So it seems like my PCs are about to head off into the open plains during a massive snowstorm. I'm wondering, how best to RP the effects of cold and such? I'm thinking of treating it as sort of a physical attack on the characters every few rounds, with warm clothes functioning as armor. Does that sound right?

I'd like there to be something to roll against, though.

A game I'm trying to start is based heavily around a chess match. Knowledge of chess will thus be something that is at times helpful (or harmful) to the players, so in their bios, I'm going to ask for the character's Elo rating if they play.

From Wikipedia, Elo ratings correspond to skill level according to:

RATING      SKILL LEVEL
2851          Highest rating in history (Garry Kasparov)
>2600   World Championship contenders (the NPCs are in this range)
2400–2600   most Grandmasters (GM) and International Masters (IM)
2300–2400   FIDE Masters (FM)
2200–2300   FIDE Candidate Masters (CM), most national masters
2000–2200   candidate masters, experts (USA)
1800–2000   Class A, category 1
1600–1800   Class B, category 2
1400–1600   Class C, category 3
1200–1400   Class D, category 4
1200-0100  novices
0100           Minimum possible rating

Logically, a high Elo rating should be correlated either to a good Scholarship score, a good score in some other category with an appropriate stunt (i.e., a high Performance score for a character who is able to throw opponents off with acting), or other stunts (like 'Chess Prodigy' or 'Extensive Training.') I'm not sure how to try mapping these, though; a 2-refresh stunt 'Chess Genius' should give more Elo points than a 1-stunt 'Talented at chess,' and a 5-scholarship character should have the possibility of being much better at chess than a 2-scholarship one (all else being equal) but I'm not sure by how much.

Does anyone have experience trying to map a rating system to DFRPG, or any ideas on how best to do so?

Sure. Like I said, I'm not all that good at statting yet.

Info on the different types of big daddies is at: http://bioshock.wikia.com/wiki/Big_Daddy

I've been considering an NPC idea for a game I'm running as such: he has the bare minimum stunts/powers to enable him to magically kill people, and no other abilities that cost fate points or refresh. This saves his FPs so that he doesn't go down to 0 refresh from taking Lawbreaker and so that he can pay off the various compels from his Lawbreaker aspects (so he can be an assassin without going full on Omnicidal Maniac). He doesn't break any laws besides the 1st (although he's broken that one innumerable times before the game even starts), and has no magical ability besides the powers that let him kill people.

* Is this viable in a Chest-Deep game? In submerged?
* What combination of powers, if any, would let him accomplish this? Is there a set of evocations or something that actually have a strong possibility of killing, say, a baseline human, without costing more refresh than he could have? What about a stronger character, assuming he got an ambush (i.e., roll 0 to defend)?
* Assuming he weren't actually targeting the PCs, at least at first, would it seem cheap to actually drop such a specialized character into the campaign?

(I know that NPCs don't technically need to have >0 refreshes, but I generally like mine to so they aren't overpowered relative to the main cast).

(And obviously I'm not going to introduce an NPC to just one-shot the PCs; that would be mean. But he might be a sort of Kincaid-like 'possible ally, but very scary' fellow).

1. The GM. If the GM goes away, the plot will have a lot of trouble continuing. Even if the GM remains, if they choose too lenient of a gaming style ('Monty Hall' GMs) it can get boring, and too sadistic or restrictive can kill interest.
2. Players. Even a great GM can't necessarily hold things together if multiple players leave. If enough leave, most of the city/NPCs will need to be redesigned (e.g., in FtF, since 4/5 of the original cast has left, most of the NPCs and folks that were developed in the city itself aren't really applicable anymore). At some point it becomes more useful to just do another 'setting the scene' round where the city and characters are setup. There's also the issue of where good 'jumping in and out' points are; if players leave during a sequence where it's not plausible for new players to be added, the whole thing can bog down and make it take longer to get to a point where new people can enter, which increases the chances of game death.
Another note on players -- players whose posts are hard to read/use poor grammar can drag down a game, simply because others won't always be as inclined to read or respond to such posts.
3. Consistency of posting. The longest game that I'm aware of on the board is the chat game, which meets twice a week at set times to post. This ensures that the game happens.
4. Setting. Some settings don't lend themselves well to being RPGed. It can work at first due to novelty, but eventually the logistical difficulties are too strong to be overcome. (This can happen in games set in other fictional universes, e.g. a Potterverse game or a Death Note game, because the magical objects in those games don't really lend themselves well to the DFRPG.) Other problematic settings are those where the main characters have to compete against each other (e.g., in the extreme, a Battle Royale game) because this makes it harder to work as a team or accomplish much of anything significant, and games where characters don't have well defined motives to continue being in the game (i.e., no explanation of why they don't quit and go home).
5. Particular difficult scenes/characters/etc. If a game features something really new, it'll take more work on the part of the GM, which can make the GM want to drag their feet a bit as that scene approaches. They might also be forced to pause the game if they can't stat or create something in time. (E.g., in my game, the characters may eventually need to fight armored vehicles, which are challenging to stat because there aren't examples in the books. So, if I didn't stat those ahead of time and suddenly the characters had to fight a Blackhawk or a battleship or something, I might have had to pause the game to hash out the stats, maybe even run a combat test or something). Long enough pauses can derail things.
6. Inclusion of gimmick elements. If Slenderman shows up out of nowhere, great, but it just got a lot harder to run and play the game because you've got a gimmick that risks overwhelming the whole plot. This can be mitigated if the gimmicks are built into the campaign (e.g., in the above, a major adventure, designed from the beginning, is dealing with Slenderman), but then it's not really a gimmick anymore.
7. Mismatched setting/character level. Some settings are better for, say, 'Chest deep' campaigns than 'Snorkeling' or 'On the Beach.' If the plot is 'the characters are generals in Winter's army as they march against summer', statting them as 'On the Beach' will limit what they can realistically do against Summer's forces. On the other hand, 'Snorkeling' characters are liable to walk all over, say, the characters from some of Harry's earlier adventures, and this can make the game boring.  

As a GM, sometimes I want my NPCs to invoke certain aspects that the players haven't assessed yet. Examples might include if an NPC has a Demonic Co-Pilot that the PCs haven't figured out yet, or if one of them is secretly a psychopath, or something. Normally, invoking implies that you declare which aspect you're using. Is it okay in some circumstances to say "My NPC is invoking his Trouble aspect" without saying what it is?