Somewhere I saw cap-and-trade referred to as cap-and-auction (can't find the link and don't want to look)... as if the trading were a bad thing (I'm not sure if this has been proposed but I'm knee jerk reacting) .
More or less, cap-and-trade would achieve CO2 reductions efficiently (i.e., at least cost) if permits are allocated as giveaways (say, as a percentage of historical output) or with auctions. Under giveaways the low cost abaters have an incentive to sell their permits to the high abaters. The trading feature acts as the abatement cost information elicitation device. The knock on giveaways are the windfall profits that accrue to polluters.
Under an carbon permit auction firms have incentives to bid in line with their abatement costs and permits would go to the high cost abaters (who would be free to emit CO2). The low cost abaters would lose out on permits and reduce their CO2. The final allocation of permits should be about the same and firm/industry abatement costs are ultimately revealed. The knock on auctions is political -- more resistance from polluters.
Auctions would reduce the additional efficiency generated by trading if permits are passed out but the trading feature should not be replaced. Gains from trade post-auction could be had by unexpected technological innovations and other surprises. Without the trading feature the dynamic efficiency aspect, i.e., firms trying hard to find ways to cut abatement costs, of incentive-based policy could be lost.