Strong jump shifts and raises are similar in that responder describes his hand and usually relinquishes captaincy. Since opener need not describe his hand in these auctions, they are perfect candidates for the use of relays. Danny Kleinman has pointed out that using relays, one jump shift is sufficient to show all strong jump shift hands; further relays can allow many raises to be shown with the remaining jump shifts.
Building on these ideas, we have constructed several sets of raises that start by having responder jump shift and opener rebid the next step. In general, we won't have several rounds of relays (although they could be used) so these methods are more appropriately called "puppet" raises. After each of the four jump shifts (including into notrump) opener generally accepts the puppet, and then responder rebids to show which type of hand he has.
It would be nice if one of the types of hands responder could show is a weak jump shift into the puppet suit, with which he will simply pass the puppet, but the ACBL General Convention Chart (GCC) only allows artificial jump shifts to be "either raises or forces to game," so in GCC events, including weak jump shifts is not permitted.
Which raises or forces to game ought to be included within the puppet structure is a personal choice; I don't claim to know the best set. Indeed, in different partnerships, we have made different choices. Here is an example set.
The responses to 1 are similar, although Jacoby 3 needs to be used. Versions of Jacoby 2NT like Bergen's adapt well to the loss of one step; "normal" Jacoby can be tinkered with easily.
Followup auctions are pretty much the same as if the puppet round had been omitted. For example, after an LROM, if one plays Mathe over normal limit raises, it is still on, new suits are cue-bids or whatever, etc.
Again, the exact choice of which hand types one feels are most important to include in the puppet structure is a personal choice, but using this approach will allow a lot more hands to be shown than can be in standard methods. Furthermore, a few more game forcing hands can be included in the structure; for more information about possible alternatives and followups, see http://www.gg.caltech.edu/~jeff/#science.
Some disadvantages of Puppet Raises are (1) a fair amount of memory burden, (2) a (very) few sequences end a level higher than they would playing direct raises, (3) some of them are vulnerable to 4th chair's preemption, and (4) you can't play weak jump shifts. The memory burden can be severe, depending on how many hand types one wants to fold into the structure. Just adding all strong jump shifts into a Bergen Raise structure after 1 openings, for example, costs very little. The extra space taken by a few sequences (for example, if one plays Jacoby 3) isn't too much of a problem, since most of those sequences were not available in the first place. In designing a set of raises, one must consider preemption; several cases all similar folded into one bid is not good, but combining vastly different hands works well. One ought to discuss sequences after interference. Losing weak jump shifts is a cost others have been willing to pay before. Hopefully, Puppet Raises will increase the return on that cost.
Thanks to Walter Hamilton, Mike Shuster, and Roberto Scaramuzzi for suggestions on how best to use these ideas.