Attaching an updated patch for Py2.7.

* Kept OP's simple constructor call but renamed it from counts() to
Counter():
     item_counts = Counter('acabbacba')
* Followed Guido's advice and avoided subclassing from defaultdict(). 
Instead, subclassed from dict, keeping its familiar API.
* Avoided KeyError issue by defining __missing__().
* Added most_common() method to address a principal use case --
patterned after Smalltalk's sortedByCount() method.
* Added elements() method requested by Alex Martelli -- patterned after
C++, Smalltalk, and Knuth's examples.
* Made necessary subclass overrides to dict methods: __repr__, update,
fromkeys, and copy.
* Wrote docstrings, doctests, and motivating examples.
* Verified that instances are copyable, deepcopyable, and picklable.

Working on docs and unittests.

Nice example (most common words in a text file):

>>> import re
>>> words = re.findall('\w+', open('hamlet.txt').read().lower())
>>> Counter(words).most_common(10)
[('the', 1143), ('and', 966), ('to', 762), ('of', 669), ('i', 631),
('you', 554), ('a', 546), ('my', 514), ('hamlet', 471), ('in', 451)]