Several news agencies have recently reported on our calculation of bottom US cities by broadband value which gives North Carolina the dubious distinction of having seven out of those ten worst cities.
- http://arstechnica.com/tech-policy/news/2011/03/cable-backed-anti-muni-broadband-bill-advances-in-north-carolina.ars
- http://www.geekosystem.com/broadband-h129/
- http://www.wgnc.net/index.php?option=com_content&view=article&id=9375
Bottom cities by Value
| City | Value |
|---|---|
| 1. Anchorage, AK | $65.50 per mbps |
| 2. Greensboro, NC | $62.03 per mbps |
| 3. Winston Salem, NC | $61.86 per mbps |
| 4. Columbia, SC | $61.35 per mbps |
| 5. Raleigh, NC | $60.98 per mbps |
| 6. Cary, NC | $59.53 per mbps |
| 7. Durham, NC | $59.18 per mbps |
| 8. Wilmington, NC | $56.83 per mbps |
| 9. Charlotte, NC | $52.73 per mbps |
| 10. Shreveport, LA | $52.72 per mbps |
Let’s take a closer look at the data and how we calculate out that value.
Value is simply defined as cost / speed. The North Carolina cities that made the list have poor upload speed, paired with relatively high Internet prices. Upload and download speeds factor equally in our calculations.
Our lists at broadband.com are calculated using last months data (currently February 2011) from NetIndex by Ookla and aggregate Internet speedtest results.
We determine value as the cost of the Internet connection divided by its aggregate speed.
NetIndex gives us upload and download speeds per city per day as well as the number of speedtests.
Our Calculation
- We multiply the number of tests times the value.
- We sum that up for the month.
- We divide that back out by the number of tests in the month. That gives us an average value for both upload and download speed per city for the month. To get a single average speed for a city we just average upload and download speed. (As noted, this is where NC cities start to look bad.)
This data would be more useful if you just considered download speeds, or used two data sets, one for download and one for upload. Mixing upload and download really muddies the waters.
I still don’t understand your methodology. How are you calculating the relative value of upload vs download speeds? Because I live in Harrisburg, NC getting internet from Time Warner out of Charlotte, NC. And I get 10Mbps down and 1Mbps up for $41.95/mo. And in speedtests I get consistently very close to those numbers – remember that it’s practically impossible to hit those numbers exactly. I get somewhere around 960Kbps upload.
If I used that number alone, without factoring in download speed at all, I still am paying about $44.16/Mbps upload which is far less than the $52.73 you list. And that doesn’t factor in the fact that I get about 9.6Mbps download.
Can you provide the actual formula that you use for weighting download and upload speeds in order to determine the aggregate?
Wait. I see where you say you average it. So, for me:
9.6 + .95 = 10.55
10.55 / 2 = 5.275
$41.95 / 5.275 = $7.95
This number still doesn’t come even close to what you’re arriving at.
Additionally, looking at the Ookla source that you provide, I get something very different for N.Carolina: http://www.netindex.com/value/3,62/North-Carolina,-US/
We don’t look at individual speedtest results, but averages for the entire month
for the entire city. Internet speeds and costs in an area can vary greatly. The
key takeaway is that, on average, NC cities tend to pay more for less.
Ookla’s representation of the data looks at their entire dataset (not just
February 2011) and only factors in download speed. We factor upload/download
speed equally because, with the advent of cloud storage like dropbox, upload speed is becoming
more and more critical.
Do you provide your raw data? Alternatively do you provide the methods by which you extracted this data from Ookla?
Ookla actually provides the data at Net Index – Source Data. Their zip file contains a number of csv (and zipped up csv) files that we use in our calculations. There’s a lot of great data in there!