...I bought the game therefor i own the game, ...
Technically, this isn't correct. I can guarantee that the EULA for every single piece of software you have installed on your PC (even free software) has a term which states that you don't own the software you have installed. Rather, what you own is a license to use that software. For example, this is from the Steam Subscriber Agreement (section D):
All title, ownership rights and intellectual property rights in and to the Software and any and all copies thereof are owned by Valve and/or its licensors.
There have been multiple reports in the New Vegas forums of the option switching back to auto-update mode. Sometimes it seems to be because Steam itself updated, sometimes it seems to be at random. Nobody has been able to find a particular thing these people are doing wrong, and the reports come from elsewhere as well, so the best guess is that the function still has flaws that cause it to not "stick" properly for some people.
Have these people tried contacting Steam's support and telling them about the problem? The more people that complain about this problem, the more likely/sooner it will be fixed.
By requiring people to be on their site to play their games, Valve probably experiences a dramatic increase in sales compared to if they did not require people to be on their site. I hypothesize the the primary motive for Steam's online accounts and game validation system is the cyber-equivalent of 'point of purchase' advertising. It is a brilliant strategy to foster brand loyalty (in this case, the 'brand" being the distributor) and it seems to work.
...Overall, you do make a good point and your theory does make sense. It's just that I don't see anything in the Steam client that screams "Attempted Marketing" (except maybe the News panel, but that can be turned off). Also, I don't believe that this would be their only reason for implementing DRM (in particular, piracy definitely plays a role).
I hear many guys saying they buy ALL their games through Steam.
Most likely a case of, "I have the facilities to buy games through Steam, therefore Steam is the easiest way for me to acquire games."
It is common sense to assume the world is flat but the evidence indicates pretty clearly that it is not.
Very well, where is the evidence which proves that most pirates forgo pirating popular games in favour of pirating games that are unpopular? In the meantime, http://www.tweakguides.com/Piracy_1.html basically proves the opposite. In fact, page 4 contains a table of the 10 most pirated games of 2008 created by a popular pro-piracy website and then makes an anolysis of it. The author basically concludes:
So far the information in this section has provided us with enough data to make some initial observations:
...
More popular/desirable games are pirated more heavily than less popular games. The entire top 10 pirated games list doesn't contain any truly unpopular games, indeed some of the most popular good quality games of 2007/08 appear on the list. Similarly when searching torrents, I found more popular games have far more individual torrent listings than less popular games. This clearly contradicts the claim that 'good games get pirated less' - we see more evidence of the fallacy of this claim throughout this article.
Finally, on the contentious topic of DRM, aside from Spore whose audience may well have fallen victim to DRM-induced hysteria, the presence of intrusive DRM appears not to increase piracy of a game. For example Call of Duty 4, Assassin's Creed and Crysis all have no intrusive DRM whatsoever: they all use basic SafeDisc copy protection with no install limits, no online activation, and no major reports of protection-related issues. Yet all were pirated heavily enough to have the dubious distinction of being in the Top 10 downloaded games list. But strangely absent from the list are several popular games which do use more intrusive DRM: BioShock, Crysis Warhead, and Mass Effect. This indicates quite clearly that intrusive DRM is not the main reason why some games are pirated more heavily than others.
Another notable fact about that table is that even Far Cry 2 and Fallout 3 made the top 10 list even though they were released less than 2 months before the table was created, they are very popular games (especially FO3) and they both use only basic disc checks for DRM.
Spore was poorly received, but was evidently intensely pirated.
The article I linked to already deals with both the Spore controversy and SecuRom issues adequately, so I'll just recommend that you read the article rather than copy-paste stuff here.
Apparently a large fraction of the lost revenues from piracy are thought to occur during the first 24 hours after a games release, before its actual popularity is established. Thus, the hype or expectations of a games popularity may in fact account for a significant fraction of the variance in piracy, but I still believe that actual popularity would only show a moderate to weak correlation.
The article also deals with day-zero piracy and a game's popularity's effect on piracy.