At 7:18 AM 24/1/96, Jeff Cochrane wrote:
>The good Dr. Lisse politely points out:
>> By the way, there is currently no UN Internet effort in Africa
>> that is even remotely concerned about sustainability...
>My sense from discussions in Malawi late last year was that the UN
>staff there were genuinely concerned about the problem of
>sustainability. Whether their effort in Malawi and elsewhere proves
>to be sustainable remains to be seen. But reports reaching me
>suggest they are now at a stage where they are listening to a lot of
>people and perhaps rethinking their strategy with an eye toward
I have not said that UN people somewhere are not concerned. Our UNICEF
office in Windhoek is very good and what they do is quite admirable.
However the major projects that are happening at present are RINAF,
CABECA and SDNP. But that's about it.
I have criticized RINAF severely in the past, but they still asked me to do
a system operator's course in Senegal last month :-)-O. They have seen
the errors of their ways, but I now understand why they went about
certain things in certain ways.
CABECA is well intended but poorly funded and quite ineffective.
It may just happen that I will have to eat my words about SDNP,
but we'll see :-)-O
Please not in this context the CC list :-)-O
>What seems missing in most debates is a clear idea of how donors
>with investment funds can apply their capital sustainably. Suppose a
>donor has, say, US$600,000 to "provide full Internet access" in a
>country. How should these funds be spent to "provide" the access?
Wrong assumption. Should not do that, "provide full Internet access" in a
As soon as you start funding the operating costs you stop any effort
towards sustainability dead in its tracks. The service is free, it is
"theirs" and nobody really gives a damn anyway.
If it is "ours", and we have had to put in blood, sweat and tears, we can't
just allow it to fail...
>I was recently with a colleague reviewing a World Bank report (while
>seated in their coffee shop -- good cofee, by the way, and reasonably
>priced) that suggests recurring costs of about US$130,000 a year from
>a 64kbps VSAT hub. Those figures seem about right to me. In a demand
>assessment I wrote for Niger, I worked with the proprietor of a local
>computer firm to draw up a business plan, and calculated recurring
>costs of about $160,000, but that included a roughly US$40,000
>allowance for capital depreciation.
I totally disagree with these figures, see below.
>As an aside, I reworked the figures on a VSAT full-Internet
>business to compare with a Fido or UUCP dial-up email/document
>exchange business, providing full international email and local
>html/ftp document access to about 40 computers in a country,
>requiring a revenue of only about US$20,000 a year to break even.
>You can have a look at my numbers for this type of enterprise on
The document is hardly readable and to ask developing countries to get a
more modern reader instead of just formatting it properly is downright
Anyway, I have downloaded the file and had a look at the numbers, they
don't make much sense other to prove my point, that you can not do it
other than on a non profit basis.
With regards to calculations, I will use 500CFA equalling 1$US. This may
or may not be correct but it makes easy calculation and comparison.
> COST PARAMETERS
> Cost of International Telephone Polling 150 + 4 per mailbox
150$ per month and 4$ per mailbox is just no way of calculating the
anticipated mail volume.
> Cost of International Email Gateway (Overhead) 10
Not that it matters, but you get that for free.
> Cost of System Equipment ($5000, 3-year depreciation) 140
5000$ (2.5 Million CFA) buys you some decent system, I quite agree with
that, in fact it buys you two computer systems with ethernet cards, three
modems each and one tape drive.
However you don't need to buy the systems. This can very well run on an
existing system at a university, or an NGO or an idividual.
> Cost of Office Equipment ($1000, 3-year depreciation) 30
1000$ (540000CFA) for office equipment? What equipment would that be?
> REVENUE PARAMETERS
> Number of Mailbox Rentals
> Year 1: 20
> Year 2: 30
> Year 3: 40
These assumptions are way too low.
> Price of User Mailbox Rental 30
30$ per month? Who would pay 15000CFA up front per month?
> Price of User Installation 300
This is utter nonsense! 150000CFA to run a SETUP.EXE or an INSTALL.BAT?
> HYPOTHETICAL THREE-YEAR INCOME STATEMENT
> YEAR 1 YEAR 2 YEAR 3
> Mailbox Rental 7200 10800 14400
> Installations 6000 3000 3000
> TOTAL 13200 13800 17400
> Telephone polling 1380 1620 1860
> System Equip Deprec 1680 1680 1680
> Office Equip Deprec 360 360 360
> Personnel System
> operator $800/month 9600 9600 9600
You got to be kidding me. 400000CFA per month for the SysOp to do what?
> User support
> $5/user/month 1200 1800 2400
What would that be?
> Facility Rental 1200 1200 1200
What facility? You need to put the machine into an existing room, even if
you are commercial you can't ask for rent again.
> Telephone 1200 1200 1200
Who do you want to phone? I grant you the basic phone fee and
installation fees which should be set up in the telephone dialup fees,
but users should phone you not you them.
> Utilities 600 600 600
What utilities? Water and Lights? Come on.
> Office, Misc. 1200 1200 1200
Sure, this I agree to, you got to send out the acounts :-)-O
> Ordinary taxes
> (stamp, sales) 600 600 600
What would that be for? The customers pay that directly.
> TOTAL 19200 19840 20680
> GROSS PROFIT (LOSS) (5920) (6040) (3280)
Oh no, this is not loss, this is profiteering!
What you got to do is to use existing facilities first.
You need of course a Unix system to run Taylor UUCP and smail with batched
gzip. Linux is one of several extrelemy cheap versions, a set of 4 to 5
CDs cost 35$US and icludes source code for everything, documentation and
tons of stuff. It can run on a 33Mhz 386 as in linux.lisse.NA, the former
national and now coastal gateway :-)-O. A 66 Mhz 486 or a nice Pentium of
course is quite ok too... These things can be repaired locally or you can just
rip the disk and put it into another box.
I don't believe in billing by volume because it creates lots of hassles and
work. However you can take the price per KB and I am positive one can do
it for 10 CFA per KB (or less) and just double it, the rest going into the
kitty for hardware.
You can then charge each subscriber a monthly fee calculated on an average
consumption previously which is payable in advance, failure to do so will
automatically block the access. This is fairly easily to implement under
Unix and can be connected to the accounting system, so if the check doesn't
come in access is blocked. Even on the other side smail can be made to
refuse certain mails and uux could be used to configure this on the fly
That way you only have to monitor for excesses and those you can sort out
very quickly indeed.
Once the user base is high enough you can get yourself a router, wire it
into the ethernet, connect it to the leased line and run uucp over IP
until you figure out DNS. Actually it is quite nice because it will run
as redundant backup, if the IP UUCP can't connect it may just dial the
phone ever so often :-)-O
We first used my own computer then the University's system and then bought
our own from membership fees.We went from 20 users to 100 after one year
and now after 18 month we are at 300 accounts/domains but with between 500
and 800 users. We pumped 40MB per month after one year (email only,
gzipped which expands to 120MB) and now with the IP line our router figures
show between 1 and 4 MB per *HOUR*, or 1440 MB per month (which of course
includes everything, TCP UDP and whatever).
More than half of the expenditure in the esample put up on the USAID server
is for the operator/owner of the system, hidden or direct. I find this nothing
The line, whoever puts it up first, should be a national infrastructure, accessible
to anybody at cost plus maintenance, not a commercial monopoly.
If you had 600000$ you should offer it for hardware and training under the
condition that the country (Government, national TeleCom, Federation for Science
Development etc) concerned guarantees the leased line costs for lets say 6 years.
100000 per year will go a long way.
The local NGOs usually know very well how is capable, corrupt (or both :-)-O). They
can form such a service NGO very conveniently splitting the overheads among them
(one puts the machines up, the richest one dows the mailing, one does the accounting
Manpower? We have universities. The Computer Science Departments and
Computer Centers will jump at the chances getting involved. Student labour
is cheap and they will learn. Who do you think owns the hugely profitable
Internet Providers which dominate South Africa now when Internet really
started there only 1992?
Hardware? Software? See above.
Being an NGO has also the advantage of qualifying for the educational discount.
I have said this before, a VSAT line can be had for less than a decent Mercedes
or BMW for the presidential motorcade per year. It just needs the political will!
If the government of Niger told Sonatel: "Put the VSAT line up or else!" they
will hop to it.
Dr. Eberhard W. Lisse \ / Swakopmund State Hospital
<[log in to unmask]> * | Resident Medical Officer
Private Bag 5004 \ / +264 64 461503 (pager) 461005 (home) 461004 (fax)
Swakopmund, Namibia ;____/ Zone/Domain Contact for the NA-DOM
Vice-Chairman, Board of Trustees, Namibian Internet Development Foundation,
an Association not for Gain. NAMIDEF is the Namibian Internet Service Provider.