Let's first look at what might have happened if the 2007 Assembly election had been run with five seats per constituency instead of six. Since the quota - the number of votes you need to get elected - is defined by dividing the total valid votes by one more than the number of seats, that means that the quota in each seat changes from just under 14.3% to just under 16.7%. (Though because of the effect of transfers, the real electoral threshold is considerably lower - all candidates who managed at least 9% of their constituency vote in 2007 got elected.)

I went into this assuming that a cut in the number of seats would hit small parties worst. As I ground through the numbers, it became apparent that that is not necessarily the case, for two reasons. The first is what one might call the 'tallest poppy' effect. In a constituency where one party is dominant over the others, its last seat may well be vulnerable, particularly if was rather narrowly gained - if you are defending two seats, you need transfers to take you to 33.3% instead of 28.6%; if you are defending three seats, you need 50% instead of 42.9%; if you are defending four seats, you need 66.7% instead of 57.1%; if, like Sinn Féin in West Belfast, you are defending five seats, you need 83.3% instead of 71.4%. On that basis it's fairly clear that five-seat constituencies would indeed have penalised Sinn Féin in West Belfast, but also the DUP in East Belfast, East Londonderry, and Strangford, and the SDLP in South Belfast and Foyle.

But the more important effect is what happens to the mathematical positioning of the communal divide. In Fermanagh and South Tyrone, for instance, the total Unionist vote in 2007 was 46% - 3.22 quotas, if there are six seats - and the total Nationalist vote was 53% - 3.71 quotas. Not surprisingly, Nationalists won three seats and so did Unionists. But had there been only five seats at play, the Unionist vote would have been 2.66 quotas and the Nationalist vote 3.18 - Unionist parties as a whole (and the DUP in particular) would have lost a seat. I would therefore add Fermanagh-South Tyrone, West Tyrone and, narrowly, Newry and Armagh to the DUP's losses (where the the DUP were the smaller Unionist party) and on the same basis would call East Antrim, North Down, Upper Bann and probably South Down as losses for the UUP. Similarly on the Nationalist side, North Belfast, North Antrim, and South Antrim would have seen the SDLP lose their seats.

There are two remaining constituencies. The situation in Lagan Valley is the only one drastically affected for this debate by the new boundaries, used for this year's Westminster election but not for the 2007 Assembly; on the 2007 votes as cast, Alliance would probably have lost their seat, but on the new boundaries, Alliance would certainly hang on at the expense of Sinn Féin. And in Mid Ulster, where both in 2007 and in 2010 the total Unionist vote was 33% and the total Nationalist vote 66% - just under two and four quotas respectively for a five-seat constituency - it is really impossible to call which side would lose the sixth seat; it is easier to state that the loser would either be Sinn Féin, as tallest poppies, or the UUP as junior Unionist party.

I've also looked at the 2010 Westminster election results, where there is considerable blurring due to obvious tactical voting and withdrawal of particular parties' candidates which presumably will not be repeated in the Assembly elections next year. The only seat where the 2010 votes give a clearly different outcome is Newry and Armagh, where a swing towards the Unionists (comparing 2010 votes with 2007) would mean the SDLP rather than the DUP losing out in a five-seat contest.

So, if the 2007 Assembly election had been run on five-seat rather than six-seat constituencies, I make the difference as follows:

DUP | SF | UUP | SDLP | Alliance | Oth | |

Seats really won in 2007 | 36 | 28 | 18 | 16 | 7 | 3 |

Notional 5-seat constituency seats | 30 or 31 | 25/26/27 | 13 or 14 | 10 or 11 | 6 or 7 | 3 |

Change | -6 | -1/-2/-3 | -4 or -5 | -5 or -6 | 0 or -1 | 0 |

% seats really won in 2007 | 33.3% | 25.9% | 16.7% | 14.8% | 6.5% | 2.8% |

% seats notionally won | 33.3% or 34.4% | 27.8%/28.9%/30.0% | 14.4% or 15.6% | 11.1% or 12.2% | 6.7% or 7.8% | 3.3% |

difference | 0 or +1.1% | +1.9%/+3.0%/+4.1% | -2.2% or -1.1% | -3.7% or -2.6% | +0.2% or +1.3% | +0.6% |

Of course, all of this is more than a little theoretical. The votes cast in next year's Assembly election, never mind the first election under the proposed new arrangements in 2015, will be different from those cast in 2007 or 2010. And in any case we will be in a totally different ball game with respect to the constituency boundaries; the shift from the current 18 Westminster seats to 15 is pretty certain, but any shift from six-seaters to five-seaters from the Assembly is highly theoretical, and would require at least the DUP and SF to agree on it and probably at least grudging acquiescence from the other parties. But I think the basic lines set out above - that such changes will affect either the tallest poppies, or the more vulnerable party in the smaller bloc - are sound, and so it seems to me a fair conclusion that Peter Robinson's proposals will do the DUP no relative harm, will be bad news for the UUP and SDLP, and ironically may actually benefit Sinn Féin most.