|
Comments Round 1
The top 6 warriors from Round 1 all use a precision scanning
technique, and I thought that corewarriors who are not familiar with
it might like a short introduction.
A precision scan finds the long QS code, then scans for the beginning
(or end) of it so that the black warrior knows *exactly* where to find
all the components of the white. This makes it pretty easy to boot a
cleanup warrior somewhere safe and pick them off at leisure; you can
even kill a replicator with DAT bombs when you know exactly where to
hit.
I first met it a couple of tournaments ago when Paul Kline used it to
win a white warrior round by a landslide. Most long white warriors are
best dealt with this way. I have often used in in White warrior
rounds, and my multiwarrior :-= used a similar "precision fang"
approach to get on to the multiwarrior hill by exploiting a single
known opponent.
All the warriors are on available at koth.org, so I will just give
some comments.
Force feedback uses a 0.5c scanner. I am a little surprised that it
beat Darkside because the longer the scan takes, the more time the
replicator has to overwrite the black warrior. OTOH it is shorter,
which makes it less vulnerable to both the QS and the replicators, and
it looks as if it comes out a little ahead on balance. If you have not
seen a decoy generator before, watch it in cdb. The QS spacing
makes -300 a very effective place to put a decoy because the QS will
usually scan it 1 cycle before it would have found your real warrior.
Darkside uses a 0.8c scan to find the QS faster, but it is longer and
so more vulnerable. It is also a bit difficult to reset the pointers
if I happen to scan a replicator before I get the QS and so the scan
pattern is not always perfect. I discovered that short DAT carpets
would kill the replicators very quickly, so I did not bother to boot
the bomber but maybe this was a mistake.
JedimPURGE is a 0.8c scan coupled with a short 2c QS -- more for fun
than effect, I suspect! It SPL carpets the replicators before starting a
spiral clear. This requires greater length, but the spiral clear can
win even if the scan has missed the QS and found a replicator.
myBlack is different, because the QS is handled by precision spacing
of the components; the warrior will often be hit but because the bombs
are known to be exactly 12 apart, the components are aranged so that
one scanner and payload will (always?) remain intact. This is another
neat example of the kind of power that perfect knowledge of your
opponent gives you. The warrior is otherwise very similar with a 0.5c
scan booting a spiral clear.
Tsetse takes the usual approach, but performs rather worse than the
first three. I think this is because it uses a short but slow decoy
generator which gives the QS too many chances to hit before the decoy
is in place. This is an odd decision because AFAIK Paul invented the
3c decoy generator. It is safe to use a long decoy generator because
even if it is spotted by the QS the bombs start to fall on it after it
has finished running, and don't reach back to the main warrior.
However, the resulting warrior is very compact.
Jedi Hunter follows a 0.5c scan with a scan for the end of the QS.
However, unlike all the others that use a 1.0c linear scan, it uses a
a longer-but-faster stepped scanner. I suspect that this is not a very
effective use of space, and that it would have been better to use the
space to raise the initial scan to 0.8c, or simply to shorten the
warrior. The attack is a SPL/JMP bomber, which again is a bit bulky
compared to the other kinds of attack. Finally, the biggest problem is
that there is no decoy at all! The warrior is long, so it suffers many
hits from the QS and, unlike myBlack, it is not very resistant to DAT
bombs.
For comparison to the precision scans, Hyperclear achieves excellent
results for a general-purpose warrior but the gap between it and the
precision scans is huge; it score is about 75% of Force Feedback,
compared to 90-99% for the other precision scans. In some ways grey
warriors (in which the constants are unknown) are a richer challenge
because you have to write something that resembles a normal warrior
rather than just rehashing the precision scan theme. If the boot and
replicator steps were not known then something like Hyperclear would
probably be the best approach.
Conclusions?
-- Precision scan is the best way to handle any long white warrior.
-- White Quickscans are almost inneffective.
-- Decoy generators are great because the decoy is always ideally
placed.
-- Cleanup is easy once all the components have been found.
Regards,
Robert Macrae
|