The answer depends in part on what is meant by "most important," in terms of what criteria signal that a book or article is "important" to a field:
Important to whom? Americans? English-speaking audience? Foreign researchers? Mainstream scholars or scholars taking more alternative/critical perspectives on topics? (some things are taboo, even in the presumably democratic Western science, due to institutional biases and paradigm norms shared by the intellectual elite in a given field at a given time). Paradigms vary, especially in more subjective fields like social sciences.
For example, a psychologist who considers him/herself a "behaviorist" would send you to read B.F. Skinner, whereas a sociocultural theorist will send you to read L. Vygotsky. Are both "important?" Indeed, but in different ways and to different extent, depending on the needs and likes of their respective audiences.
As you observed, there is a lack of systematic structure in scientific intellectual development and inquiry patterns, despite much work and honest effort aimed to maintain it. The real world is messy. Good ideas may become simply 'unpopular' and 'old-fashioned' for practical/political reasons like shifting of funding agencies' priorities, or the development of a certain methodological approach, or simply higher salaries for researchers with skills in some specific narrow areas (e.g., theories that can be tested through empirical quantitative analysis as opposed to theories that can be tested through logical inference or decade-long ethnographic observation).
Important when? Today, last week, in 2014, in the past decade, the 20th century, or since the inception of that discipline?
As you can see, there is possibility of stumbling into bias at every step of the way, and no single definition or 'correct' answer exists for determining the "most important" works in a given field. The only correct answer is, it depends.
That said, a few guidelines might help focus your efforts on initial stages of inquiry in a given discipline:
- Wikipedia it. (Granted, some bias is most likely already present at this step!) Wikipedia, although somewhat 'pedestrian' in its popular, mass-appeal model, can be a great way to discover key figures and references (at the bottom of the pages) in a given discipline. Just google any discipline (e.g., "economics") and browse away. The more narrowly you define your search, the narrower the focus of the theories and publications that you will see. Keep this in mind, and play with the breadth of your search terms. You can also straight-up google "most influential publications (books, articles) in _____ (your field of interest). For example, trying this for "economics" produces this list: http://en.wikipedia.org/wiki/List_of_important_publications_in_economics
Alternatively, try searching for something like "most controversial figures in economics" and see what you dig up. This could be fun!
(Note: The real fun of science is in taking long shots, i.e. tacking big-payoff mysteries from non-traditional perspectives. It takes a real scientist-at-heart to have the guts, means, and intellectual stamina to make a dent in traditionally established scientific views, and such individuals often risk being discredited in the process, earning a 'reputation' and becoming outcasts. What makes them deserving of respect is that they don't care. Plus, they also get to do real science. One possibly relevant example here, although the jury is still out: http://www.alternet.org/food/meet-controversial-mit-scientist-who-claims-have-discovered-cause-gluten-sensitivty).
For a more 'academic' approach, and in regard to journal articles in particular, you may try some academic databases or publishers that can fetch top-cited articles in a particular discipline or journal. The advantage of using a database is the ability to refine your search (e.g. papers with a specific word in the title or abstract, written in a certain range of years).
One example: http://top25.sciencedirect.com/archive/57
Ask real humans. Talk to someone (or better, many different someones) who have been in that field for some time and just ask them point-black what they consider most significant works in the field. You might also ask what works made the greatest impact on them, but be ready to get different answers (hopefully, the different answers will contain some interesting surprises).
As far as forums, it really depends on the field. In #3 above, ask folks what forums (or email lists) they read and consider useful for broadening one's perspective in a discipline.