27 ms·

There are some plots [1] of the time taken for the first 100 solutions to part 1 (silver dot) and part 2 (gold dot). They're a good visual indication of the rel

There are some plots [1] of the time taken for the first 100 solutions to part 1 (silver dot) and part 2 (gold dot). They're a good visual indication of the relative difficulty of each question and how the difficulty of the second part increases. [1] http://www.maurits.vdschee.nl/scatterplot/ http://www.maurits.vdschee.nl/scatterplot/

Strilanc 5y ago
I like that the data clearly shows the server issues on day 1.
djaychela 5y ago
Those plots are really comforting for me, as on the days where a part was difficult (such as day 20 part 2 last year) I didn't find a solution. Thanks, I'll be keeping an eye on them this year as things progress (or not!)
wpietri 5y ago
Ooh, that's really interesting. And given how many folks here talk about dropping out as it gets harder, that suggests the later problems look easier than they are.
- Jtsummers 5y ago
  The weekday problems tend to be easier so you get a ramp up, but with drops in difficulty. A lot of people give up after one or two of the hard days and don't continue, but there's no actual requirement to do them in order. If they'd skip the hard ones they'd probably find several more solvable (for them) problems remaining in the year.
- pedrosorio 5y ago
  These plots only include the first 100 people. Those dropping out as it gets harder were unlikely to be be represented in the top 100 for later problems either way.
artursapek 5y ago
...is the y axis showing DD:HH:MM or HH:MM:SS?
- Jtsummers 5y ago
  HH:MM:SS, the first 100 solvers are usually done within minutes, hours at most, and never days. None of the problems are that hard.
  - artursapek 5y ago
    wow 0:
Volundr 5y ago
Is this really saying for 2021, the first solution to the first problem came in 30 seconds after being posted? That's insane!
- Rietty 5y ago
  People are incredibly quick at both skimming/parsing the problem, coupled with a language like python where you can generally express stuff incredibly concisely and not have to worry too much about types, it makes it very quick.
- notafraudster 5y ago
  I'm in a really bad time zone for AoC right now, but here was my solution in R for day 1 problem 1. library(tidyverse) # pre-typed lines = readLines("data/day1.txt") %>% as.numeric() # Read data diff(lines) %>% .[. > 0] %>% length() # Part 1 This didn't involve me being a sophisticated or good programmer, and if I was in a time zone to be competitive I also would have written a function to submit my answer from within my IDE (which can also be pre-written). I could absolutely buy people wanting to be competitive and being able to do this in 30 seconds. I actually could have pre-written the second line of three, in point of fact.
  - jazzyjackson 5y ago
    whoa I didn’t even consider automating my answer submission, suppose it’s just a curl away
speedgoose 5y ago
And that's for the top 100. I would assume that the differences would be even larger for the average developers.
oceliker 5y ago
That's crazy. I have no idea how anybody can solve this problem in 258 seconds: https://adventofcode.com/2020/day/20 https://adventofcode.com/2020/day/20 Did they hope the corner tiles had unique edges and look for those? (which would be unlikely under normal circumstances, considering there are more than a hundred tiles and only 256 different possible edges...)
- jcranmer 5y ago
  From the problem statement: > Tiles at the edge of the image also have this border, but the outermost edges won't line up with any other tiles. So they didn't need to hope; the problem guaranteed it for them.
  - oceliker 5y ago
    Ah, sorry, I missed that. That does make it easy.
- yesbabyyes 5y ago
  While I agree four minutes is very quick, if you take into account that no side fits with more than one other side, and that the sides (#, .) may just as well be interpreted as binary numbers, it becomes an issue of parsing, mostly, and then matching numbers. I remember being flabbergasted at first, and then realizing the above. My solution is here: https://github.com/linus/aoc/blob/main/2020/js/day-20/app.js https://github.com/linus/aoc/blob/main/2020/js/day-20/app.js
  - oceliker 5y ago
    So it turns out I missed those parts of the problem statement. That makes it fairly easy. I probably still can't write working code for that in four minutes, but it does sound humanly possible, at least. Also, that's really nicely written code :)
- CameronAavik 5y ago
  Funny you mention that, the person who solved part 1 in 258 seconds recorded themself solving that one live: https://www.youtube.com/watch?v=ODUkgEeF7sc https://www.youtube.com/watch?v=ODUkgEeF7sc

Strilanc 5y ago

I like that the data clearly shows the server issues on day 1.

djaychela 5y ago

Those plots are really comforting for me, as on the days where a part was difficult (such as day 20 part 2 last year) I didn't find a solution. Thanks, I'll be keeping an eye on them this year as things progress (or not!)

wpietri 5y ago

Ooh, that's really interesting. And given how many folks here talk about dropping out as it gets harder, that suggests the later problems look easier than they are.

Jtsummers 5y ago
The weekday problems tend to be easier so you get a ramp up, but with drops in difficulty. A lot of people give up after one or two of the hard days and don't continue, but there's no actual requirement to do them in order. If they'd skip the hard ones they'd probably find several more solvable (for them) problems remaining in the year.
pedrosorio 5y ago
These plots only include the first 100 people. Those dropping out as it gets harder were unlikely to be be represented in the top 100 for later problems either way.

artursapek 5y ago

...is the y axis showing DD:HH:MM or HH:MM:SS?

Jtsummers 5y ago
HH:MM:SS, the first 100 solvers are usually done within minutes, hours at most, and never days. None of the problems are that hard.
- artursapek 5y ago
  wow 0:

Volundr 5y ago

Is this really saying for 2021, the first solution to the first problem came in 30 seconds after being posted? That's insane!

Rietty 5y ago
People are incredibly quick at both skimming/parsing the problem, coupled with a language like python where you can generally express stuff incredibly concisely and not have to worry too much about types, it makes it very quick.
notafraudster 5y ago
I'm in a really bad time zone for AoC right now, but here was my solution in R for day 1 problem 1. library(tidyverse) # pre-typed lines = readLines("data/day1.txt") %>% as.numeric() # Read data diff(lines) %>% .[. > 0] %>% length() # Part 1 This didn't involve me being a sophisticated or good programmer, and if I was in a time zone to be competitive I also would have written a function to submit my answer from within my IDE (which can also be pre-written). I could absolutely buy people wanting to be competitive and being able to do this in 30 seconds. I actually could have pre-written the second line of three, in point of fact.
- jazzyjackson 5y ago
  whoa I didn’t even consider automating my answer submission, suppose it’s just a curl away

speedgoose 5y ago

And that's for the top 100. I would assume that the differences would be even larger for the average developers.

oceliker 5y ago

That's crazy. I have no idea how anybody can solve this problem in 258 seconds: https://adventofcode.com/2020/day/20 https://adventofcode.com/2020/day/20 Did they hope the corner tiles had unique edges and look for those? (which would be unlikely under normal circumstances, considering there are more than a hundred tiles and only 256 different possible edges...)

jcranmer 5y ago
From the problem statement: > Tiles at the edge of the image also have this border, but the outermost edges won't line up with any other tiles. So they didn't need to hope; the problem guaranteed it for them.
- oceliker 5y ago
  Ah, sorry, I missed that. That does make it easy.
yesbabyyes 5y ago
While I agree four minutes is very quick, if you take into account that no side fits with more than one other side, and that the sides (#, .) may just as well be interpreted as binary numbers, it becomes an issue of parsing, mostly, and then matching numbers. I remember being flabbergasted at first, and then realizing the above. My solution is here: https://github.com/linus/aoc/blob/main/2020/js/day-20/app.js https://github.com/linus/aoc/blob/main/2020/js/day-20/app.js
- oceliker 5y ago
  So it turns out I missed those parts of the problem statement. That makes it fairly easy. I probably still can't write working code for that in four minutes, but it does sound humanly possible, at least. Also, that's really nicely written code :)
CameronAavik 5y ago
Funny you mention that, the person who solved part 1 in 258 seconds recorded themself solving that one live: https://www.youtube.com/watch?v=ODUkgEeF7sc https://www.youtube.com/watch?v=ODUkgEeF7sc