On the measurement of cost/speed/bundle.
(And lets try and keep emotions or networks out of this discussion for a minute

It's about how to measure stuff.)
Comparing data costs give a relative value for each service. While this is fine when you buy the service the first time, most users will expect to be able to use their data at the best possible speed.
Say you buy a 1GB bundle per month. In theory, you'd be happy with a download speed of ~3kb/s as this will be enough to use your bundle in the month and that's all you paid for, 1GB of data.
But obviously no-one is going to be happy with this, downloading at 3Kb/s. Although you only have a 1GB bundle, you want to download at as high a speed as possible, whenever you choose to do so. If you can download the whole bundle in minutes, it's your choice.
As an example, say you have two services where a 1GB bundle costs R100 but the one has double the download speed of the other.
Which one provides the better value for money?
This is measured by cost/speed/bundle and such a measurement is needed when you try and normalise different offerings to be compared in the same table.
Thus the author created 2 columns. One is the outright cost and does not take speed into account and here a service like iBurst or Cell-C will do well.
But if you bring speed into it, the higher speed networks like 4Mb/s ADSL will do better.
What's missing form the normalised value is an indicator of up-link speed. Again, all else being equal, how do you differentiate (in a single number) the two services if only the up-link speed is different?
Something like cost/downlink/uplink/bundle now should be calculated.
Obviously one can now bring issues like latency, reliability, coverage, etc. into it as well, but because they're typically not stated, tend not to be used.