The Ulster Unionists will get that many votes but in a small number of seats in Northern Ireland. The Conservatives might just win by a small amount but in a large number of seats, each of which translates directly to an MP in Parliament.
For anyone who hasn't watched them , I can wholeheartedly recommend CGP Grey's "Politics in the Animal Kingdom"[0] videos for an brief overview of why FPTP voting is as bad as it seems.
Suppose you have 20 districts, each electing its own MP, and each with exactly four citizens. One citizen in each votes UKIP, one votes Labour, and two vote Conservative. Conservatives will win 20 seats, despite getting only 50% of the national vote, and the other parties will have none despite getting 25% each.
Because it depends on geography as voting in the UK is based on who you vote for your local MP.
If UKIP's four million votes were entirely within 20 or 30 constituencies, they'd have won them all. But because their votes are spread out across the entire country, other parties pick up those seats instead.
The parties that do best are either those that get a ton of votes everywhere, or those that dig in and solely focus on specific locations where they can get a majority (the Lib Dems used to be good at this).
It's not random at all, it's just that the national vote doesn't factor at all (only the vote in each seat does).
It has the (often overlooked, IMO) benefit of a strong 1:1 connection between constituent and MP: even if you didn't vote for the MP, a majority of your neighbours did, and the MP is still your representative in parliament. Organising locally to boot out your MP is much, much simpler than organising nationally to shift balances of power. That has got to be a factor in keeping MPs (more) honest.
In PR, you can be in the deeply problematic situation that you like candidate A and dislike candidate B from party X, but you vote for A actually ends up going to B - in effect, you can only really vote for a party, not a particular candidate. This is a significant shift of power to the parties as they generally control (or heavily influences) who runs where and how the ranking that controls "spill-over" votes will end up. Parties are often run opaquely, through back-room deals and patronage, not terribly democratically.
The existence of "safe seats" in the current FPTP system, into which parties can (and do) parachute their key players makes your take on FPTP too optimistic. See my comments elsewhere for links about voting systems, already in use in the UK, that seek to balance local representation and proportionality. PR (and FPTP for that matter) can encompass a variety of actual voting systems with important differences between them.
I did not mean to imply that FPTP is flawless, it certainly is not - merely that it has some benefits and PR has some drawbacks that often get overlooked in the pursuit of "numerical fairness".
You've got 650 marginal elections where people can only pick one party - and not a coalition through prioritized votes. Compare this to the US where each congressional election is usually between two or three candidates/parties.
This makes forecasts primed for surprises due to margins of error alone, not including other factors.
