Just to add a little to the discussion of categories vs. no categories:

I just had a look at the results of last years coding competition.
The number of valid votes per category varies from 51 to 96.

When using no categories an effect of this could be, as beresk_let already mentioned, that submissions get overrun by other submissions that simple happen to fall into a more popular category.
E.g., could the top placed submission in an unpopular category never win as it gets overrun by even the last placed submissions in a popular category.
On the other side, this may in fact reflect the absolute "popularity" of submissions overall and not just relatively to each category.


Combining submissions grouped in categories with a single overall voting imho is a bad solution as it gives the impression that categories are important whereas they don't.

Similarly, I consider a "pre-selection" bad as well as this involves some "elitist"/"privilidged" board to meet decisions and exclude submissions on it's behalf. Imho the election of winners should be entirely up to the community and not be (event partly) influenced by some other authority.


So, from my perspective, it boils down to the question of categories vs. no categories. There are reasons for each variant. The most prevalent, from my point of view, is that an overall voting would really reflect the absolute popularity of an app. On the other hand, this results in unpopular application fields being entirely overrun by popular ones.

Another, addmittedly theoretic result of using an overall category is that, as far as my understanding of the voting process we use goes, we would need votes for at least 25 applications. These do not need to be a single vote but all people must have voted in total for at least 25 applications in order to ensure that an order amongst participants can be established to distribute all available devices (This gets even worse if we consider additional prices like money etc.).
E.g., imagine there are only 10 popular apps that receive votes and other apps do not receive votes. Then we could easily establish a ranking between these 10 apps. However, the remaining apps would all share the same rank as they are all placed last. In this situation it would be impossible to determine the remaining winners.
This is admittedly a very theoretic consideration and is also only as far as my understanding of our voting process goes. As we never did a similar voting with such a high number of candidates we do not have any experience from past votings.


These are my, more or less, two cents on this topic.