Mark, Good comments.
The reference to "In the fall of 1999" is directly quoted from the book, perhaps you can forward your comments to the publisher, Addison Wesley, to see if your suggested additions can influence the next revision, if they have one.

"In the fall of 1999 IBM and Cisco met to discuss ..."
You might like to add that HP and Adaptec were also there, indeed Adaptec had an almost product that did SCSI over Ethernet at the time.
SCSI over some network protocol wasn't new, in the late 90's I worked at a small company that had been doing SCSI over UDP and IPX since 1994 and there were others out there, too.
However, iSCSI was a thing who's time had come. I wasn't at the Haifa meeting, though I was at the Pittsburgh iSCSI BOF and all subsequent IPS IETF and interim meetings.
Actually IBM's first iSCSI product caused more harm than good, it was based on a very early unratified version of the specification and left some inter-operability problems behind it. Where the community should thank IBM is when IBM decided to give John Hufferd resposibilities in iSCSI and he became a co-ordinator in the IPS workgroup and a very vocal proponent of making iSCSI useful to the market.
Prior to John's involvement the IPS group was dominated by the research teams of IBM, HP, EMC and Lucent. Their vision of iSCSI was that it wouldn't be useful until 10Gb/S Ethernet was around and that it was an ideal stepping stone from which to design the necessary hardware offloads and RDMA protocols.
Those of use trying to promote the idea of iSCSI in small enterprises without the whizzes and bangs, where even 100Mb/s Ethernet was good enough were drowned out and side-lined until the arrival of John Hufferd.
John saw the best way for iSCSI to succeed was to make it an ubiquitous technology. Microsoft went on board with that and produced their initiator. Those two factors are probably more important than anything else, any one with a Windows server or an XP workstation can deploy a simple iSCSI link in seconds.
The IPS work group has gone on to do good work with regards to RDMA and mapping iSCSI over RDMA. However, without the likes of John and those of us who created and then promoted iSCSI in to the small business space, the IPS efforts would still only be a hypothetical exercise for academics. Or at best, iSCSI would be a very small niche market unheard of by the mainstream.

I would differentiate in the sense that gateways allow for LUN masking- you allocate all the LUNs you want available over iSCSI to the gateway which dictates which iSCSI initiators can see which LUNs.
Thad said, this definitely could be considered a switch- I take back my last comment :)

OSG, if it is a switch, isn't that method 2, a switch that allows iSCSI-based servers accessing FC-based storage? Both Qlogic and Brocade are switch vendors, so I would not really call these gateways.