Say that a one-celled organism is sitting here. It is complete, it is all. Now something happens. Suddenly a mutation that happens to not kill it off or do nothing instead lets it have the "information" needed to become something else, two-celled for instance.
No-one has yet presented a decent definition for what "information" is, or what a change in it means so let's put the whole concept aside and look at the facts that we know. When a single-celled organism reproduces it does so by doubling it's genetic material and then splitting into two halves (loosely speaking - they're not always equal in size) each containing a full set of genes. Trouble is the copying process by which genes are doubled is not perfect so there are errors in the copy. We call these errors "mutations". Most mutations have little or no effect, some are "harmful" and reduce the ability of that single celled organism to survive and reproduce in it's environment others are "beneficial" and increase that organisms ability to survive and reproduce.
(I've put "harmful" and "beneficial" in quotes, because they're very much value judgements rather than absolute qualities - what is harmful in one circumstance may be beneficial in another, and most changes in fact involve a mixture of pros or cons)
Now, it doesn't take a genius to figure out that an organism that is better at surviving and reproducing is going to have more descendents that one that is worse at it. This is Natural Selection.
Chances are the shift from single-celled to communal organisms was made because of a change in the environment. Just as in the example posted above.
It seems to me that evolution is sort of like playing yahtzee for a billion years and rolling all good numbers the entire time, blindly and with no purpose.
No. It's not. Mutation is random, and produces a mixture of good and bad rolls. Natural Selection weeds out the bad rolls and keeps the good and, thus, over time will optimise an organism with the best genes for living in the environment it is found in.