In reply to beardy mike:
> In short I would say a magic plate is not the right tool whenever:
> Anchors are questionable at all.
Agreed! But you have to have a good strategy for keeping the load off them...
> There is a stronger than normal chance of your second falling.
I think it depends a lot on the nature of the climb and the intentions of the climbers. There is little question that having to lower a climber hanging from a guide plate is a bit of a pain. So I'd say that it isn't falling that is an issue but lowering. If the second is going to fall, hang there, and continue on, then no problem. But if the second will have to be lowered, either because the second wants manage the moves free or because a fall puts the second in a position they can't recover from without being lowered, then the guide plate is a poor option.
> When your second might need to retreat from a position to either rest on a ledge, or to work out a sequence of moves - a traverse would be an example of that.
Amen to that! Personally, I find many of the issues involved with the plates in guide mode come from the way belayers use them. I really dislike being continually pulled and not being able to step down when I want to without weighting the rope. These are just stylistic annoyances until the rope either runs diagonally, horizontally, or over an overhang, in which case the constant rope tension of this type of belaying can very well pull the second off and leave them hanging, either on a blank section below the traversing route or out in space under the overhang. At that point, any of the possible time savings obtained from the use of the guide plate are more than squandered, as the second either has to be lowered (hopefully without being dropped) or else has to prusik up the rope until they can get re-established.
Many of these issues can be avoided by attentive belaying, with the belayer testing the second's progress by periodically lifting the second's rope without instantly and automatically pulling the slack through the plate and so trapping all available slack. The problem with this is that the plates are used to facilitate what can only be called inattentive belaying, in which the belayer is multitasking and not fully attending to the climber. So the tendency is, when the belayer does check on the climber, to rapidly pull up whatever slack has accumulated until the rope goes tight.
I think the plates are most useful on long routes where the time-saving forms of belay inattention are actually going to be fruitful, and the second is trying to move as quickly as is reasonable and doesn't at all mind a little help from the rope. As the climbs get shorter and the need for speed evaporates, I see less and less value and more and more drawbacks to guide plate belaying, although I do it sometimes anyway because it does tend to result less back strain in some situations.
In addition to guide plates not locking if the loaded rope runs horizontally from the plate, the belayer has to be careful about putting the plate in a rock configuration that could interfere with the plate, either by preventing the rotation that enables the locking effect or blocking the rotation required to unlock the plate. The main problem is when the plate is in a corner (this could be very shallow) and one wall of the corner can interfere with the operation.
These concerns are not hypothetical; I know a guide whose plate jammed in a one-inch corner when the client fell off an overhang and ended up dangling. The client had to be lowered, but the plate wouldn't budge. The guide has to set up a 3:1 hoist to unweight the plate and change over to (in this case) a redirected hip belay. Needless to say, this sucked up a lot of time, but it happened on a three-pitch cragging route so that wasn't an issue.
If there is going to be a horizontal load on the belay or if the anchor is in a corner or slot or other configuration that could restrict plate movement, then either a more conventional belay or a Munter hitch on the anchor seems to me like the way to go.