After SUPER shotgun, why go back (some do)?

Maes said:

(900 HP, the health of an undamaged ArchVile).

I think it's "only" 700 hit points, so maybe it could work.

[WH]-Wilou84 said:

I think it's "only" 700 hit points, so maybe it could work.

YEah, my bad. Still, the maximum theoretical of 300 HP is impossible in vanilla. The maximum achievable with the normal RNG table is 240-250 HP depending on circumstances, so if you happen to get 3 maximum damage shots in a row (or at least three whose sum is 700, which means two maximum and one average) it's theoretically possible.

But maximal damage hits are rare to begin with (I recall about 2-3%, I must look at the tables again if you want split-decimal accuracy) so getting two in a row is at most 3% x 3% = 0.09%, while getting a THIRD max damage hit in a row would be another submultiplication which would take us to an astonishing 0.0027% or about a 1:37000 chance.

Since you need two maximal damage and a 200-ish shot (you have more than 40% chances of a shot being > 200 HP), odds would be about 0.036% ot 1:2800. Even if you triple those odds (since you can yield the damage in any order such as "MAX-200-ish-MAX" or "MAX-MAX-200-ish" or "200-MAX-MAX", that's still ~1:900. YOu gotta kill A LOT of archviles to see that happen, and make absolutely no mistake/miss a single pellet on any of them. You're more likely to witness a single-shot lost soul kill.

If there are still any doubts, I can run some simulations for over 1 million consecutive double, triple, quadruple etc. SSG shots and reply with hard numbers (basically I'll measure how often the total damage would exceed certain thresholds like e.g. the life of a Revenant, of an arachnotron, of an archvile etc.)

Well, I suppose I would be curious to know the exact odds. It sounds like it's a rare enough occurrence that I should post the demo. I'll put it in the misc demos thread in the demos forum.

xit-vono said:

Well, I suppose I would be curious to know the exact odds. It sounds like it's a rare enough occurrence that I should post the demo. I'll put it in the misc demos thread in the demos forum.

I finished the tests (implemented Doom's rand calls and sequence in a simple Java program, as per the simple shotgun) and took 10000 samples of firing the SSG consecuteviely 2, 3, 4 and 5 times (I let the RNG run freely in the mean), add the damage for each run, and print it out. The position of the Doom's RNG was randomized with Java's rand()*256 every time a single shot of barrage started, to simulate different positions where it might be found during gameplay.

The full results are in an XLS file along with the code that generates the graphs and the Java code that does the sim. Keep in mind that these were obtained using DOOM's RNG, with an as close simulation as possible of its usage within the firing functions.

If you want a summary analysis:

Damage range for 2 consecutive SSG shots is from 340 to 495 (nowhere near 600!). This means that Revenant kill chances with two shots are 100%, since all possible values are >300. Chances to kill the next most tough monsters, Cacos and PEs are 79.58%. Not bad.

Damage range for 3 consecutive SSG shots is from 525 to 720, with values over 700 being at the very tail, with a 16/10000 or 0.16% chance. This doesn't rule out 3-blast archvile kills, but we're talking about an extremely rare occurence. However killing a Hell Knight/Arachnotron with 3 SSG shots is always possible, and the chances of killing a Mancubus are 83.245%

Damage range for 4 consecutive SSG shots is from 725 to 945. This means that an archvile will always die from 4 shots (OK, maybe with 1 million sims an occurence with all minimal damages could occur), but not a Baron of hell, even with all shots giving maximal possible damage.

I also did sims for 5 shots: damage ranges from 905 to 1050 (notice the overlap with 4 shots): the chance of killing a Baron of Hell is 87.75%, so stumbling upon a "tough SOB" kind of Baron is not entirely negligible.

You can download the sim bundle from here and run/correct/study the results yourself. I stopped at 10000 sims per case because Excel has a limit of 16000 or so rows per table, and to do more I'd have to use my own stat gathering functions, which is more than I can be arsed to do right now. I don't think the stats would change dramatically beyond this point, though.