I totally agree that combining 2.5G with WLAN hotspots does not make 3G redundant -- despite the fact that Columbitech, a wireless VPN company where I am CEO, was once called a "3G killer" by the WSJ, since we offer a solution for seamless roaming between WLAN and GPRS with a VPN connection.
But Wireless LAN combined with 2.5G, while not making 3G redundant, is a major problem for the 3G carriers.
Why were 3G networks planned in the first place? For data services -- that's what mobile carriers are expecting will boost the ARPU (average revenue per user) and fuel new revenue growth. Sure, the 3G networks can be used for voice, but voice alone will not pay back the massive investment in networks and frequency licenses that are required to deliver the service.
Carriers are counting on consumers picking up data services, from sports video clips to mobile email, and count on consumers wanting those services wherever they are -- requiring full, national ground coverage. This is where I think wireless carriers go wrong, and where WLAN becomes such a headache for them.
I think consumers (and business users for that matter) only have the time and attention to enjoy a high bandwidth wireless application when they are not moving. That means that it is really enough to have high bandwidth in places where people are likely to sit down and be in the mood for a high bandwidth wireless experience: in train stations, cafes, malls, airports, busy shopping streets, and other high density places.
When people are moving about they can pay much less attention, and for the most part a low-bandwidth wireless experience will be enough: for example, getting notifications for emails and other messaging services. Walking and talking is easy -- but try walking and watching a videoclip or playing an interactive game!
All this boils down to bad news for 3G carriers: WLAN operators can install their data networks in relatively few key locations, and still compete for the majority of the high bandwidth wireless data business.
In fact, all mobile operators are thinking about how to handle WLAN, even those that won and paid for a 3G license: "Should we do WLAN? But it cannibalizes our 3G business! Hey, if we don't cannibalize it someone else will! OK, so should we package WLAN with GPRS?..."
In my view, the mobile operator is the only player in the ecosystem that has a chance of building a profitable WLAN network operation (aka WISP): They have the customer relationship (if you can call sending bills a relationship) and can do targeted marketing to heavy users; they already have billing and infrastructure, etc. I find it hard to see how a greenfield operation could work when the target niche (work a lot on the move + can't wait to be connected + requires high speed) is still pretty small.
True, there are issues with WLAN that are still to be solved. You mention the lack of roaming agreements to tie together the archipelago or WLAN hotspot islands, for which there are actually a number of promising initiatives, from Boingo in the U.S. to Excilan in Europe.
Security is also a major issue, but one that won't stop corporate users from accessing any WLAN network they like. The wireless-connected road warrior that roams networks like WLAN, CDPD, GPRS (and 3G for that matter!) should be secured end-to-end with a wireless VPN solution. Then you are secured and independent of the network.
So a seamless combination of WLAN and 2.5G will not bring good voice services, and will not replace 3G carriers, but it will create a strong wireless data service, and I am certain that it will take big bites out of the potential 3G business.