think of DP and states..... Nowadays I have started solving the contests from 2012-2014 which had classical problem variants, just to battle the disappointment from these CF rounds.
Have people already considered a contest format where you could not see how many people solved a particular problem? This way, people could not deduce from results of others that problems like today's 1455C - Ping-pong is a one liner.
I think such format would be better because knowing how many people solved a problem is a huge hint to how complicated a problem is.
You can’t solve it without thinking doesn’t mean a contest is bad, instead it means the contest is good.
Copy my post, paste in notepad, search for negative adjectives like "bad", "worse", "not good", "poor", let me know the number of occurrences of such words.
Huh, my advice would be practice harder problems than you currently do.
sidenote:
i hate these problems where you have to get that one observation and then it becomes trivial and if you don't it's basically gg,
i hate these problems where you have to make some wild unproven guess and submit or be overtaken by 2000 guys who made some wild unproven guess and submitted while you were trying to prove it
Exactly! This makes rating gain/loss so inconsistent. I feel like I know too much extra than what was required to reach CM.
It's so weird, but seriously nowadays one can reach CM or even master maybe without even knowing what segment tree is. Just a few lucky observationForces contests.
From my CF days, I guess thinking is what matters nost. Yeah, I reached CM without ST. But, observation is the funniest thing in CF for me. There is other thing, let's put a ST problem, you won't put a simple RMQ, you would put something where you need to think, you need to observe. Let me now if it's really suitable for anything upto div2D
Yeah, rating curves look funny, look at mine for example
I reached master without knowing segment tree 7 years ago. Then I learned segment tree and got to red. It doesn't work the other way around though. If you can only apply data structures and standard algorithms you are not a good CPer.
How important is algorithm knowledge at your rating?
There are problems which cannot be solved without advanced data structures, sure. There are much more problems (I want to say all of them but that wouldn't be true) which cannot be solved without thinking.
Isn't "thinking" or more appropriately "observation" very hard to improve? I mean one can definitely set a deadline and try to cover(read) all advanced data structures within that time, but one can definitely not do that for observation.
"If I get 1 year of training, I'll read all the advanced algos and ds"
vs
"If I get 1 year of training, I'll start making these observation naturally."
Does that mean we can conclude that , given a fixed amount of time (like 3-4 years) it's certainly guaranteed that some of us won't reach our target goal (like red)? :(
The motivation I got from anime is shaking at it's root nowadays.
I'm not sure. I think that just by solving a lot of good problems you acquire experience and learn to recognize patterns which is the basis of "observation". For some reason people think that if something can be put in a concise lecture or a book it is a more valuable skill.
You probably can read the book with all the advanced algos and ds but it doesn't mean you'll learn how to apply them to problems.
How confident were you when you started CP? I mean , did you start being fully aware of the fact that one day you will be one of the best?
Of course not. I just liked the process. I still do. If one day I'll understand that I don't enjoy it anymore, I'll stop.
Don't judge me XD, here take my "Orz".
In my opinion, ad-hoc observation-based problems are kind of what gives CF its charm. They tend to be super open-ended and actually fun to think about
on the contrary, problems which don't contain observations frequently lead to moments where I'll say to myself "well shit, I've completely ran out of ideas and it's only been 10 minutes." this is especially painful in Codechef long challenges, where I get stuck on day 2 and then have to wait an entire week to see the solutions lol
In this type of problems you need kind of guts..if you are good, you would to prove it before submitting and check the edge cases ..while someone with not much thinking ability might simply type it in the guess and get lucky..for example in the first problem one could guess it's equal to number of digits from sample cases..but it takes a while to prove it..but lot won't care to prove it in their head...they would rather happy go lucky..that said these problems are definitely enjoyable and thrilling
The ppl who solve problems only on the basis of guesses don't have a concrete performance graph... So chill bro ... In the long run, it doesn't help...
like me (:
Well its opposite for me lol ... Each time there was a observationforces .. Its guarenteed I lose atleast a 100 of rating. Just see my contest page smh
I was able to prove the first task on the competition but on the third I had no idea why would the solution work. I just saw the test sample and thought "well, high risk high win" and submitted and was left jaw-dropped when it gave AC.
One good thing in that type of problems is the joy of discovering the observation. Also it may be better for developing problem-solving in general, not just in the context of competitive programming , but , in real life, or even real research problems you usually don't know exactly what you should do. Instead, you just keep exploring in organized way.
Regarding how to solve it, I think one need to carefully analyze different parts of the problem and related issues in different ways, concretely, graphical or mathematical, without sticking to one path for too long. So, a period of good exploration and brainstorming should result in many observations, which will interconnect and then lead to more observations, and so on.
To make your exploration effective, One very very valuable skill that can be developed, is asking the right question. This saves me a lot. It certainly improves with practice and more exploration.
For example, The problem you mentioned, all I did is just analyzing some instances of the problem, then the idea pop in my mind that bob can win one game with every single unit of power he has. so, the question is are there many different ways to reach this result for bob, some of which may be better than others for Alice? easy reasoning leads you to the fact that there is only one way bob can achieve his optimal result. so..
On the other hand, I struggled with Problem B in same contest, because I took many bad paths and ask many wrong questions.
Logical problems which require critical thinking should be the best choice for contests than DS/algo problems which require nothing but knowledge. CF is doing good these days by bringing such problems in contests. I like atCoder for such problems.
Thanks. You made me feel, I am not alone struggling these days.
Observation-based/ ad-hoc problems are either really cool (when you have that "Eureka!" moment) or very annoying when you realise it was just a simple observation which you missed, so it's no surprise that the CF community is kind of split up regarding the question of whether there are too many ad-hoc problems in div2 rounds. I myself got quite angry because of some problems of this type and I would prefer a bit less ad-hoc/observation stuff in div2 A-Es, but I think there really is no good solution: many good problems involving interesting CP algorithms/data structures are just too difficult for the first 3-5 problems in a div2, so less ad-hoc would more likely mean more standard problems, which is much worse than just an observation-based task.
I only learn a DS or Algo when I encounter it while solving. Due to such contests, its been a while that I learned any new algo but then again these kinds of problems make me realise that I am simply not good enough for them yet. Kinda like if u are nothing without Algos, then u don't deserve them ;_;
As a coordinator of the round, I can say the following:
Ad-hoc problems will continue to appear in ER A-C, but I agree that yesterday's round was too overloaded with them. I assumed that A and D would be less ad-hoc (for D, if you check the editorial, our main solution was two-state dynamic programming instead of greedy), but it turned out to be false. In the next rounds, I will try to limit the number of ad-hoc problems more strictly. I am sorry that my wrong expectations about how people solve the problems made the contest too boring for some participants.
But I don't agree with the people saying that at least one from A-D problems in each round should contain some sort of data structure. First of all, it's bad to have complicated data structures like segment trees in problems before E slot, so we are somewhat limited by the standard data structures (for example, the structures from C++ STL) — and even those structures are better suited for slots C and D, in earlier slots they can make the problem too difficult. And having a data structure exactly in C-D slot in each round is kinda boring. We will try to make the topics of the first four problems more diverse, but it doesn't mean that each round should (and will) contain a data structure problem.
In my personal opinion (which may differ from the opinions of other authors), data structures and algorithms are too overrated in competitive programming. Most ICPC subregionals/regionals contain much more thinking-based and observation-based problems, and for me, in most cases when I couldn't solve a problem in an official competition, it was because my lack of observation skills, not because I couldn't implement something. Furthermore, since most CF rounds are conducted online, all participants can copy-paste the implementations of various data structures, so problems should be a bit more than ``write some data structure and make trivial queries to it in the main program''.
So, I think there will still be ad-hoc problems in ERs, but I agree that there were too many of them in yesterday's round, and we will try to avoid repeating this situation.
I agree that Ad-hoc has become a very important component in problems nowadays, and I am not even against it, but some "observation" problems require some amount of "luck" as well, people say with better practice comes better observational skills, but I think, still, there will be a factor of luck. There are some clever observations, but many are just about pattern finding. And the thing with observation is you may or may not get it, in round 682 I solved 3 observation problems in 12 min, yesterday B alone took me 30 min, because unless you draw samples for n at least up to 7, it is difficult to see the observation.
And there are Ad-hoc problems that require a good amount of analysis as well, where we have to break down the problem, form conclusions, prove it, and stuff, I would prefer this over one major yet not so obvious observation for a problem. Ad-hoc's are fun but I hate some Ad-hoc where we can see the observation without any analysis.
yes, when i was in university, and did codeforces, i remember that the tricks were straigtforward, nowadays is more constructive kind of problems or observation based.
I may be stupid but I solved that problem by just brute forcing the game and looking at the output