The main ways of memory retrieval are recognition and recall.
Why has it been found that recognition is "easier" to perform, meaning it is usually faster or is more likely to yield an accurate retrieval?
I'm not familiar with the paper Ofri cites, but will agree with the OP that recognition is generally considered to be an easier task than recollection, and successful recognition considered weaker evidence for any particular memory phenomenon. One common explanation is that recognition can manifest psychologically simply as a result of the increased cognitive fluency brought about by priming -- see for example Alter & Oppenheimer 2009. Priming effects are extremely robust -- subjects will show priming on an image they saw once, a year ago, for less than a second. (See Brady et al. 2011 for review.)
If recognition amounts to the psychological state wherein some stimulus is slightly easier to process than another, and this effect can be achieved with essentially no cognitive work, but rather as a manifestation of a kind of mere exposure, then it seems clear why recollection -- which requires a complex dance of inhibition and excitation in order to activate some target thought or memory from amongst a host of competitors (see for example Anderson et al. 2000) -- should be the more difficult of the two processes to accomplish, as, in fact, it generally seem to be.
Alter, A. L., & Oppenheimer, D. M. (2009). Uniting the tribes of fluency to form a metacognitive nation. Personality and Social Psychology Review, 13(3), 219-235.
Brady, T. F., Konkle, T., & Alvarez, G. A. (2011). A review of visual memory capacity: Beyond individual items and toward structured representations. Journal of Vision, 11(5).
Anderson, M. C., Bjork, E. L., & Bjork, R. A. (2000). Retrieval-induced forgetting: Evidence for a recall-specific mechanism. Psychonomic Bulletin & Review, 7(3), 522-530.
It is not. At least not always: In this famous experiment Tulving and Thomson show that under certain circumstances recall can be better than recognition.
It seems that the reason why recognition is usually more accurate than recall, is the context. Usually, the context in the recognition test is very similar to the conditions in the learning phase - the subject sees a list of words. In the free recall test, the context is different - the subject is not given any external stimulus.
Tulving and Thomson changed that in their experiment: the context of learning was more similar to the recall test, which resulted in higher performance in the recall.
This indicates that the encoding in memory has some context specificity. Items that were encoded using a visual stimulus will be easier to recognize or recall if the same or similar stimulus is presented during the test.
Note: I interpreted the word 'easier' in the question as 'more accurate' because typically accuracy is measured, and not the 'ease' of the recall/recognition, which could be defined in many different ways.
Tulving, E., & Thomson, D. M. (1973). Encoding specificity and retrieval processes in episodic memory. Psychological review, 80(5), 352-373. PDF
There is a lot of evidence that our brains work most fundamentally as pattern-matching engines ( simplistic, but sufficient for the purpose ). Which would definately favour recognition, because this involves matching a newly perceived pattern with an already stored one, something that we do well.
Recall, OTOH, involves keyed data access - that is, identifying the links via another piece of information - when you saw something, or a tray that used to have objects on. The cognitive processes are not as good at this because the way the information is retrieved is not based on time or place, which are abstract entities.
So "did the face look like this one" will produce quicker responses than "how big was his nose". This is why photofit and sketches are so important - they trigger recognition. Of course, this form of recognition - of faces - is so strongly wired that it can produce a lot of false positives. But that is identification accuracy, rather than recognition accuracy.
A quick explanation for why recognition may seem easier than recall is due to advantages that come into play for recognition that aren't present for recall. In particular, on a recognition test, subjects are presented with what are called "copy cues" -- in other words, duplicates of studied items. When studied items actually show up on the test, subjects are able to use two types of processing to respond: (1) recollection, which is careful, controlled responding that is based on retrievable episodic and contextual detail (present in both recognition and recall) as well as (2) familiarity, which is quick, heuristic-based deciding (generally more readily available in recognition memory). These processes are not unlike Kahneman and Tversky's famous processing types.
Ofri's answer is a good one, however -- it's very interesting to spot situations where recognition is actually less successful than recall. Recognition failure of recallable words is one such instance.