My reasons for buying weren't to win a Knuth reward check. I wasn't even sure of being able to understand the book, let alone find an error.
(Unless Rs. isn't Rupee, but Google thinks it is)
You need to be able to trust a book. I received a misprint of a book I ordered online once, duplicate pages, missed half a chapter. I didn't need that particular chapter for the course I was taking then, and it was a week before the exam so I didn't send it back, it's a really cool book (Computational Geometry) that, like other interesting study books, I would pick up again and browse occasionally, if it weren't for this mess up. It's a shame, really.
That said, lunch in the company cafeteria was about 50 Rs, if I recall. So, where I would spend about 5 to 8 USD at home on a modest lunch, they would spend what amounted to 1 USD. Salaries are scaled accordingly, so yeah, those Rupees come dearly when you have to earn them locally.
100 Rs might exchange for 2 USD, but it feels like having to earn 10 USD (or more).
Also over the past year, the value of dollar has increased to Rs 67(or the value of Rupee has fallen, whatever). The point is, it burns more now. And the burn amount keeps fluctuating.
Apologies if I am pedagogical but it is "Gandhi" and not "Ghandi".
I paid about $150 for my volumes, and while it was more than I wanted to spend, that's actually quite a deal. That comes out to less than $40 per book, and each book is so dense and authoritative that another equivalent book probably costs on the order of $100. In other words, at least in the USA, if each book was sold separately as textbooks, I wouldn't be surprised if it cost $400 in total.
As you have detailed, Knuth's work here is seminal, and he will probably continue writing until the end of his life (and TAoCP won't be completed by him). If you intend to truly read and understand all of the material he has presented, then you're in it for the long run. Unless you are very educated or a researcher in the field, this is probably years worth of work. This includes reading all the mathematics, understanding it, and doing the exercises. (Of course, not all exercises can be reasonably done by anyone, as some of them are actually open research problems.)
Because you'd be in it for the long run, cheap paper copies are a bad idea, unless you don't mind pages falling out and deteriorating. Doing a semester-long course in some subject? Okay, that's fine. But purchasing for personal reasons the opus magnum of one of the greatest computer scientists? That feels a little different.
Although you've already paid a measly one-tenth of the price already, I'd consider paying the full price, and take your first as just a tax on the lesson learned. In the USA, many of us would get a 5-10% markup anyway (sales tax), so it's not all that devastating.
Edit/Addendum: Lastly, one important thing. For anyone reading this comment: While there is prestige in finding errors, that should not even be remotely close to the reason you purchase the books. The checks, in their original intent, are small awards for finding mistakes, and only somewhat artificially have they turned into something more. The true reward, as cliché as this sounds, is being able to read Knuth's prose and learning from it.
If you believe that having such an achievement will slingshot you ahead in business or academia, this is almost surely false. While the award is nice to get, it won't give you much in return. I own two checks, and they're on my résumé, but I don't think they've carried much weight, if at all, at the multitude of companies I've applied for and been at. I think I've been asked about them only once, and even that time, the interviewer wasn't interested in hearing about numerical algorithms or boolean satisfaction.