June tournament post-mortem

58 views
Skip to first unread message

Shawn Moore

unread,
Jul 1, 2006, 1:08:54 PM7/1/06
to
First of all I'd like to thank paxed and dtype for putting up with the
June tournament. To anselmus for giving me the idea (Monks do not suck,
thank you very much!). To rast for pushing for the lowest turncount
trophy (which in turn allowed best behaved and most gold). And, of
course, a big thanks to all the players.

We smashed just about every record:
* daily ascension record of 14 on June 2nd (previous was 9 on April 7th
and March 19th)
* monthly ascension record of 188 (previous was 111 in May)
* daily game record of 1481 on June 20th (previous was 1089 on
September 1st, 2005)
* monthly game record of 18748 (previous was 14241 in December, 2005)

Before I continue I'd just like to say that yes, this *will* be an
annual thing. There are good arguments on both sides for hosting it in
May (closer to halfway to the devnull tournament) and July (fewer
people in school) but (for now) next year will be in June.

Of course, as you may have noticed, tournament wasn't without its
problems. There were a lot of problems with unique deaths. During the
final few days of the tournament I had to exclude every death that
included " called " and " named ". People were abusing the scripts by
dying to small critters named "(with the Amulet)". I couldn't rework
the regex in time, so this was the next best solution. For next year,
I'll be writing the regex from scratch so such abuse will be minimized.
I'll also comb the source for any deaths that the devnull people might
have missed (and there are some, such as "killed by sitting in lava"
which reid achieved).

There was also a problem with players whose NAO names have more than
ten letters. For some reason, the dgamelaunch software allows players
to have up to twenty characters in their names, but NetHack can only
support the first ten. So in the logfile, squidlarkin's games were
recorded under "squidlarki". This is a huge problem in (at least!) two
ways: first, there's no easy way of automatically expanding
"squidlarki" to "squidlarkin" -- the lastgame dumplog (which I needed
to grab and parse for ascensions) uses the full "squidlarkin". My
solution during the tournament was to hardcode in certain expansions
(so squidlarki becomes squidlarkin, GmonSquare becomes GmonSquared,
etc). I'm not going to get into the other problem (paxed and dtype know
of it, I've submitted a patch, etc.) but thankfully it wasn't an issue
during the tournament. Hopefully by next year NAO is patched to include
the full name in the logfile or some other solution is made clear.

For some reason a few ascensions have been missed by the scripts. For
example, brooder is the real winner of highest-scoring ranger. I'll fix
this problem as soon as I'm done posting here.

Some stuff for next year:
* Single ascension trophies will be awarded to each role (so you'll see
"Best Behaved Valkyrie" and "Fastest ")
* Definitely fastest realtime ascension (even if that means I have to
calculate this one manually!)
* If NAO is patched in the meantime, a whole lot more! (fewest
keystroke ascension, anyone?)

Shawn M Moore

Shawn Moore

unread,
Jul 1, 2006, 1:15:22 PM7/1/06
to
Oops, missed two things!

Shawn Moore wrote:
> First of all I'd like to thank paxed and dtype for putting up with the
> June tournament. To anselmus for giving me the idea (Monks do not suck,
> thank you very much!). To rast for pushing for the lowest turncount
> trophy (which in turn allowed best behaved and most gold). And, of
> course, a big thanks to all the players.

Also to Calamarain for making the trophy images. (sorry I forgot you,
man)

> * Single ascension trophies will be awarded to each role (so you'll see
> "Best Behaved Valkyrie" and "Fastest ")

Read that as "Fastest Samurai"

Shawn M Moore

Rast

unread,
Jul 1, 2006, 3:40:42 PM7/1/06
to
In article <1151773734.2...@75g2000cwc.googlegroups.com>,
sar...@gmail.com says...

> First of all I'd like to thank paxed and dtype for putting up with the
> June tournament.

Here is the list of winners:
http://sartak.katron.org/nh/tourney/trophies.html


And here is the full sorted list of ascension turncounts:
http://sartak.katron.org/nh/tourney_sturns.txt

Median: 40.0k turns
Top 16%: 23.7k turns
Bottom 16%: 64.3k turns

Of course, these are not representative of all ascensions, or even all
NAO ascensions...


--
"Sometimes I stand by the door and look into the darkness. Then I
am reminded how dearly I cherish my boredom, and what a precious
commodity is so much misery." -- Jack Vance

Zarel

unread,
Jul 1, 2006, 6:19:41 PM7/1/06
to
Rast wrote:
> In article <1151773734.2...@75g2000cwc.googlegroups.com>,
> sar...@gmail.com says...
>> First of all I'd like to thank paxed and dtype for putting up with the
>> June tournament.
>
> Here is the list of winners:
> http://sartak.katron.org/nh/tourney/trophies.html
>
>
> And here is the full sorted list of ascension turncounts:
> http://sartak.katron.org/nh/tourney_sturns.txt
>
> Median: 40.0k turns
> Top 16%: 23.7k turns
> Bottom 16%: 64.3k turns
>
> Of course, these are not representative of all ascensions, or even all
> NAO ascensions...
>
>
Question: What do the achievement trophies mean? I'm confused about them.

--
____ (__)
/ \ (oo) -Zarel
|Moo. > \/
\____/

Shawn Moore

unread,
Jul 1, 2006, 7:03:18 PM7/1/06
to
Zarel wrote:
> Question: What do the achievement trophies mean? I'm confused about them.

Single-game achievements:
* You get the Gold Star for making it to the Plane of Earth.
* You get the Platinum Star for making it to the Astral Plane.
* You get the Dilithium Star for ascending.

Multiple-game achievements:
* You get the Birdie by ascending both genders.
* You get the Double Top by ascending both genders and all three
alignments.
* You get the Hat Trick by ascending both genders, all three
alignments, and all five races.
* You get the Grand Slam by ascending both genders, all three
alignments, all five races, and all thirteen roles.
* You get the Full Monty by ascending both genders, all three
alignments, all five races, all thirteen roles, and all twelve
conducts.

For multiple-game achievements, you get "with bells on" if you also
have no unnecessary games in your run for the trophy. All deaths count
as unnecessary. Say you're shooting for the Hat Trick and ascend (in a
row): Human, Elf, Orc, Orc, Gnome, Dwarf. That wouldn't count because
you had the unnecessary doubling of the Orc. However, you can keep
going with the last three (i.e. ascend another human and elf) to get
the hat trick with bells on. That run would of course count for hat
trick without bells (provided you got both genders and the three
alignments)

You can see it phrased differently at the devnull page, here (near the
bottom):
http://groups.google.com/group/rec.games.roguelike.nethack/post?inreplyto=40a3b201cce003fb&reply_to=group&_done=%2Fgroup%2Frec.games.roguelike.nethack%2Fbrowse_thread%2Fthread%2Ffe2c2ecf094769f3%2F7407c0c0328378fd%3F&

Also the tournament information page (which answers a lot of questions
you folks might have) is at:
http://sartak.katron.org/nh/tourney/

Shawn M Moore

Jack Bolsen

unread,
Jul 2, 2006, 7:51:17 PM7/2/06
to

>>
>> And here is the full sorted list of ascension turncounts:
>> http://sartak.katron.org/nh/tourney_sturns.txt
>>
>> Median: 40.0k turns
>> Top 16%: 23.7k turns
>> Bottom 16%: 64.3k turns
>>

Curious, I would have thought that better players would spend more time
on their ascensions than weaker players. The first ascensions I ever did
were fairly quick - between 25,000 and 30,000 turns. However, my
ascensions lately have taken anywhere between 40,000 - 50,000 turns or
more (I spent 55k turns on my healer ascension for the tourney and about
56k for my barbarian).

I wonder if this means I'm actually getting worse?

Jack

Philipp Lucas

unread,
Jul 3, 2006, 5:11:28 AM7/3/06
to
On Sun, 2 Jul 2006, Jack Bolsen wrote:

> Curious, I would have thought that better players would spend more time
> on their ascensions than weaker players. The first ascensions I ever did
> were fairly quick - between 25,000 and 30,000 turns. However, my
> ascensions lately have taken anywhere between 40,000 - 50,000 turns or
> more (I spent 55k turns on my healer ascension for the tourney and about
> 56k for my barbarian).

The first times I came to the planes, I was underprepared and died. So I
decided to grab more and more wands, increase my HP even higher and so on,
which takes time, and my first ascension had about 65,000 turns. With
time, I got better, and realised that I do not need to spend so much time
preparing; also, the habit of staying too long in the midgame levels to
create better armour, wait for a particular artifact or maybe a wish
vanished, so the next ascensions were around 35,000 turns.

If this is representative, it makes sense that better players have
ascensions in the 20,000--30,000 area, whereas worse players need more
turns.

25,000 turns in a first ascension is quite impressive.

--
Philipp Lucas
phl...@online-club.de

chee...@hotmail.com

unread,
Jul 3, 2006, 3:01:27 PM7/3/06
to
Shawn Moore wrote:
> First of all I'd like to thank paxed and dtype for putting up with the
> June tournament. To anselmus for giving me the idea (Monks do not suck,
> thank you very much!). To rast for pushing for the lowest turncount
> trophy (which in turn allowed best behaved and most gold). And, of
> course, a big thanks to all the players.

Wish I'd had time to compete this time.

> Of course, as you may have noticed, tournament wasn't without its
> problems. There were a lot of problems with unique deaths. During the
> final few days of the tournament I had to exclude every death that
> included " called " and " named ". People were abusing the scripts by
> dying to small critters named "(with the Amulet)". I couldn't rework
> the regex in time, so this was the next best solution.

Another solution would be to call this what it is--cheating--and hand
out disqualifications.

*pounds fist on table*

--
kb

Stabwound

unread,
Jul 3, 2006, 3:48:21 PM7/3/06
to

"Jack Bolsen" <jbo...@verizon.net> wrote in message
news:VZYpg.885$5i3.709@trnddc01...

Most people would consider a fast player to be superior to a slower player.
To get through the game faster, you have to sacrifice certain things, to
make the best use out of what you have, and to know the minimum
stats/equpiment you can get along with. Throwing out a 40-50k turn game
usually means you are inexperienced or are overpreparing a LOT.

Your case is actually pretty odd. 25k-30k turns is roughly the average for a
good player, and in my opinion _excellent_ for an early ascension. There is
no need to spend more than 35k turns in a normal ascension. I'd have to see
your logs, but I'd guess that you spent a lot of time overpreparing in
getting your AC low and HP high; not really needed when you are comfortable
with the game.


Kent Paul Dolan

unread,
Jul 3, 2006, 8:11:56 PM7/3/06
to
"Shawn Moore" <sar...@gmail.com> wrote:

> Also to Calamarain for making the trophy images.

Which, from a graphics geek, are plain awesome,
especially considering the limited time frame:

http://sartak.katron.org/nh/tourney/trophies.html

Well Done!

xanthian.


--
Posted via Mailgate.ORG Server - http://www.Mailgate.ORG

sjde...@yahoo.com

unread,
Jul 3, 2006, 11:58:35 PM7/3/06
to

IMO, the two signs of a great player are either many consecutive
anscensions (mrivan always, or anselmus here) or fast anscensions (Rast
is the king here).

It's _much_ harder to ascend in < 15,000 turns (or <7,000, as Rast has
done a few times) than to ascend in 20,000+ turns.

But ascending 13+ in a row is also incredibly hard.

Calamarain

unread,
Jul 4, 2006, 1:00:34 AM7/4/06
to

Kent Paul Dolan wrote:
> "Shawn Moore" <sar...@gmail.com> wrote:
>
> > Also to Calamarain for making the trophy images.
>
> Which, from a graphics geek, are plain awesome,
> especially considering the limited time frame:
>
> http://sartak.katron.org/nh/tourney/trophies.html
>
> Well Done!
>
> xanthian.
>

Thanks :) I enjoyed making them. All done in Bryce and GIMP :)

Cal

Rast

unread,
Jul 4, 2006, 1:56:33 AM7/4/06
to
In article <VZYpg.885$5i3.709@trnddc01>, jbo...@verizon.net says...

> >> And here is the full sorted list of ascension turncounts:
> >> http://sartak.katron.org/nh/tourney_sturns.txt
> >>
> >> Median: 40.0k turns
> >> Top 16%: 23.7k turns
> >> Bottom 16%: 64.3k turns
> >>
>
> Curious, I would have thought that better players would spend more time
> on their ascensions than weaker players.

There are only a certain amount of things ones needs to do to ascend.
The weaker players tend to waste many turns doing things which don't
improve the chances for ascension.

I _suspect_ that the median turn count for endgame ("with the Amulet")
deaths is at least 60k, but I don't have a way to find out.

Rast

unread,
Jul 4, 2006, 3:02:29 AM7/4/06
to
In article <MPG.1f13cd1ec...@198.186.190.61>, ra...@hotmail.com
says...


> I _suspect_ that the median turn count for endgame ("with the Amulet")
> deaths is at least 60k, but I don't have a way to find out.

Following myself up here...


http://alt.org/nethack/logfind.php reveals 16 .*(with the Amulet) deaths
since they started saving dumplogs (June 1). Two of those are fakes
involving "clever" food or monster names, and one somehow didn't save a
dumplog. The remaining 13, in chronological order:

http://alt.org/nethack/chardump/mr0t/mr0t.20060603-055001.txt
http://alt.org/nethack/chardump/tournatek/tournatek.20060604-134501.txt
http://alt.org/nethack/chardump/pbrane/pbrane.20060604-150502.txt
http://alt.org/nethack/chardump/Allie/Allie.20060608-005501.txt
http://alt.org/nethack/chardump/Sijorian/Sijorian.20060608-011501.txt
http://alt.org/nethack/chardump/spaceMAN/spaceMAN.20060608-060502.txt
http://alt.org/nethack/chardump/zem/zem.20060613-092501.txt
http://alt.org/nethack/chardump/mr0t/mr0t.20060615-021001.txt
http://alt.org/nethack/chardump/soze/soze.20060619-001502.txt
http://alt.org/nethack/chardump/danguy/danguy.20060622-024001.txt
http://alt.org/nethack/chardump/tournatek/tournatek.20060623-071501.txt
http://alt.org/nethack/chardump/smitty/smitty.20060630-210001.txt
http://alt.org/nethack/chardump/Angcrist/Angcrist.20060702-174501.txt

Turns, sorted:

33486
35688
43347
47150
47777
48604
55498 <--- median, 15k turns slower than the median ascension
56234
56475
56905
60249
65674
68501

One can glean all sorts of interesting data from those 13 YASDs. For
example, only two were wearing rings of conflict, although many had them
in inventory or bag. Eight had been lifesaved previously that game; this
may or may not show the true value of "oLS. Four had AC:-32 or better,
one had AC:-25, and the remaining eight had AC:-22 or _worse_.

Rachel Elizabeth Dillon

unread,
Jul 4, 2006, 8:53:06 AM7/4/06
to
On 2006-07-04, Rast <ra...@hotmail.com> wrote:
> In article <MPG.1f13cd1ec...@198.186.190.61>, ra...@hotmail.com
> says...
>
>> I _suspect_ that the median turn count for endgame ("with the Amulet")
>> deaths is at least 60k, but I don't have a way to find out.
>
> Following myself up here...
> Turns, sorted:
>
> 33486
> 35688
> 43347
> 47150
> 47777
> 48604
> 55498 <--- median, 15k turns slower than the median ascension
> 56234
> 56475
> 56905
> 60249
> 65674
> 68501
>
> One can glean all sorts of interesting data from those 13 YASDs. For
> example, only two were wearing rings of conflict, although many had them
> in inventory or bag. Eight had been lifesaved previously that game; this
> may or may not show the true value of "oLS. Four had AC:-32 or better,
> one had AC:-25, and the remaining eight had AC:-22 or _worse_.

Wow. I tend to overprepare because I enjoy building up characters to how
I want them (though I should really work on ascending faster and becoming
less dependent on exactly what I want --- my Healer ascension, in the
tournament, had not yet seen Sokoban at 22000 turns) and because I am
paranoid. But I think AC -22 is not overpreparing. As an overpreparer, I
would consider AC -22 _underpreparing_.

Overpreparing is AC -50, wished-up Eye and Orb of Fate, max HP around
500, XL 30, all stats at max, and 0% failure on level four spells as
a combat wombat. I would classify AC -22 or worse at >50000 turns,
unless going for some kind of wonky conduct, as preparing poorly. My
recommended overpreparing is not necessary either (though it gave me
the resources I needed to tame Famine and Pestilence, which I thought
was a lot of fun and thus worth the extra time), but I think it is less
likely to lead to death.

Based on my own experience, I expect that players will, as part of the
learning process, start ascending faster, or at least taking more risks,
once ascending itself becomes not that big of a deal. Before that, I
expect that they will make the mistake of preparing poorly, and still
take a long time to do it. My only endgame death (so far!) was poorly
prepared in a large amount of time, much like these games. From that
death I learned what things I want in my ideal character, and rather
than figuring out exactly what I need to squeak by, I've just been
constructing my ideal character and sending it up to Astral every time.
This is a fine plan for racking up an ascension every month or so but
is not going to win me any major tournament trophies.

(There's still no excuse for not seeing Sokoban at turn 22000 though. :)

-r.

Calamarain

unread,
Jul 7, 2006, 2:21:47 PM7/7/06
to

>
> Wow. I tend to overprepare because I enjoy building up characters to how
> I want them (though I should really work on ascending faster and becoming
> less dependent on exactly what I want --- my Healer ascension, in the
> tournament, had not yet seen Sokoban at 22000 turns) and because I am
> paranoid. But I think AC -22 is not overpreparing. As an overpreparer, I
> would consider AC -22 _underpreparing_.

It's amazing how you get less and less paranoid the more you know about
the endgame and the more you've done it. My first ascension - I was
utterly paranoid, asking for help on NAO a lot, using ring of conflict
freely, and being very scared of the endgame riders to the point of
panic.

Now, I respect the harder enemies at the end and am cautious, but not
so afraid. But I would definetly say not seeing Sokoban by 20k turns is
overpreparing, unless you've spent an inordinate amount of time in the
mines doing other stuff.

> Overpreparing is AC -50, wished-up Eye and Orb of Fate, max HP around
> 500, XL 30, all stats at max, and 0% failure on level four spells as
> a combat wombat. I would classify AC -22 or worse at >50000 turns,
> unless going for some kind of wonky conduct, as preparing poorly. My
> recommended overpreparing is not necessary either (though it gave me
> the resources I needed to tame Famine and Pestilence, which I thought
> was a lot of fun and thus worth the extra time), but I think it is less
> likely to lead to death.

Yeah, I'd definetly count that as overpreparation :) 250HP is generally
more than sufficient. Means you can at least waltz through the endgame,
assuming no careless mistakes.

> Based on my own experience, I expect that players will, as part of the
> learning process, start ascending faster, or at least taking more risks,
> once ascending itself becomes not that big of a deal. Before that, I
> expect that they will make the mistake of preparing poorly, and still
> take a long time to do it. My only endgame death (so far!) was poorly
> prepared in a large amount of time, much like these games. From that
> death I learned what things I want in my ideal character, and rather
> than figuring out exactly what I need to squeak by, I've just been
> constructing my ideal character and sending it up to Astral every time.
> This is a fine plan for racking up an ascension every month or so but
> is not going to win me any major tournament trophies.

Try setting yourself goals - e.g. Sokoban by 10k turns at the latest.
Castle by 25k. Wizard by 40k. Ascend by 50k. Those aren't too
unreasonable, but they would prevent "overpreparation".

> (There's still no excuse for not seeing Sokoban at turn 22000 though. :)
>

There's a perfectly good excuse for it. Finding an amulet of reflection
and a bag of holding elsewhere ;)

Boudewijn Waijers

unread,
Aug 30, 2006, 9:15:54 AM8/30/06
to
sjde...@yahoo.com wrote:

> It's _much_ harder to ascend in < 15,000 turns (or <7,000, as Rast has
> done a few times) than to ascend in 20,000+ turns.
>
> But ascending 13+ in a row is also incredibly hard.

My ascensions *always* take 13+ turns...

--
Boudewijn.

Confusius says: may the RNG be good to you!


Reply all
Reply to author
Forward
0 new messages