How would someone actually implement an effective mechanism for controlling suspicious voting? For example, if 10 people log on from the same corporate network,

How would someone actually implement an effective mechanism for controlling suspicious voting? For example, if 10 people log on from the same corporate network, say Citibank headquarters in NYC, doesn't Stackoverflow/HN/Reddit see them as coming from a single location since they won't have distinct IP's coming through the corporate gateway? If they all vote in a similar pattern, that is certainly an indication of collusion. But what if it's not? I realize we're not deciding people's lives here, and yet it is a noble goal to try to create an online world more perfect than the one in which we live. If it is possible to identify voting collusion, can we not remove downvoting and be confident that we are close to a virtual utopia? As confident as Joel sounds in this post, I suspect the identification process is not quite so seamless. I don't know how many other technical scenarios exist similar to the Citibank one above. Perhaps someone here has tried to solve this problem and could shed some light? While Reddit would be an entirely different thing (mostly better) without downvoting, I have often wondered what HN would look like if no one had downvoting capability - is it needed at all? I would think the best comments would bubble up as they do currently and the worthless or offensive ones would either get forgotten or responded to, without the need for being downgraded.