How would you change Hextech?

I've been enjoying the new patch immensely, but Hextech is not making everyone happy. I've had games where it targets a nothing character with one item and I steamroll like nothing. I had a game where I had 5 item holders, each spread so that only one would get hit by Hextech (2). My opponent had a Hextech (4), fluked my center unit and locked all 5. It was a painful way to get 2nd (40% to only affect 2 items, 40% to affect 3 and 20% to hit all 5 and he hit the jackpot). I wasn't THAT mad (after a 5 minute cooldown), but I can see how the inconsistency could be frustrating. You can position around Assassins and Blitzcrank (another contentious unit), but Hextech is just random chance.

So.......what are people's proposed fixes? I see lots of complaining, but few ideas. Here are a few quick ones I thought of that preserve the spirit:

1. Keep as is but shorten length of time items are disabled

• I don't like this option as the "all or nothing" randomness of the effect can be frustrating, but seems to be the simplest and the one that comes up the most.

2. Hextech 2/3/4 - Deactivate 1/2/3 random items equipped to enemy champions for X seconds

• Weakens the synergy A LOT. Probably too much. Maybe increase to 2/3/4 or 2/4/6. There's still randomness, but it's less likely to completely neuter a hero.

3. Hextech 2/4 - Deactivate the 3rd/2nd and 3rd item slots of all enemy champions for X seconds

• If the goal of Hextech is to have a counter to item stacking, this could accomplish the goal in another way. It would also have counterplay itself as the order you place items on your units would matter (key item in first slot). You could reverse the order (1st/1st and 2nd) to power it up considerably (and potentially encourage stacking instead of discouraging it.

4. Hextech 2/4 - Deactivate 1/2 random items on all enemy champions for X seconds

• A variant on 3). Again, could encourage stacking instead of discouraging it as a way to "shield" key items from the effect.

Duration has been left blank because that really needs testing to iron out. Thoughts?