Some ideas make so much sense to me that it is a constant frustration that they never seem to catch on.

My favorite beautiful-but-largely-rejected idea is something called "final offer arbitration." When labor-management negotiations reach an impasse, each side makes a final, detailed proposal that it offers as a settlement. An independent, jointly selected panel of arbiters then chooses one of the proposals as the contract.

The arbiters may not pick and choose among the features of the final offers. One or the other of the proposals, taken intact, will become the contract. The beauty is that it encourages each side to drop its more outlandish demands, lest the other's proposal be adopted; in many cases, the final offers are virtually indistinguishable. The frustration is that, for reasons I cannot understand, hardly anyone wants to use it.

Here's another brilliant idea that probably will never be used: "cake-sharing" disarmament. The basis is the solution to the problem of two children sharing a piece of cake: one divides the cake, the other gets first choice. The idea as a strategy for nuclear disarmament has been kicking around at least since 1963, but an Edinburgh University engineering professor named Stephen H. Salter has refined it to the level of mathematical elegance.

Here's how it would work, as outlined in a recent edition of a Los Angeles newsletter called Leading Edge: The United States and the Soviet Union each makes a list of its own weapons and assigns each weapon a proportionate value based on its usefulness in the eyes of its owner. The total values for each superpower would add up to 100.

Then each side chooses a tiny, mutually agreed percentage of the other's weapons for dismantling.

The proposal neatly sidesteps the problem of weapons comparison, brought on by the fact that the two sides will tend to assign different threat/security values to their various weapons.

Under the "cake-sharing" scheme, the differences in perceived value cancel each other out, since each side would demand elimination of the weapons it found most threatening. There would be no point in either side's deliberately distorting the military value of any particular weapon, and each side would feel it had gained an advantage from each successive reduction.

Leading Edge describes an example provided by Salter: "A mobile, forward- based quick-launch unhardened missile with accurate terminal-guidance can very easily be mistaken for a first-strike deterrent destroyer without providing much protection for its land of origin. On the other hand, a less-accurate submarine-based weapon, which is safe from detection and could be used at leisure for retaliation as a second-strike weapon, is a splendid protection for its owner without appearing too threatening to the target."

"Cake-sharing" also avoids many of the anxieties related to the difficulty of verification. To begin with, the disarmament steps are so small that the overall balance would remain essentially unchanged, and if either side suspected cheating, further reductions could be halted until the matter was settled.

In addition, the very slowness of the process would offer more time for verification while allowing the money saved on weapons to be spent, say, on surveillance satellites or on "intrusive" inspection visits by neutral countries.

"Cake-sharing" seems in all, as sensible an approach to nuclear disarmament as final-offer arbitration is to collective bargaining. And it is just as likely that both sides will reject it out of hand.