Broadband prices compared

[ic]

ic, I asked you something here. Sorry if you missed it.


http://mybroadband.co.za/vb/showpost.php?p=1901045&postcount=84

Ta. I've downloaded a report from 2007, but it doesn't have the raw data, the details of the 'algorithm', the polls, who conducted the survey etc. Where can I find that?

The report is just that - a report and not raw data - reports usually do not list a bunch of raw data, but rather seek to provide useful information that is based on raw data that has had some information processing applied to it [the raw data].

If anything is unclear, I suggest you first read the posts in those threads I linked to in that Google search, and if you're still unhappy then p.m. rpm and ask for more information or raw data or whatever.

 

[Gatecrasher]

The problem is that there is no way that a multitude of different weighted [human] viewpoints [subjective statistical interpretations] can be represented in a news article.

For MBB's Broadband Report, rpm asks forumites to provide their own personal preferences for such weighted factors - their own personal opinion of which are more important relative to one another, and then rpm presumably takes an average representative weighting of all forumite responses, to apply within the algorithm that is used for MBB's Broadband Report.

Well, I think you have to pitch any comparison like this at the average Joe, or even the better informed average Joe.

It would be very interesting to see the poll results if forumites, for all their broadband requirments, had to choose, without reference to the company, between:

a) a 500MB monhtly cap at 1.0Mbps for R149pm
b) a 500MB monhtly cap at 1.8Mbps for R189pm
c) a 500MB monthly cap at 3.6Mbps for R238pm

I strongly suspect that the current ranking (in the article) would be reversed.

 

[vodacom3g]

Well, I think you have to pitch any comparison like this at the average Joe, or even the better informed average Joe.

It would be very interesting to see the poll results if forumites, for all their broadband requirments, had to choose, without reference to the company, between:

a) a 500MB monhtly cap at 1.0Mbps for R149pm
b) a 500MB monhtly cap at 1.8Mbps for R189pm
c) a 500MB monthly cap at 3.6Mbps for R238pm

I strongly suspect that the current ranking (in the article) would be reversed.

It's important to not decide (or try and predict) the outcome of a measurement before you create the tool that will give the result

Get the tool right, do an accurate measurement and then accept the outcome.

Since pricing (in this discussion) is a given, one can maybe do a poll on how important speed is (on some scale) and also how important cap is, on the same scale.

This can then be plugged into the formula with the actual pricing to get to a weighted 'value' rating.

 

[ic]

I strongly suspect that the current ranking (in the article) would be reversed.

From one of my previous posts:

Considering that the Cost column is in the article, and that very few forumites seem to understand the significance of the "Cost/Mbps/GB" column, and that the ZAR amounts shown in the Cost column are actually higher for Vodacom compared to MTN, I tend to think that people reading the article are more likely to go for MTN instead of Vodacom, which would suggest that the author must be biased in MTN's favour...but whatever as they say.

Also, in the article, under the cost column, iBurst is shown to be the least expensive for 500MB as well as for 1GB, followed by "Telkom DSL 384" for 2GB, and finally iBurst again for 3GB.

IMO the article is hardly weighted in Vodacom's favour - in fact the inclusion of Vodacom's HSDPA 3.6 with its ZAR49/month VAS fee, acts more as a penalty against Vodacom in the article since it pushes Vodacom's pricing up, which would not have been the case if Vodacom's HSDPA 1.8 had been used...

 

[Gatecrasher]

Since pricing (in this discussion) is a given, one can maybe do a poll on how important speed is (on some scale) and also how important cap is, on the same scale.

But the two are interrelated. The smaller the monthly cap, the less important the speed. The higher the monthly cap the more important the speed.

 

[Gatecrasher]

IMO the article is hardly weighted in Vodacom's favour - in fact the inclusion of Vodacom's HSDPA 3.6 with its ZAR49/month VAS fee, acts more as a penalty against Vodacom in the article since it pushes Vodacom's pricing up, which would not have been the case if Vodacom's HSDPA 1.8 had been used...

Yes, those may well be the facts, but that is certainly not the impression the article gives, as it ranks Vodacom HSDPA top in each of the first three categories.

 

[milomak]

On the 3G issue - If every Vodacom 3G base station is rated at 3.6Mbps (the minimum for Vodacom 3G towers from what I understand), is it not fair to rate the MTN service at its lowest speed especially given that the provider has given no information as to the speed of the majority of its towers? It seems consistent to me. And I am with MTN before someone accuses me of bias. It seems to me in this particular speed issue, MTN has not been treated unfairly.

As for rated speed vs actual speed, surely a site like myadsl is not expected to prove the one network is closer to achieving rated speed as opposed to the other. As an aside, did MTN and Vodacom (and iBurst) fall under the same judgement where Telkom couldn't advertise as 4Mbps? If so, then there is ground to dispute my original statement in this paragraph. But does that mean we should assume say 60% of stated speed?

 

[rpm]

I think I must just clearly state that we did not give preference to any provider. We have been using Cost/Mbps/GB for a while already, and until we get a better method to calculate value for money I cant see a better way to compare services. With these articles I generally sit with a spreadsheet and a formula or two and the results are then reported on – no fine-tuning of results.

The weighed model suggested by Gatecrasher may indeed provide a better/more balanced view, but we will have to ensure that it is indeed an easily defendable model without obvious errors. I also welcome suggestions of this nature – it will indeed be great to have a very objective formula which clearly shows value for money which we can then use for local and international benchmarking.

 

[milomak]

Benefit = average data received. If you receive 1 gb every day of a 30 day month (slow uncapped connection), the benefit is 15gb (0.5 + 1.5 + 2.5 +...+ 27.5 + 28.5 + 29.5)/30.

Benefit = (1 - cap /(2*max))*cap, where cap = max = 30, thus
Benefit = (1-30/60)*30 = 15gb

I can't help but feel real thick, but how you get to 15GB benefit is not immediately obvious to me. Nor the 29.5GB in your later example.

 

[ic]

Yes, those may well be the facts, but that is certainly not the impression the article gives, as it ranks Vodacom HSDPA top in each of the first three categories.

If one looks and understand the meaning of the last column, otherwise consumers are more likely to concentrate on the Cost column where Vodacom fares badly compared to the other services.

The extra R49 a month fee was not included in the original article was it. It was in fact left out originally.

Read what rpm has already posted in this thread - it was a mistake and was quickly corrected, what more do you want.

PS: you are starting to sound very much like a certain someone that had to be banned for consistently bad behaviour and totally untrue insinuations.

 

[Gatecrasher]

I can't help but feel real thick, but how you get to 15GB benefit is not immediately obvious to me. Nor the 29.5GB in your later example.

Let's restate it: Cost = monthly expense. Benefit = average data received.

So a new month begins. On my slower uncapped connection, I pull 1 Gb during this day. At midnight I had 0Gb downloaded. By Noon I had 0.5Gb downloaded, By the next midnight I had 1 Gb downloaded. So on average on the 1st day I had the benefit of 0.5Gb. The next day, another Gb is downloaded, my benefit rises to 1Gb. I continue downloading 1Gb per day. By the end of 30 days, I have downloaded 30Gb, and my benefit has risen to 15Gb.

On a graph, see y=x, a line starting at the origin, (0,0) at a fixed gradient, ending at (30,30). The benefit is the average area under the curve, or more formally 1/30th of the intergal of the curve y=x over the interval x=0 to 30. This equals 1/2 * (30^2)/30 = 15.

In the second case, my super fast capped connection pulls 30GB on day one. At midnight I had 0Gb downloaded. By Noon I had 15Gb downloaded, By the next midnight I had 30 Gb downloaded and I'm capped. So on average on the 1st day I had the benefit of 15Gb. Since I'm capped I can't download any more, so for each subsequent day of the month I have the benefit of the 30gb I downloaded on day one. The benefit over 30 days = 29.5Gb.

On a graph, see y=30x, a line starting at the origin, (0,0) at a fixed gradient, ending at (1,30). For x>1, y=30, a flat line. The benefit is the average area under the curve, or more formally 1/30th of the intergal of the curve y=30x over the interval x=0 to 1 and the integral of y=30 over the interval x=1 to 30. This equals [30/2 * (1^2) + 30 * 30 - 30 * 1 ]/30 = 29.5GB

I hope this explains it better, rather than making it even more confusing...

The best way to look at it is to imagine data recieved on the y axis of a graph, and days of the month on the x axis. Benefit is the average amount of data received so far, averaged over the course of the month.

 

[Gatecrasher]

If one looks and understand the meaning of the last column, otherwise consumers are more likely to concentrate on the Cost column where Vodacom fares badly compared to the other services.

Well, this consumer here would take far more cognizance of the order in which they are ranked and value used to derive the "value-for-money" ranking.

So we'll just have to agree to disagree on this point.

 

[vodacom3g]

But the two are interrelated. The smaller the monthly cap, the less important the speed. The higher the monthly cap the more important the speed.

Then use a weight for each.

The point is that people will want to plug their personal preferences in, so if both are important you give them both a weighting of 1.

 

[milomak]

Let's restate it: Cost = monthly expense. Benefit = average data received.

So a new month begins. On my slower uncapped connection, I pull 1 Gb during this day. At midnight I had 0Gb downloaded. By Noon I had 0.5Gb downloaded, By the next midnight I had 1 Gb downloaded. So on average on the 1st day I had the benefit of 0.5Gb.

I am really going to be an irritant I can sense it. So I will try to get a grip of your concept step by step. If I download 1GB in a day (24 hour time period), how can the average/day be anything less than 1?

 

[Gatecrasher]

I am really going to be an irritant I can sense it. So I will try to get a grip of your concept step by step. If I download 1GB in a day (24 hour time period), how can the average/day be anything less than 1?

Because we are not measuring the average downloaded each day. We are measuring the average of what has already been downloaded. At the start of the month I have downloaded a total of 0GB. At the end of the month I have downloaded a total of 30GB. Over the course of the month, the average data downloaded is 15GB.

Similarly, on Day 1: Start of day, I have downloaded 0GB. At the end of the day I have downloaded 1GB. Over the course of the day, the average data downloaded is 0.5GB.