I was using crashing into a pole because it's something where you can see where you could plausibly have a huge disparity in sentencing outcomes in relative terms (there are always going to be thorny issues when you kill someone because even if you weren't driving drunk they're going to be picking apart all the different fault and negligence aspects anyway). For killing a pedestrian, you'd presumably make it so that killing someone while driving under the influence is presumptively negligent homicide as opposed to something like vehicular manslaughter.
Additionally, it's disingenuous to say there's a "near 100% chance of endangering people around you". That's saying that there's a 100% chance that you've increased the risk by some amount. There's nothing even close to a 100% chance that you will actually do any damage to people around you, or crash. According to [1] there are 112 million self-reported episodes of impaired driving, and 1.4 million arrests, but 10,000 deaths. Obviously these are not hard numbers, but my guess is that driving under the influence likely triples or quadruples an existing (small) risk that you would kill or harm someone. That's not a trivial thing, but it would still mean that in the vast majority of cases, if you drive under the influence, everyone gets home safe.
And in any case, I'm not in favor of people driving under the influence. Similarly I don't think people should drive while tired (it's a similar rate of impairment), but the simple fact is that these crimes by themselves don't have victims, they are just occasionally the cause of actions which do have victims. You can certainly make the case that drunk driving should be against the law for any number of reasons, but that's a separate issue as to whether that crime actually has a victim.
1. http://www.cdc.gov/MotorVehicleSafety/Impaired_Driving/impai...