I think the point Paul_S was probably trying to make is that although there is a difference in how one connects to the Internet (ie. access mechanism being either 'mobile' or 'fixed') the underlying bandwidth requirement should be viewed as the same. I could be wrong, but that's how I read his point.
Some areas are not part of Telkom's 'golden' plan for DSL rollout until it suits them (if at all) while other areas have no decent wireless signal coverage. Thus you find a lot of people using wireless (be it iBurst / MyWireless / HSDPA if the signal is good enough, or 3G or even plain GPRS if all else fails) simply to have any connection at all and the fact that it is 'mobile' has absolutely no relevance on the choice of package.
That said, at the end of the day - what is the real difference in cost for the underlying bandwidth used between fixed / fixed-wireless / mobile-wireless ... is there a true justifiable explanation at all?
If I use 3Gb of bandwidth per month - in our market (with all it's current flaws) should it make a difference and cost me more / less than if I use 3Gb via a fixed-wireless or true mobile-wireless compared to my ADSL service? Sure, access charges should cover the extra expense for going true mobile-wireless, but surely the bandwidth cost should likewise be similar in price per Gb / Mb regardless of platform used? There will already be issues of speed / signal strength vs contention vs attenuation to differectiate the various connection packages, but 1mb of bandwidth remains 1mb of bandwidth - you don't get any other types of 'bandwidth' do you?
Compare it to fuel. Regardless of whether you drive a sportscar or an entry level 125cc motorcycle, you need fuel. In this case petrol = mobile and diesel = adsl in terms of price comparison. The bandwidth here = the crude oil that is used to provide both the petrol and the diesel. Without the oil, you wouldn't have fuel. Without bandwidth, you wouldn't have internet service. You'd still have an access mechanism to the internet (same like you'd still have the car / motorcycle) but you wouldn't go anywhere without bandwidth (fuel).
It seems the industry should relook at how bandwidth is provided and adjust it for our 'unique' market. Providers like Telkom are always telling us that we cannot have equal level of services and prices (as overseas markets) because our market factors differ substantially from those overseas markets - so why then do they make profits hand-over-fist providing bandwidth and services that does not effectively take our unique factors into account? We all know international access is costly. But that doesn't justify why local access (and especially local bandwidth allowance and local speed / quality) cannot be on par with overseas markets or at least much closer to it than we currently have). If prices for local is dropped to affordable levels for the masses (and I'm not talking R99 for 150mb, I'm talking about R99 (fullprice) for 1Gb unshaped plus R50 premium for using a fixed-wireless access mechanism for a total of R149 / plus another R50 for true mobile-wireless for a total of R199) then a lot more consumers can subscribe for internet services and grow the market until such time as it becomes feasible to compare ourselves directly with overseas markets and structure services and packages along similar lines.
As South Africans, we're used to getting shafted by service providers, but let's not forget the extent to which we get overcharged for sub-standard services. Bringing service levels and allowances under control and within accepted levels does not mean that costs have to remain overpriced - it merely means ensuring more consumers take up your service (by lowering overall costs and ensuring a quality service). In fact, more profits will be made this way by having a medium- long-term approach instead of the short-term method used today.
Regarding the article, it worries me that Sentech wants to invest a further R3Bn to leverage the R10mn already invested in their first-generation 3G network when clearly they are unable to retain more than 4k/5k clients as it is due to their lack of understanding of the product provision vs market needs. Admit their first attempt as a failure and rather cut the R10mn loss and start fresh using the latest technology with the R3Bn tax-funded investment and ensure that the platform address not only current issues, but will also still be usable in the future when the rest of the world moves to 4G / 5G wireless standards.
Of course, this does mean that oversight be enforced to ensure the R3Bn doesn't get wasted or slipped into oily palms / backpockets. Another smart choice would be to outsource the deployment of the new network and client services division to companies that have extensive experience in those fields. Sentech's track record proves that internally they're not up for the task. It's one thing providing services in a protected environment (ask Telkom) but a completely different story when there's already competition in the marketplace.