Piracy is a difficult issue. Since this topic is about how piracy should be handled, and not about whether or not it is right or wrong to pirate, I'll try to explain a few things with economic concepts.
I'll start with empirical data (usually it should be the other way around, first develop theory, then look at data, but this case is different). The reason for that is that it's important to realize we don't really have anything to go on here. Piracy is a difficult phenomenon to measure or estimate and it's even more difficult to estimate the effect of piracy on sales. Measuring piracy is difficult because it's illegal and has somewhat of a stigma and some people think it's cool to do. This means surveys are probably going to result in answers which are far from the truth. Pirating games by using the copy of a friend can't be measured, and torrent downloads can be legal downloads (in some countries it's legal to download a copy of your game if you own it, and even if it's illegal the "pirate" may still own the original).
One report by the U.S. GAO mentioned that
Three widely cited U.S. government estimates of economic losses resulting from counterfeiting cannot be substantiated due to the absence of underlying studies.
An oft cited report by the FBI reporting losses of $200-$250 billion per year "has no record of source data or methodology for generating the estimate and [...] cannot be corroborated."
That is to be expected. Not only is measuring or estimating piracy difficult, estimating the effect of piracy on sales is even more difficult, if not impossible. To give an example, a friend of mine had his HDD full of pirated movies, games and music. However, he never watched any of those movies, played only a few games (mostly for only a few minutes to try them out) and listened to only a few songs. He pirated just because, really. If those movies and games had not been available for free he wouldn't have bought any of them. Yet, in the (torrent) statistics, he would've downloaded movies, games and music worth thousands of dollars.
I myself, on the other hand, pirated games a few years ago and actually did play them (examples include GTA:San Andreas, Mafia and Oblivion). I didn't have much money, so one pirated game wouldn't have been equal to one lost sale, but some sales were lost (I have since bought most games I ever pirated and since I've been using Steam I don't pirate any more at all).
These examples show that it's difficult, if not impossible, to properly estimate the substitution rate, since they are different per downloaded copy (depending on the item itself and the person who downloaded it). To give another example: if one would regress the number of sales on number of copies pirated, there may well be a positive relationship. However, "number of sales" is NOT the variable you'd want to use: you'd want to use a variable like "willingness to pay". Usually, willingness to pay is measured by looking at... sales and prices. But because there is now a free alternative, it can't be measured. People pirating a certain game might well be the result of their unwillingness to pay for it.
Since empirical research isn't going to get us anywhere, let's look at economic theory to try and find out what the effect is and, knowing the effect, what can be done about it.
The first thing which anyone discussing piracy should acknowledge is the fact that there _is_ piracy, no matter what. No DRM has managed to make piracy impossible. There is, in other words, always a free alternative (of course, there are a few exceptions, like games which are fairly obscure; however, next to all games we play are available for free as well).
Knowing that there is a free alternative, WHY would anyone PAY for anything? There are two main explanations: different (better) product and social reasons. I buy games because I want to support good developers and publishers. I even backed 8 projects on Kickstarter for that reason, while I'm actually not interested in the project itself in a few cases. Buying games makes me feel good about myself (to put it in microeconomic terms: I gain utility from paying for the product).
This social (or ethical) factor can be influenced by publishers and developers. I would never pirate a Double Fine game because I'm in love with those guys. I wouldn't pirate a game from CD Projekt either. Both would make me feel like a dick. Pirating from Ubisoft or EA, on the other hand? I have no issue with that. The only thing that's keeping me from doing so is the fact that they haven't released any games I'd even want to pirate. Treating customers well is a way to increase the cost of piracy and increase the benefit of buying games. (That's not an opinion, BTW: that's microeconomic theory (cost/benefit)).
Second way to reduce piracy is by offering a superior product. This is what Gabe Newell is saying when he states that "piracy is a service problem."* This is something Microsoft, with GFWL, doesn't seem to comprehend. For example: I heard GTAIV was a bad port and decided I didn't want to buy it without knowing whether I liked it. So I pirated the game, played it for a few hours, loved it and immediately bought it off Amazon. I uninstalled the pirated copy, tried to install the LEGITIMATE copy, and I couldn't get it to work. It first it asked me to create several accounts, which didn't activate. When after literally half an hour or so I managed to activate them, but then GFWL stated the password was incorrect. An hour later (no joke), that problem was solved as well. GFWL then demanded to download the latest patches and wouldn't let me do otherwise. It stopped downloading halfway through. During the second attempt it did download the update, but then it crashed when it tried to install the patch. And again. And again. Ultimately, I quit. I tried it again a few days later and it still didn't work. A few months later I found the courage to try again, and it worked (probably due to a GFWL update). I haven't touched games using GFWL since.
Long story short: it was easier for me to pirate (including the latest updates!) then it was to get a legitimate copy to work. Obviously, this only encourages piracy since pirates offer a superior product at lower cost! No matter how bad piracy is, DRM which hurts paying customers more is always a bad thing to do (and one reason why I never play Ubisoft games, I would want to try Assassin's Creed II but I won't due to their stupid DRM).
So, legitimate copies should offer a higher utility than pirated copies by including DRM which is somewhat annoying to pirates (if only because it makes it more difficult to update to the latest patch because the right crack is unavailable or difficult to find) and unobtrusive to customers, include no DRM, or include very bad DRM which hurts pirates even more than paying customers, if and only if the loss of sales to legitimate customers put off by the draconian DRM if offset by the increase in sales from pirates who don't want to put up with the even worse problems they face if they try to get their pirated copy to work. The last option is fairly unlikely IMO and would be difficult to measure.
DRM is not the only way to improve (or decrease!) the utility of legitimate copies relative to pirated copies. Another way is to include extras which can't be offered by pirates. Achievements, save file back-up in the cloud, communities, multiplayer and support as well as physical items, such as nice boxes, manuals, maps and other goodies. Steam offers some of the first few items, while hardly any publisher releases collector's editions for PC games, which is kind of odd considering the fact they always complain piracy is really problematic on PC. A collector's edition can be a reason for a gamer to buy a copy instead of pirating (and it's no more expensive to sell a PC collector's edition if you also sell collector's editions for PS3 and/or Xbox 360 and there is a regular PC edition as well).
Bottom line is: treat customers well, use no DRM or unobtrusive DRM (unless you really know what you're doing, which you don't) and offer services or items pirates can't provide.
There is one more option, of course, which is suing pirates. The effect of this is unambiguous: lawsuits cost money (lots of it) and might decrease the social or ethical cost of piracy (because people think the company suing individuals are immoral bastards), though it might also scare pirates (however, this may not actually result in actual sales if the substitution rate is zero, or in very few extra sales if the substitution rate is very low). However, lawsuits don't seem to scare away pirates (piracy, as measured by torrents, didn't go down after such cases AFAIK).
Of course, the government could do something about piracy as well. In the grand scheme of things, however, piracy is unlikely to be a problem. If consumers don't spend their money on a game they spend it on something else, so other sectors profit at the expense of the entertainment industry. Plus, piracy increases welfare. And there are severe privacy implications for most measures. The result is a trade off, which depends mostly on one's preferences (besides the effectiveness of such measures).
*he goes on to say it's "not a pricing problem", which is rather odd since it's typically about the price/performance characteristics, the two being two sides of the same coin. However, if the prices are low enough to make the pricing factor insignificant compared to the social/ethical factor, the product/service is more important than the price, but the social/ethical "price" and benefit are still very important.