My thoughts:
(1) When it comes down to it, I don't actually care about ice bolt being less powerful than the other bolts due to the fact that it's water, which is less of a spell-damage element thematically than fire (or dark, for that matter), and already has access to extremely effective CC (squid, congeal, permafrost shield to the extent that it counts).
(2) With that being said, it is less powerful. I think it should remain less powerful, but there's room to boost it and still have it weaker than the other two.
(3) Splash damage doesn't make sense to me thematically, whether splash to creature or splash to player. If we were going to add splash damage to one of the bolts, my vote'd be fire bolt.
(4) Splash to player wouldn't be that bad, but it'd be more of a buff than I'd really want to give the card. I would see that effect as quite strong, because it would functionally mean that (assuming you had quanta), the card would be "kill target creature and do ~10 damage to player".
(5) Splash to creatures is interesting but complicated to implement, and actually kinda weakens the bolt, since you are no longer able to easily take out creatures with any reasonable amount of HP without saving up a lot of quanta, assuming you meant that it splits its damage. If it doesn't split its damage and just does 100% to target and 50% to each adjacent creature, you've buffed it way too much, since it does 133% of fire bolt in total damage and becomes the highest-damage mass CC card in-game by a wide margin.
(6) The buff I like is making freeze guaranteed but for a lower number of turns. I'd be fine with guaranteed two turn freeze, but something I thought of that I kinda like better is a randomized freeze chance - something like 50% - 1 turn, 30% - 2 turn, 20% - 3 turn. That'd work out to 1.7 turns on average, but I'd be open to plenty of other breakdowns (provided there was a quite low chance of 3 turn freeze so as not to step on the toes of the freeze card). This also keeps some of the chance-of-freeze flavor of the current card.
So, basically, I'm gonna vote for the guaranteed low-turn freeze on the poll, but I think it'd be more interesting if it were a randomized duration (weighted to average low-turn) rather than a flat amount of turns guaranteed.