more recent news | previous news

Top maps played. This list of top maps being played is from Nix's tracker. Note that people tend to mess up filenames a lot when resaving maps. If you added "Chew Toy 3" and "Chew Toy III" it would actually be the most popular map on the tournament games.

Updated: Thu Oct  9 16:53:37 EDT 1997

Tournament Games                Open Games
221   DH-Oil Rig �              366   Everard Island
178   Chew Toy 96                25   another harley's checkered map
157   Everard Island             25   Wash DC map
154   Xenophobia                 25   Flame War
150   Chew Toy 3                 24   Mutatis Mutandis
106   Etaka                      22   DH-Oil Rig �
 95   Baringi                    19   Blobbomint
 92   Skew Toy 2                 18   Tralfaz
 77   Chew Toy III               18   BESTEST MAP IN THE UNIVERSE
 71   Easter Island III          18   -Son of Pillbox of Death
 67   Hathec                     16   TixToxPillBox
 65   1x1 Lock II                16   Chew Toy 96
 59   Skathe                     15   Dream Soda V
 57   Dream Soda V               14   Gothic Industrial
 57   Chew Toy 2                 14   Arena
 56   Dream soda                 13   airstrip
 55   chew toy 3                 13   Skew Toy 2
 54   Filled Chew Toy            12   Arabu
 52   cubosh                     11   Guido's Enchanted Land
 49   Cubosh                     11   Easter Island III

Battle of the Bots. Preliminary tests on the new aIndy 3.0 released yesterday are disappointing. In 4 consecutive games, teams of version 2.6 bots beat teams of equal numbers of version 3.0 bots. We set up the games such that each computer ran 1 old and 1 new bot, with computers at the same site here at Duke and over Ethernet. This way there is no bias due to different computers or networking conditions. The bots never took up more than 20% of system resources according to Bolo's own estimates. We played 2x2 and 3x3 games, 3 games on DHOR and 1 on Crankcase.

Almost invariably, the 2.6 bots jump out to a better base grab. The 3.0 bots can do some interesting stuff, including base raping, decoying, even an occasional Jolo Maneuver (drawing enemy pillfire into an enemy tank), build blocks to take pills (but not often and not well), build blocks to refuel off spiked bases, etc. The 3.0 bots consistently won in jousts against 2.6 bots, with some really impressively ruthless shooting. One of my 3.0 bots had a 2:1 kill ratio in a 2x2 game, where by chance it should've been the other way around at 1:2.

In spite of all these new features, however, they kept losing. Pill placement seemed to be poor, reminiscent of Indy-style fortressing. There was far too much hardlining going on, the futile sort where they shoot a pill halfway down, take way too many hits, then wander far away to refuel to the top, and of course by then the enemy has fixed the pill. Likewise the base rape feature means they ignore pills almost entirely, and although 2.6 bots don't have a rape feature, they do hate to lose bases, so you can get a bit of a rape fest going on, and more often than not, the 3.0 bots end up losing both pills and bases!

We'll do more tests to see if we can come up with some consistent patterns. I think people should get the new bot and try it out for themselves. I saw some highly amusing and encouraging behavior in aIndy 3.0 - things I never saw in a bot before. Maybe we can take some notes and help Paul improve it.

If you're interested in learning more about Bolo brains, see the brain section from the rec.games.bolo FAQ or Paul Joswig's Bolo Brains Guide, both hosted at the OBHP.

In case you need to be convinced it's worth the effort to upgrade, here are all the changes from the author's original dox:

Changes from aIndy 2.6 to aIndy 3.0.

This version was many hours in the making.  Listed here are most,
probably all, of the major improvements, and some of the many minor
improvements.

Added a new Alert box and logic to deal with non advantage mode
(i.e. aIndy displays the alert, then shuts down).

Improved the "factory defaults" for the menus.

Modified pill to attack selection code so that early in the game we
try and target closer neutral pills more (i.e. weighting of other
factors was too high before).

Modified code to go get the LGM so that if we don't have enough
shells to reach him, go and refuel first.

Similar to above, if we don't have enough shells to pickup a dead
pill we have targetted, refuel first.

Added logic to avoid certain processing tasks if CPU is too low
(i.e. slow Mac, or too much going on to give aIndy enough time).

Changed the way tree farming is done to be more eco friendly; this
has positives and negatives I guess, I haven't watched it too much
to see if it's overall a regressive change.

Patched Route Finding a little so that if we are examining a square
with pill with no strength, the cost is minimal.  This way aIndy
"automatically" picks up other pills (sometimes?) when on the way to
get others.

aIndy communicates it's current pill target, and "cost" estimate of
getting there when picking up a dead pill and when fixing a pill.
This info is used by a team of aIndies to try and limit the number
of aIndies heading for any particular dead pill or pill in need of
repair. (Previously, all aIndies would flock to such "attractive"
targets).

Similarly, aIndy communicates it's current neutral base target
during the base grab, to try and prevent many aIndy's heading for
the same base.

Played around with Route Finding so that aIndy can shoot through
single barriers when on a boat, if the boat is needed after the
wall.

aIndy watches friendly tanks that may be causing a danger to our LGM
fixing some pill, and tries to avoid losing man this way.

aIndy watches allied men, and tries to avoid driving on squares that
would endanger them (only looking at the pill the allied man is
going to fix).  aIndy won't "flee" from these situations yet though.

More readily shoot at friendly pills to piss them when sitting on a
base.

Will now shoot friendly pills to piss them, in order to protect dead
pills (previously aIndy just tried to fetch the dead pill without
using it's pill resources for protection).

Modified KillTank so that the LGM isn't sent out when it's
dangerous.

Discovered that the way the selection of a tank to kill code was
written meant that we only ever select the first "known" tank in our
data, and this might not even be killable!  Replaced this with code
that considers all enemy tanks, orders them, then checks if we can
kill the most attractive enemy tank.

aIndy is now much safer at picking up trees in desperate situations
(e.g. need to plant pill and have no trees), since it looks around a
bit for a good tree instead of getting the nearest one (which may be
covered).

aIndy tries to avoid picking up dead pills that are covered by hot
hostile pills.

aIndy can now rebuild dead friendly pills, and tries to do it if it
is judged that building would be better than attempting to pickup.
Previous aIndies and Indy 2.02 were not programmed to build dead
pills (amazing huh); if you saw a man build a dead pill it's because
the man was sent before the pill as actually dead.

Refueling is changed such that: 1) aIndy doesn't try refueling at an
occupied base (ie one which has an ally sitting on it), except in
some special circumstances.  2) if a "manual" ally sits next to the
base an aIndy is on for 1 second, then aIndy will vacate that base
so that the ally can use it, except in special cases.  Is 1 second
too long or short?  3) if a non-manual tank, carrying pills, is
sitting next to our base then vacate after 1 second, unless we are
also carrying. This is to give preference to allies that are
carrying when good bases are scarce.  It works because if there was
a good base for the aIndy ally to refuel at, it wouldn't be trying
to get onto ours (see point 1).

Improved decision when to go for a pill, if sitting on a base while
carrying pills, aIndy will get more armor than it used to.

Implemented a Base Killer mode, in which the aIndy will target enemy
bases as first priority.  Base Killer mode is entered/exited
manually or automatically:

One aIndy on a team will automatically become a Base Killer iff the
number of players on the team is at least 2 more than the total
number of hostile players.  It exits Base Killer mode when there are
no enemy bases.  So, playing against 3 aIndies should be much more
challenging now.

Modified the way aIndy attacks bases which are covered by enemy
pills, to allow Base Killer aIndies to be more successful.

A human ally can also command one or more aIndy tanks to enter/exit
Base Killer mode. This is done using a modified form of the "kill
base" command, in which you replace the base ID with ON or OFF as
required.  Note also that the kill base command can now be given in
the short form using just k, e.g. !k 5, !k on, !k off (see the
section on command message changes for other changes).

Added commands to support pill massaging, i.e. to get aIndy to auto
re-ally you after leaving the alliance. (see section on commands).

Added menu item "Pause at Startup".  When checked, pause will be
switched on every time aIndy is started.  Handy for starting games
with/against bots, so you don't have to set "pause" on for each one
by hand.

Added memory of deepsea squares that have been seen, this enables
aIndy to tell if a dead pill is on deep sea or not (it is difficult
otherwise to tell believe it or not).  aIndy still doesn't use the
information as well as it should though.

Extensively reworked parts of the route finding, including coverage
system - aIndy now uses information regarding enemy pills "blocking"
other enemy pills.

Implemented new Path Checking code for the LGM, which takes account
of the fact that the LGM can move around objects; previously the
linear path was checked, and if any blocking terrain encountered the
path was considered impassable.

Using similar code to Path Checking, the Kill Man code works much
better - it can predict much better where an enemy LGM will be
(previously linear path was assumed).  Other good improvements were
made, and overall, LGM killing is much better.

Improved KillTank aiming.

When enemy men are dead aIndy more aggressively targets enemy pills
(i.e. in preference to neutrals).  I have now extended this so that
if *all* enemy men are dead, it's even more aggressive.

When allied men are dead, aIndy computes their return time, and
sends this to allies.  I think it also sends return time for dead
enemy men. the return time is only sent if you told aIndy you are a
"manual" player (by sending a "!mytype manual" message).

New FixPill.  It is now heaps better :-).  One aspect is that it
avoids using the pill it is trying to fix as a barrier to hide
behind if in range of en enemy pill (which was usually fatal to the
LGM ;).

Update the message system so that aIndy 27 can tell who is an allied
aIndy 27, this is the use of the /goodbye message when you shut down
aIndy.

Added code to detect switching pill targets based on distance.  If
we try switching back based only on distance, it isn't allowed.  It
works, only for switching between two pills though, but that's much
better than no detection.

Improved GetPill:

new code for determining when to turn and run from a pill we are
shooting at, resulting in faster pill takes.

improved the code that determines if we fix the terrain of the
square we will be attacking a pill from.

much better aiming for edges of pills

tank drives out from behind a barrier to get a better shoot angle if
possible; drives behind barrier if that's useful.

fine tuned attack positions data.

tries much harder to use n pill as a barrier in preference to a
wall.

And ...  drive not so close to target pill, shoot earlier, flee
earlier if the pill is strong, decelerate sooner when fleeing,
checks for danger and blocks when fleeing.

Implemented basics of pill massaging, but this isn't ready for prime
time yet, so it is unused code.  Key problem is WHEN to try a pill
massage.

Added reading of the Bolo prefs file, to get the keys that are set
up for Bolo, and use them.  Thanks are due to the person who
submitted this code by to brains list: Craig Hickey,
hickeyc@cadvision.com or cmhickey@acs.ucalgary.ca.

Added code which specifically deals with the case of numEnemies ==
0. In this case aIndy now looks much more intelligent when playing
against no one.

Implemented code that finds source and dest of each shot visible.
This is used to detect if an allied tank is firing at one of our
pills, and if so don't rush over to fix it.  Similarly, don't rush
to an fpill to fix it if it is being hit but there are o enemy tanks
or pills around.  This all prevents loss of LGM by attempting to fix
pill being picked up; also prevents frustration of human allies ;-).

Choosing which base to build pills at: tries to only build two pills
at each base instead of three, also uses KevDogs idea for building
closer to enemy territory.

Before shooting a pill to piss it off, checks if anything else is
already shooting that pill.

Modified shooting of neutral pills to kill enemy tanks - now only
does this if enemy tank is very close to neutral pill and we are far
away.


Fixed bug in recording of x,y locations of tanks as given by Bolo,
causing aIndy to aim poorly.

Fixed bug in parsing of messages.

Fixed two instances where the "ninth" surrounding square was being
examined.

Fixed bug where Pause/Cont messages were not adjusting state of
Pause menu item.

Fixed a "silly bug in refuel" (didn't record what it was).

Fixed bug in Refuel where the terrain to build a barrier on wasn't
exhaustively checked, so sometimes aIndy would build a boat on
river, instead of a road, then a wall.

Fixed bug in which pills could become momentarily forgotten.

Fixed handle leak in Prefs code form aIndy 2.6.

Fixed bug in GetTankFromXY where last tank wasn't being considered.

Fixed bug in the code for costing pills to attack where the number
of covering pills wasn't being considered.

Fixed a divide by zero error.


New/changed messaging:

Can now optionally use '.' character as prefix for commands.

Added "very short forms" for kill base, get pill, and suicidal get
pill commands.  Examples: !k 2, !g 1, !s 7

Modified Kill base command to switch base killer mode on and off. 
This is achieved by replacing the 'ID' with either 'on' or 'off'. 
e.g. !k on

Added Re Ally function.  This is the where you tell aIndy to request
alliance with you as soon as you leave the alliance.  This way you
can use aIndy as a pill massage partner.  Can be invoked as a one
off, or switched on and off.

As if that wasn't enough, CF's father did this gorgeous 3D rendered picture of a Bolo battlefield. It's a high-quality JPEG image, so it may look a little cruddy on your web browser, but if you save it to your disk and open it with JPEGview it'll look much better.

Speaking of the Rolodex... I forgot to mention that the Bolo Rolodex is constantly being updated. Within the past few weeks I've put in new addresses for MegaWatt, masher, Cat_butt, >, Kovas, and many others.

Usenet crosspost spam filter. Tired of seeing on rec.games.bolo dreck cross-posted to 25 newsgroups? If you're using a filter-capable news client such as Yet Another NewsWatcher, set up a filter where the "Newsgroups" field contains the following "regular expression":

.*,.*,.*,.*

By the way make sure you are using version 2.4.0 of YANW, the last version ever now that the author abandoned the project. To learn more about the various NewsWatcher clones, check out Claude Errera's NewsWatcher Tracking Page.

Bolo Rolodex hits 250. The Bolo Rolodex has over 250 players listed now. Note this is not a mailing list, just a way for you to look up individual players such as to arrange games, discuss strategy, set up a romantic rendezvous, etc.

And before you ask, no I don't have AA, nor do I know where/how/when you can get it, so don't even waste your breath. I do have some screen caps which I'll be putting on line soon. Stay tuned. :)

After messing around with AA, we settled down and played some good ol' fashioned Mac Bolo, or at least we tried. A recent upgrade of many of our Macs to OS8 somehow totally messed up their ability to Bolo, probably some bizzare extension conflict. After an hour of continuous splits, we moved mecca and pins to our second lab in the building where Max works, and finally got good 7-player games going with mecca, pins, Max, and Jolo at Duke, Noriko from Lawrence Berkeley National Lab (don't ask), Hoov from Michigan, and sluggo from UCLA. We played 3 good long games, and I actually went 2-1! Here's a tracker list:

Host: 128.97.2.230 {50000}  Players: 7  Bases: 16  Pills: 16
Map: Jannark  Game: Tourn.  Mines: Visible  Bots: No  PW: Yes
Version: 0.99.7  Tracked-For: 5 min.  Player-List:
   bolo://Warhawk.seas.ucla.edu/ ....... sluggo
   bolo://lizard.lbl.gov/ .............. Lawrence Berkeley National La
   bolo://emmac.biop.umich.edu/ ........ Hoov
   bolo://odin.mc.duke.edu/ ............ Max
   bolo://ripley.mc.duke.edu/ .......... Jolo rox the bolo ladder
   bolo://sabrina.mc.duke.edu/ ......... Duke University

We started at 8 pm or so and went until 11:30ish, then most of us left and Max stayed on for another 4 hours or so. All in all I enjoyed playing for the first time since March's bolofest. Who knows, you may even see more of me out there on the tracker. :)

OAK CANYON RULES THE OPEN GAME! From a tracker output taken today... looks like Arizona has a new rival.

Host: 168.174.120.222 {50000}  Players: 2  Bases: 6  Pills: 11
Map: Everard Island  Game: Open  Mines: Hidden  Bots: Yes  PW: No
Version: 0.99.7  Tracked-For: 9 min.  Player-List:
   bolo://168.174.120.222/ ...................... Tucson Unified School Distric
   bolo://com87.oakcanyon.alpine.k12.ut.us/ ..... Oak Canyon Junior High
   bolo://com103.oakcanyon.alpine.k12.ut.us/ .... Oak Canyon Junior High

Host: 205.118.17.85 {50000}  Players: 8  Bases: 0  Pills: 0
Map: Everard Island  Game: Open  Mines: Hidden  Bots: Yes  PW: No
Version: 0.99.7  Tracked-For: 40 min.  Player-List:
   bolo://com85.oakcanyon.alpine.k12.ut.us/ ..... Oak Canyon Junior High
   bolo://com86.oakcanyon.alpine.k12.ut.us/ ..... Oak Canyon Junior High
   bolo://com88.oakcanyon.alpine.k12.ut.us/ ..... Oak Canyon Junior High
   bolo://com91.oakcanyon.alpine.k12.ut.us/ ..... Oak Canyon Junior High
   bolo://com92.oakcanyon.alpine.k12.ut.us/ ..... Oak Canyon Junior High
   bolo://com93.oakcanyon.alpine.k12.ut.us/ ..... Oak Canyon Junior High
   bolo://com94.oakcanyon.alpine.k12.ut.us/ ..... Oak Canyon Junior High
   bolo://com98.oakcanyon.alpine.k12.ut.us/ ..... Oak Canyon Junior High
   bolo://com99.oakcanyon.alpine.k12.ut.us/ ..... Oak Canyon Junior High
   bolo://com102.oakcanyon.alpine.k12.ut.us/ .... Oak Canyon Junior High
   bolo://com104.oakcanyon.alpine.k12.ut.us/ .... Oak Canyon Junior High
   bolo://com107.oakcanyon.alpine.k12.ut.us/ .... Oak Canyon Junior High

Duke AA/Bolofest. No you didn't read that wrong. Chris Howard aka mecca of AA fame is visiting our area on business. We'll have a little Bolo gathering at 8 pm tomorrow (Tuesday the 30th) and maybe he'll even show us what AA really looks like.

Bolo Tournament! Damon Taaffe aka Kax <dwt@condor.sccs.swarthmore.edu> from Swarthmore is organizing a 1x1 and 2x2 tournament. Check out his announcement at his home page, which will serve as the tournament headquarters with all the latest info and outcomes, etc., or read rec.games.bolo. Email Kax to register or to get more info.

Jolo vs. Joel. Yes it's finally going to happen, we think. Chris Howard aka mecca is coming to town on business next week. He has promised to bring his PC laptop so I can finally play Armoured Alliance against He Who Cloned Bolo. People in the NC area will probably get together for some Bolo and/or beers, stay tuned here for Bolofest news.

Yet another possible tournament. This time Athena is thinking of organizing a 2x2 tournament. Email Athena at <iistormyii@aol.com> if you're interested. As usual, only serious players, no modemers, etc. need apply. Last I heard there were 6 people who were interested.

History of Bolo update. You can now get a sneak peek at the Chronology of Bolo page as it continues to grow. Don't link to it yet, this is just a work in progress.

This guy has not produced a copy of his program to date, and when pressed for details, made a series of obviously impossible claims regarding his "PCBolo". Some highlights that gave away the hoax - it was allegedly written in assembly, runs fast with 30 players at once, "uses windows95/NT for network play (TCP/IP) but will run in dos and possilby other os's", etc.

It still doesn't hurt to get us all thinking seriously about what it would take to get a real PC Bolo out there. Copyright issues have held up the PC Bolo / Armoured Alliance release for years, still with no end in sight. What if Joel or some other programmer decides to release a clone as freeware, say under the GNU licensing scheme? It is unclear what anybody can or would do to thwart that. There has been no word from Stu since he graduated a few months ago.

Well that's all I know, again don't ask me for more info or for PC Bolo itself, I can't help ya.

History of Bolo, part I. A few people including Athena, nature, and Samhain have heroically decided to compile a history of the game and its players. If you look around the OBHP, you'll see that while we have all the latest news, strategy, etc. there is really no good source of info on the "old days" - back when you could sit atop a pile of pills, impervious to attack; back before the base rape was invented; back when names like God, Santa, Red, and yes even Black Lightning were feared; back when Jolo used to play Bolo. :-)

The first task will be to make some sort of outline and organize the task a bit. If you have ideas, old logs from #bolo, rec.games.bolo or elsewhere, or other historical contributions, please email Athena at <iistormyii@aol.com> or nature at <doug@holdit.com>. I've already donated to them 150 or so logs I've collected over the years, including the infamous "Moments in #bolo" and other rgb musings. Let's all pitch in and try to compile a really kickass history of Bolo!

Bolo Buddy Page Kevin Whitty aka KevDog, the author of Bolo Buddy (the only sane way to find games if you don't have easy telnet access), has created a Bolo Buddy home page courtesy of the fine folks at Kagi. The latest beta versions will be available there if you want to help him debug stuff. As of version 1.1 Bolo Buddy will become $10 shareware but it won't be crippled in any way.

Jolo defecting to .com? I try not to spam my personal life here, but this may potentially affect the future of the OBHP so here goes. In spite of being underpaid, I was reasonably happy in my current job as an assistant research professor in the dept of radiology at Duke Univ Med Ctr, having recently received a decent raise and promotion, which coincided with buying a new house too. Then a few weeks ago a headhunter (executive recruiter) calls me out of the blue regarding a position as medical imaging scientist at a new company called Sterling Diagnostic Imaging. Sterling made its fame and fortune through leveraged buyouts of over 30 failing companies, turning each around and selling it at a huge profit before finally taking over Dupont's diagnostic imaging division. This company now has a half-billion dollar per year revenue and makes 30% of the x-ray film used in the US. I'd be helping them finalize development of a revolutionary new digital x-ray device, slated for production Q1 98. It's not really my field of specialization, but it's general medical imaging stuff that I can definitely help them with. They are desperate to fill the position immediately (like in a few weeks), and it would involve relocating to the Wilmington, Delaware area.

Normally I would just express polite interest and blow them off, but here's the catch - they're offering to nearly double my salary! Money is not a huge factor in my life, but still, I know I will probably never make that kind of money anytime in the near future staying in academia and certainly not at Duke. While the job isn't incredibly inspirational, it certainly is interesting, challenging, and could directly contribute to patient care as early as next year.

So after a few phone interviews, this past Monday (Sep 8) I flew up to Sterling for a 1-day site visit. The people seemed very nice and knowledgeable, but unfortunately I didn't see much of the product, so I have some nagging doubts about some known flaws which may affect its commercial feasibility. Ironically just 2 weeks ago, Duke announced a partnership with GE Medical Systems to test their secretly developed rival product, which allegedly fixes both of the major problems plaguing the Sterling version, plus there's GE's gargantuan marketing machine to contend with. So now there is certainly increased risk with taking the Sterling job, but then again I'd be getting in early and may well get rich off stock options if Sterling makes it. Since Sterling has only been around 1.5 years and has only a few products in its arsenal, it's impossible to predict the long-term future of this job.

I'm now waiting on the formal offer from Sterling, which should be any day now, and obviously I'd have to talk this over with my current boss. I'll keep you guys posted. If I do take the job, we may or may not have to find a new home for the OBHP and FTP archive. Let's not worry too much about it yet.

PC Bolo is out? We've all heard the "I HAVE PC BOLO AND WILL RELEASE!!!" jokes, but this time it's not a joke!? According to reliable sources, Marc Lubarsky aka ^Zer0^, a relative newcomer to the field, has allegedly been working in secret on an exact PC clone of Bolo. This is not affiliated with or endorsed by the PC Bolo / Armoured Alliance people. Apparently private beta copies got leaked and are floating around, so the author allegedly is going to release it today! Details are sketchy at this point. Stay tuned here for breaking news.

Bolo Buddy beta A major new version of Bolo Buddy has been in beta testing the last few weeks. If you're interested in helping with the testing, contact the author Kevin at <kev@ray.abo.fi> or me.

more recent news | previous news

[ Intro | News | Links | Archive | Guides | Gallery | People | Misc ]

The Bolo Home Page is copyrighted by
Joseph Lo & Chris Hwang