Search This Blog

Wednesday, April 16, 2014

A keyhole view of Item 21

EducationSuperHighway released an analysis based on the data from Item 21 Attachments.  I thought it was a quick turnaround of data from their Item 21 Portal, but no, it's from 2013-2014 Item 21 Attachments.  Apparently, they got Item 21 Attachments from 1,044 school districts, representing over $350 million in funding requests.  That's about 3% of applicants, and about 7% of funding requested.

That's more Item 21 data than anyone else has.  (Well, the New York City Board of Ed requested $605 million in funding, so technically they have access to a higher dollar amount of Item 21 attachments.)  But it's a pretty small sample, and I have no idea how representative it is, so I can't say whether the data is any good.

The data is being used to support a point of view.  Let's take a look at the "insights" and supporting data in the executive summary:
  1. We face an urgent challenge to ensure that our students do not fall further behind.    
    1. The wealthiest districts are more likely to have met ConnectED goals, the poorest less likely.
    2. Rural districts are less likely to have high-speed fiber.
    3. It will take 7 years to reach today's goals.
    4. In 7 years, our schools will need 10 times the bandwidth.
  2. Schools are not meeting the ConnectED goals because high-speed broadband is not affordable.  
    1. Schools that are meeting the goals pay on average 1/3 the price for broadband vs. those that are not meeting the goals. 
    2. Schools that are meeting the current ConnectED goals also have Internet access budgets that are on average 450% larger than those that do not.
  3. Schools that are able to afford high-speed broadband provide an actionable roadmap to enable every school to meet the ConnectED goals.
    1. Districts with fiber connections have approximately nine times more bandwidth and 75% lower cost per Mbps compared to districts without fiber.
    2. At higher speeds, which might be accomplished by aggregating purchases across multiple schools and districts, schools can reduce their costs to as little as $2/Mbps.
    3. Schools with access to competitive options pay 2 - 3 times less for their WAN connections compared to schools that are only served by incumbent telephone and cable companies.
    4. schools that have the option to take the initiative to lease fiber, self-provision a fiber network, or access an existing city network, pay the lowest prices for high-speed broadband.
  4. 96% of schools could meet today’s Internet access and WAN standards , if the FCC focused the E-rate program on broadband, but meeting the five-year ConnectED goals will likely require a combination of lower prices and more resources.
    1. Re-investing the $1.1 billion per year of E-rate funds that are spent on non-broadband services (telephony, mobile, web hosting, and email) would provide enough funding to enable 96% of schools to have a gigabit WAN connection and 100 kbps/student of Internet access.
    2. without improving the affordability of broadband, the $1.1 billion per year increase in support for broadband will still leave 80% of schools with too little bandwidth in five years
My thoughts on their insights:
  1. So vague, it's indisputable.  "Further behind" whom?  Behind in what?  The evidence presented indicates a disparity in access based on location and income level, and that we won't reach artificial goals quickly.  There is no evidence on how the disparity or failure to meet targets will cause our students to fall further behind.  Also, that "7 years to reach today's goal" figure is extrapolated from the fact that 28% of schools had 100 Kbps/student in the spring of 2013, and 37% did by the fall of 2013.  That increase of 9% (over six months) was assumed to be the annual increase for the next 7 years.
  2. Probably true in a lot of cases, but I know schools with 90% discounts, for whom money is no object, that don't meet the ConnectED goals because the network admin is competent enough to know that the schools don't need that much bandwidth.  The factoids don't support the hypothesis very well.  The first factoid results from the fact that cost/Mbps always drops drastically as the Mbps increase.  (If a 100 Mbps link cost a school $1,000/month, a Gbps link is probably going to cost less than $3,000/month.)  And the second factoid just says that schools that spend more on Internet have higher speeds.
  3. Of course schools that have fiber have more bandwidth and cheaper prices.  Factoids 1, 3 and 4 basically just say, "Applicants that have more options generally pay less."  Factoid 2 claims that aggregating purchases can reduce costs.  I agree that if 10 districts share a 1 Gbps Internet port, that will be cheaper than each of them having a 100 Mbps port.  But only the Internet port costs would go down; circuit costs would probably go up, as all the districts would have to connect to a single hub site, rather than connecting to the nearest POP.  The full report has a very interesting graph showing that applicants with over 100 WAN nodes have a per-Mbps cost that's a third of what other applicants pay.  But there are only 2 applicants in that category, which is such a small sample that it seems irresponsible to make that a category.  Hmm, interesting, the graph shows only 180 applicants with WANs.  Only 17% of applicants in their sample have WANs?
  4. I think their small sample skewed these results.  Here is what those services cost for FY2011, according to USAC: Telephony=$428 million ($452 million if you include VoIP). Mobile=$176 million. Web hosting=$27 million. Email=$10 million.  The total saved would be $669 million.  ESH's estimate, based on their sample of 3% of 2013 FRNs, is 65% higher.  The biggest discrepancy: their web hosting cost is 4 times greater than USAC's.  The fact that only 17% of their applicants had WANs would also drive down the proportion of spending on broadband.
The report is certainly worth a read, and there is some good information in there, and they do make an effort to provide clear and objective information, but some of the information is spun pretty hard.  The Executive Summary in particular is almost misleading.

1 comment:

  1. I just thought of a good analogy for why I don't like this statement: "Districts with fiber connections have approximately nine times more bandwidth and 75% lower cost per Mbps compared to districts without fiber."

    It's like saying: "Commuters who take the subway pay 75% less and take 75% less time than commuters who take the train to work." Well, that's true, but it implies a causality that isn't there.

    Subways are cheaper where they're available. Fiber is cheaper where it's available. That doesn't mean that the subway system is the most efficient form of transportation for every situation. Fiber is often the most efficient form of data transport, but it is not always.

    The reason the subway is cheaper largely is due to the shorter distance traveled. Fiber is more likely to be deployed where distances between schools is smaller.

    Fiber transport is correlated to lower cost and higher speed, but correlation does not imply causation.