The issue is seats that have significant geographical barriers to being able to being able to be formed into the usual sizes. Like the two Scottish seats, it is deemed to be unfair to merge the island with the mainland. Once you've made that decision, it's a simple choice of having a giant constituency larger than all the rest, or two smaller ones.
It went through parliament, so got the agreement of the parties. Given this was during the coalition when the LibDems reneged on the timetable, they could have very easily objected to splitting it as the price of getting the legislation through.
If it was being done purely for party-political gain as has been suggested, do you really think it would have got through?
OK, the legislation specified two severely undersized Scottish seats (that have always been there), and mandated splitting the Isle of Wight into two undersized seats. At the time it seemed almost certain that of the four seats, two would go to the Conservatives, 1 to the LibDems and 1 to the SNP (who at the time were a minor party, not a perceived threat to the Conservatives). I don't like throwing accusations of this kind of gerrymandering around, but this arrangement certainly gave a small overall advantage to the then coalition parties.
The two Scottish seats have traditionally been so small in population because otherwise you would have had a seat with an extremely large land area and extremely poor communications
within the constituency. It arguably would've been politically very difficult for the Government to have attempted to go that way.
That's not really true for the Isle of Wight. Sure, the single seat was vastly oversized and needed splitting somehow. But communications with the mainland are relatively good and the geographical areas are small. Making a seat that - for example, included parts of NW Wight and Lymington would have been awkward but no more so than some of the seats the boundaries commission has come up with. And it would have been far more consistent with the Conservatives' claimed aim to equalize population levels. The fact that the then coalition decided that awkwardness was not acceptable for the Isle of Wight (a decision which coincidentally favours the Conservatives) - and therefore specifically legislated to prevent it - but apparently decided such awkward constituencies were acceptable elsewhere - is at best inconsistent, and certainly suggestive that there may have been some political motivation (though of course it's impossible to be sure).
Of course, when you throw in other issues - like the fact that this review attempts to remove a minor pro-Labour bias in the distribution, while ignoring the vastly bigger pro-Conservative (and to a lesser extent Labour) bias caused by the whole electoral system, and the way the boundaries commission has been forced to go not by population, but by registered voters just after the registration system was changed in a way known to have excluded many voters in groups more likely to support Labour - and the argument that the whole thing is at least in part motivated by the desire for political advantage becomes very compelling.
That's why these are initial proposals. People can make their representations as to why they believe their ward belongs in a different constituency. Assuming the arguments have more behind them than "they don't vote the same as us" and they can keep the numbers balanced there's no reason why adjustments can't be made.
But representations will only be accepted and considered if they fall within the legal framework that the boundaries commission operates. That very severely limits the ability to improve some of those constituencies. My point is that some of the constituencies are going to be so ridiculous because the legal restrictions placed on how constituencies can be made up make it mathematically extremely hard to do anything else. No amount of representation is going to change that significantly.