This post used to be called Don’t Drink And Drive! However, it strangely does not have anything to do with driving. But it does deal with alcohol consumption. :D.

It all started when cfgt invited me to play a game of Counter-Strike over the hols. Not wanting to go through the hassle of Steam and HL2, we decided to play the old CS (CS1.6 I think). I’m used to CS:Source, it being more realistic, and having nicer graphics, so I discovered my skills in the old CS severely lacking. So he asked “Oi, drunk?” (note: it’s a known fact that I play FPSes like Quake 3 or Unreal Tournament in a very drunken disorderly manner. People watching me play will get dizzy).

But I wasn’t drunk. In fact, I think quite a few of my readers would know what happens when I consume alcohol. Some embarassing incidents happened before. But that gave me a very interesting question, which I would pose to myself… what happens when I consume alcohol and play a game?

And me being me, I put it to the test. With numbers.

Brief specs

• Game played - Counter-Strike 1.6.
  I originally planned to include DoTA and Magic: The Gathering into the list (because alcohol impairs judgement), but I couldn’t measure any information accurately from those games. Thanks to the mess that was left behind by Valve with their idiotic Steam components, I decided not to install CS:S on my laptop despite me being more familiar with it.
• Data Harvesting Methods - AMX Mod X + WinCSX
  I found this metamod for CS which collects data, absolutely brilliant. Props goes to the makers of AMX Mod. And also, I found a nifty little software called WinCSX, which extracts statistics from the AMX Mod statistic data file. Great works actually. I used AMX Mod because another statistics metamod, Statsme, couldn’t work (or rather, I couldn’t get it to work).
• Data Harvested - typical CS stats
  Kill Counts, Death Counts, Shots, Hits, Headshots and Team Kills (aka Friendly Fire).
• Controls - me, not intoxicated.
  But seriously, though, the controls are what a typical game looks like. Also, because the 150 rounds take place over 3 days each (for the sober and intoxicated tests), there are other forms of control, like the amount of alcohol quaffed, et al. It will be discussed at the bottom.

Long story cut short, here are my results (note, they are per-game averages):

HS stands for Headshots, TK stands for Team Kills (aka Friendly Fire). Simple accuracy is the hits per shot ratio, while simple efficiency is the kill-death ratio.

This is the short game version. I played 150 rounds of CS, split into 30 5-round games. Each time I play against the default Hard Bot. Each game is split into 5 rounds, and I am auto assigned on each team, so some times I would play on the Terrorist team, and sometimes on the Counter Terrorist team. Now, lets head on a bit to the explanation, ok?

The Score (and Explanations)

The combined score is calculated with efficiency weighted with the complex accuracy score and times a hundred for impact.

The complex accuracy score is weighted with headshots per shots (because shots and headshots are positively correlated -see graph on right - this was done so that random spraying of bullets causing plenty of headshots isn’t counted as being accurate), and weighted against team killing (so that the more team mates you killed, the lower your score will be)

I use accuracy a lot in weighting because that’s what alcohol is supposed to do - affect your judgement and coordination, hence, I take into account Headshots and Accuracy.

Team Killing was taken into account because alcohol is supposed to affect judgement. I found in previous instances, after some games, and I switched team, I have a tendency to TK. So that’s why I included TK and Auto Assign Teams into account.

Score Breakdown

My score is an of 30 games (150 rounds) played, and my combined score for being sober is 126.38 (±19.14, CI 90%) while my combined score for being intoxicated by just one glass of wine is 77.548 (±11.02, CI 90%). Also, if you would notice, my simple efficiency score increased a lot when intoxicated (because I kill more). Which was certainly most interesting. I discuss it below. I also seem to TK more when I am intoxicated.

Aditional Controls Implemented

Gamewise, I implemented a rule that only one gun should be used - the Desert Eagle. Every bot (and me) used a Desert Eagle. This is because the Desert Eagle is the single most powerful handgun in the game. And because it only caries 7 rounds, everyone will have to reload the weapon quite often, which in itself, a tactical skill. Also, it would be very easy to kill with a Desert Eagle.

The map played is de_dust, and planting the bomb is NOT the object. Who won each round was also NOT the object of the game, so it didn’t matter.

I originally had another long game experiment - involving playing very long games (100 rounds) for each experiment (sober and intoxicated), but I knocked out when I was doing the intoxicated version. I was too dizzy to continue playing.

Please do note that the above “intoxication” is very mild - from one glass of wine (2002 Shiraz). And a small one at that. Here it is:

Half that glass is roughly 125ml of wine (and yes, don’t lecture me, I know it’s not a wine glass). And to control it, I take only one glass for each experiment. The wine was also controlled, that I only take wine from the same bottle, which has 13.3% alcohol -

A great many of you readers know that I don’t take too well to alcohol (symptoms include dizziness, vomitting, and plain ‘high’ness), probably due to lack of aldehyde dehydrogenase, according to ayjk (anyone wondering why someone like me would come to own 2 bottles of wine, it’s because they were cheap at the market. Not particularly good tasting, but ah). Which was partially the reason why I only took a glass, instead of copious amounts.

Even so, with one glass of wine, it is enough to change my accuracy scores drastically (by about half)

Quirks

Naturally, I noticed some quirks in my data. For one, my kill-death ratio when intoxicated was higher (because I killed more) than my non-intoxicated state. Also, I noticed that the number of times I fired also increased (which also means the probability of getting higher kill counts increased).

This was technically not supposed to happen. In fact, I first thought that getting intoxicated would severely depress my scores (which are sad enough already), but instead, I got more kills. Curious as to why that happened, I investigated further.

Perhaps it was a good thing that I recorded my gameplay as well. I watched the recordings, and discovered, that I have been more gung-ho when intoxicated. Anyone who has played CS against me would know I am a cautious shooter, preferring to hide behind crates and barrells (with CS:Source, it’s even more hiding).

Intoxicated, I found that I played quite a bit like Abel. But still being cautious. That explained the increased shootings. I was more reckless with my bullets.

Curiously, when I was intoxicated, I played better as a CT, but when sober, I play better as a Terrorist. Haha, now you see why certain religions don’t allow alcohol in them (in case they want to run amuck and start shooting people, and driving planes into buildings, they’d do it better sober, or knurd).

Also, I’m quite consistent in my shootings. Look at the graph to the left. It shows my kill count-simple accuracy correlation. I compare my sober state correlation with my intoxicated state correlation, and surprisingly, the R² value is the same. Hmm, odd? You bet. There are other things quite consistent too, in my shooting scores. I’ll leave you to figure what.

And to those mathematically inclined, my difference ratio between my simple accuracy scores (rounded) when I’m sober and intoxicated is a nice 1.618 (which is the Golden Ratio, or phi, in case you don’t know)

Conclusion

Conclusion? Alcohol DOES affect judgement. My combined score is biased towards accuracy. The more accurate a player is, the higher score he’d get, not the kill-death ratio score that most people normally use. It does say much about alcohol affecting my accuracy (also caused by me shooting more often, because I’m more brazen). If you were me, don’t drink and drive. It’s not worth it.

A lapse of judgement in a computer game merely causes digital death. But a lapse of judgement in real life, and you’d be meeting Death (with a Capital D) real soon (you must admit, he has a pretty sweet looking scythe, with the blue blade et al).

Ah, but I guess I know what to do at the next LAN party. Not boozing of course. Can’t stand that shit. I’ll play more recklessly! Just need to get into that state.

Cool? DiggDel.icio.usTechnoratiFeedsterFurl
RSS feed for comments on this post  |  TrackBack URI for this post