Here are some thoughts on the latest points discussed.
First, I agree that the method used for handling convoys should be as simple as possible.  Or to put it another way: the basic house rule for handling convoys should be as simple as possible, and individual players can then introduce additional details if they want as supplementary house rule modifications of the basic house rule.  I also agree that “negative IPCs” sound problematic and are a hard thing to visualize.  I’d prefer to stick to the concept of a convoy being worth 3 IPCs when it’s at full strength, 2 or 1 IPCs when it’s suffered partial casualties, and 0 when it’s been completely destroyed.  Its value should not go above 3 or below 0.
It’s a valid point that the attacker needs an incentive to attack a convoy, but I’m not sure that the correct incentive would be a model in which the attacker can sink a whole convoy all at once, but in which the convoy automatically reappears on the next turn.  To me as an attacker, the automatic reappearance of the convoy would discourage me from bothering to attack it because, in effect, this would mean that I’d have to keep sinking the same convoy over and over again to keep its IPCs out of my opponent’s hands – a task which would tie down my valuable naval units, which I could perhaps use more profitably elsewhere for other purposes.  A further problem is that this model would not be realistic.  In real life, cargo ships that have been sunk by enemy attack don’t magically float back up to the surface and continue towards their destination; instead, replacement ships have to be built and launched and loaded and sent on their way across the seas, all of which takes time and money.  And to draw a parallel with land warfare (both in real life and on the A&A game map): when a territory gets conquered by an invader, the territory doesn’t automatically liberate itself after a set amount of time and the invader therefore doesn’t have to keep conquering it over and over again in order to keep it out of enemy hands.
To me, the biggest incentive to attack a convoy would actually be a model in which the convoy doesn’t automatically regenerate itself, and in which convoy losses therefore cause serious harm to my opponent.  In such a model, I wouldn’t be worried about whether I could or couldn’t knock out the whole convoy at once; any damage I inflicted, whether it be the loss of 1 or 2 or 3 IPCs by the convoy, would be worthwhile.
That being said, I think it would be a good idea if convoy losses could be replaced under certain circumstances, provided that this was handled in a way that’s realistic and that doesn’t remove the incentive for an attacker to attack a convoy.  I think the way to handle this would be as follows.
Remember that a convoy is a group of cargo ships: freighters, tankers and other merchantmen.  They’re represented in the house rule by a marker, but conceptually they still consist of ships; in a sense, the ships are really represented by the grey plastic chips under the convoy tokens.  (Perhaps more accurately, these chips represent columns of ships within a convoy, each column consisting of several vessels).  Regular A&A ships, such as naval transports, are units that can be purchased, and the cargo ships in the convoys could be thought of in the same way.  Each player starts the game with a specified number of convoys, which don’t have to be purchased – but if a player’s convoys starts taking casualties, he would be able to replace the lost cargo ships by buying replacement ones (in effect, by buying replacement grey poker chips).
There are, however, a few things that we’d need to figure out about this concept because it represents the process of building and launching replacement ships, loading them with cargo, and having them travel along their cargo route (which takes time and money in the real world).  I think there are basically three possible options for doing so:
- 
A player acquires replacement cargo ships (poker chips) in the same way that other naval units are purchased: each poker chip costs $x (we’d need to figure out a fair price), it’s built at a coastal Industrial Complex, and it has to sail from there to its designated convoy location before it’s considered fully operational.  I think this would take too much time, and it would introduce too many confusing complications by having poker chips moving across the map.
 
- 
A player buys replacement chips for $xxx – a much higher price than the cost of $x in option 1 – and gets to add the chips directly to the desired convoys, without any transit time.  This could work (though we’d once again have to figure out a fair price), but it’s not my preferred option.
 
- 
A player buys replacement chips for $xx – let’s call these mid-priced cargo vessels – and adds the chips directly to the desired convoys…but he has to wait one full turn before doing so.  In effect, he skips a turn (meaning that he has to live with an under-sized convoy for a full turn) to represent symbolically the time it takes for the newly-built cargo ships to join the operational convoys.  This is my favourite option: it imposes a time delay in the replacement process (which is realistic), but the delay is of standard length (one turn)  and it does not involve any chips moving across the game map on their own.  As always, we’d have to figure out a fair price.
 
And on the issue of fair pricing: I think that the calculation of the replacement cost for convoy poker chips is probably the best way to find the right balance between the attacker and the defender of a convoy, while keeping the combat system itself very simple.  If we determine that the attacking and defense mechanisms for convoys provide too much incentive for the attacker, then the solution would be to lower the cost of the replacement chips.  If on the other hand we determine that the attacking and defense mechanisms for convoys provide too little incentive for the attacker, then the solution would be to increase the cost of the replacement chips. Adjusting this variable, rather than adjusting the attacking and defense mechanisms themselves, would allow us to come up with a convoy combat model that is simple and realistic and which doesn’t take up too much game time.
Now to switch to a different topic: convoy “conditions” or requirements, the topic whic is addressed in the right-hand column of my draft chart.  I’m thinking more and more that we do need to define operating conditions (at least in simple terms) for each convoy on the map.  The reason is that, conceptually, a convoy needs three things to operate: a port of departure (where the transported goods come from), a port of arrival (where the transported goods are going), and of course the actual ships in the convoy (the topic discussed in the paragraphs above).  Without all three components, convoys can’t operate in the real world – so the game should reflect this.
Fortunately, modeling this in A&A both simply and realistically would be pretty straightforward.  For each convoy, we’d simply need to specify one or more ports of departure and one or more ports of destination, and require that the convoy owner (or one of his coalition partners) control one or more of the specified territories at each end of the convoy route in order to collect his bonus IPCs.  The departures and arrivals could be tied to a single specific territory (for example Java in Global 1940, which would be a single port of departure) or to a multi-territory area (for example “United Kingdom or Scotland”, which would be an either/or port of arrival in Global 1940, and which we could designate collectively as “British Isles”).
As an added refinement: if we assume that all convoys have a standard maximum value of 3 IPCs, we could retitle the second column in my draft chart as “Convoy Name”.  This would be kind of cool, and also a useful memory aid in game play.  In WWII, convoy routes sometimes had names and individual convoys usually had code numbers.  These designations sometimes referred to ports of departure or ports of arrival – for instance the Gibraltar run, the Murmansk run, the HX convoys that departed from Halifax, and the SC convoys that departed from Sidney, Cape Breton.  So we could give each convoy route its own name, as well as specified operating conditions.
Note, by the way, that an opponent would actually have multiple ways to shut down a convoy: from the sea, by sinking it, or from the land, by conquering either its port of departure or its port of arrival.  A land power like Germany, therefore, might be more inclined to try to shut down a convoy route “from the land” (by territorial conquest), while a naval power like Japan might be more inclined to shut down a convoy route “from the sea” (by using parts of its fleet).  As a bonus, this would nicely illustrate some of the relationships that exist between land warfare and naval warfare.