I can see the problem. While it doesn't seem that the starter should take 100% of the blame when a reliever lets all his baserunners score, it also doesn't seem that a reliever should be penalized for letting up a couple RBI hits when he wasn't the one who let on the runners who scored. After all, a starter can let a couple baserunners get on to lead off the inning and not get penalized as long as he gets out of it, so why shouldn't a reliever be allowed the same cushion?
What it comes down to, in my opinion, is that ERA just isn't that great a statistic for judging the effectiveness of relievers. Starters (rightfully) have the luxury of getting into jams as long as they can get out of them. If they let allow two baserunners an inning for six innings without allowing a run, then more power to them - they've done their job. The ERA would be 0.00, and at the end of the day, that's really what matters. (Of course, the odds aren't too high that a starter would continue to put up a low ERA when he's allowing two baserunners an inning, but that's another story.)
With a reliever, on the other hand, the ERA doesn't really show much of the picture. The fact of the matter is that the reliever's job just isn't the same as the starter's job. Sure, there are times when a reliever will come in to open the inning and simply finish it cleanly himself. However, there are also plenty of times when he comes in to bail out a starter or another reliever. For the reasons that this article pointed out, ERA isn't going to be too reflective of the reliever's effectiveness. WHIP, on the other hand, is a lot more useful. The main reason why that's the case is that when a reliever comes in with runners on base, his job is to prevent any more runners from reaching base. If he lets up two hits and allows two runs to score, then even if he walks away from the inning without any damage to his ERA, he hasn't done his job.
There are other factors to take into account also. It's important for a reliever not to allow hits with runners on base, but it's also better to let up a single than a home run. So stats like HR Allowed, SLG against, etc. are also useful. If there were a version of WHIP that were weighted to assign more weight to extra-base hits, then that would be that stat I'd use to evaluate relievers.
Of course, all of this doesn't change the point that it's not really fair to the starter when a reliever allows all of a starter's runs to score. Even if we were to judge relievers by WHIP rather than ERA (meaning that the fact that the reliever's ERA comes away clean doesn't seem quite so unfair, since that's not what people would be judging him by anyway), that doesn't change the fact that the starter got screwed. However, if you calculated the league average percentage of inherited runners scored, then you could create a version of ERA that assumes that the league average percentage of inherited runners scored. It would be a little complicated, since the percentage would be different depending on the exact number of baserunners and outs that the reliever inherited, but it seems like it should be pretty do-able. It still wouldn't be perfect, since it wouldn't take into account things like the ability of the hitters that the reliever had to face (although I suppose that the average OPS of opposing hitters could be factored in there also).
Anyway, that would be my solution. I'd use a version of ERA that assumes the league average percentage of inherited runners scored to evaluate starters, and I'd use a version of WHIP that assigns more weight to more powerful hits to evaluate relievers.