Super Fast Brainfuck

[tog]

Good God, do I have to beat tail's solution to reach that city? This seems to be very hard for a challenge which promises to be central for follow-ups.

And this is again a point where I really don't like the king-of-the-hill system: If I try hard and eventually beat tail's current solution, he definitely will try and most probably succeed to beat mine. So I reach that city, but it's only for short. I only have access to follow-up challenges until someone beats my solution, which is kind of awkward to me. And what if I quickly solve a follow-up before my solution gets beaten?

The same problem also holds for the forum: It's quite irritating to have only temporary access to king-of-the-hill-solved topics. Eventually your own post will block you from that topic because it inspired others for better solutions (which otherwise is a good thing). In the end, this prevents people from discussing their solutions in the solved-forum.

I understand the motivation for the king-the-hill-system (not to have endless chains of new challenges resulting from new records). But I think it shouldn't be employed so inflationary. After all, kind-of-the-hill challenges will always become very difficult.

And speaking of king-of-the-hill challenges: How is the score handled for these challenges? If my solution gets beaten, do I also loose the points?

---
I just realized that post became quite OT, so bear with me!

[gfoot]

I half agree, but so far it looks like adum has been careful not to make anything else depend on solving KOTH challenges.

The forum issue is more real though - early solvers have an advantage because, with poor solutions, they may have got access to hints from other solvers which latecomers won't see. The main thing I don't like is losing access to threads I've already posted in - it's made me not want to post at all about KOTH challenges any more because that's frustrating.

I don't have a solution to that, except to not discuss KOTH challenges at all. However, I like the concept and am enjoying solving some of them.

Regarding score, I've been criticized before here for apparently caring too much about it. I honestly don't really care much, but it is a motivator and I do think it would be nice to get partial credit for passing milestones in KOTH challenges, equivalent to there actually being several challenges with different targets. I don't think the first people to attempt a challenge should retain any special credit - so early solvers' scores should reduce over time as the benchmarks move. Maybe the points for these challenges could be partially awarded to 2nd, 3rd, etc places in the same way as they are for the puzzles.

[tog]

I half agree, but so far it looks like adum has been careful not to make anything else depend on solving KOTH challenges.

Then the lower left will be the first city that's a deadlock ...

I don't think the first people to attempt a challenge should retain any special credit - so early solvers' scores should reduce over time as the benchmarks move. Maybe the points for these challenges could be partially awarded to 2nd, 3rd, etc places in the same way as they are for the puzzles.

But then again, early solvers would have an advantage, because once a new record is set, they still keep points for a solution that wouldn't be accepted from later solvers anymore.

[adum]

i hear your concerns.

i didn't intend the fast brainfuck to block a city -- it won't be a choke point. none of the KOTH will be.

i definitely agree that the KOTH is pretty severe. especially with tails competing =) i needed something to stop the overnumerous challenge sequences questing to get ever smaller or faster. and to clear any mystery up: when your solution has been beaten, you don't get points for it any more.

there are two mitigating factors:
1) there's a window other solvers can be within to get credit
2) i'm planning (but haven't done yet) to add 'middle' challenges that require a certain set speed. i don't want to do this too early, because i want to see what the range of solutions evens out at. for example, for fast sort, i'll probably add a 'solve in 5k cycles' range. i don't want to make the KOTH windows too big, because that removes some excitement.

for the forum posting -- as solutions improve, the level of discussion will go up. if old solvers could still see it, they'd possibly get hints. a good example here is Super Small Hello World where the people who have gotten 29 have discussed how pretty openly.

[tails]

Only about the city: The "String Reversal" challenge is actually playing the roll of a city, so how about to move it there, to clarify the intention? The "Mod" sequence will have to move elsewhere, though.

[efe]

Just to inform you: I set a new KOTH score for "Super Fast Brainfuck" !

It was really fun to code this and there is still a lot of optimization possible.
I not even used threading!

[tails]

*phew* I have finally caught up to you, efe!
This time I didn't use threading, either.
I come up with a few more optimization, but not very much. I think this is nearly my best.

[Hippo]

i hear your concerns.

i didn't intend the fast brainfuck to block a city -- it won't be a choke point. none of the KOTH will be.

i definitely agree that the KOTH is pretty severe. especially with tails competing =) i needed something to stop the overnumerous challenge sequences questing to get ever smaller or faster. and to clear any mystery up: when your solution has been beaten, you don't get points for it any more.

there are two mitigating factors:
1) there's a window other solvers can be within to get credit
2) i'm planning (but haven't done yet) to add 'middle' challenges that require a certain set speed. i don't want to do this too early, because i want to see what the range of solutions evens out at. for example, for fast sort, i'll probably add a 'solve in 5k cycles' range. i don't want to make the KOTH windows too big, because that removes some excitement.

for the forum posting -- as solutions improve, the level of discussion will go up. if old solvers could still see it, they'd possibly get hints. a good example here is Super Small Hello World where the people who have gotten 29 have discussed how pretty openly.

I understand that I am a bit late with my proposal ...

As server remembers best so far solution for a solver there would be possibility to have solved threads for the challenge & score. Solver would not have access to threads with better score than he have achieved.

(Yes, it would make solving Super Small Hello World much more easier as the 29 solution is combination of 2 "main" tricks when each of them allows different 30 solution and both were mentioned in the thread before 29 solution was discovered ... and actually one trick was there from the 32 solution)

[AMindForeverVoyaging]

Well if adum gets around to do something, I rather hope he adds more challenges