Monday May 30th - Friday June 4th -- Accra, Legon

A picture of Omar | A picture of RJ and collegues | Samir, RJ and Omar

I've been a little bit lazy about the journal this week. Sorry about that. We've gotten a lot done on some things, and not much on others. The strike at Legon persists, so we moved some equipment to BusyInternet and Ashesi and set up DiSC there. We went to the "opening ceremonies" of AITEC, an annual ICT conference in Accra and made lots of connections, did around 250 surveys at four different Internet Cafes (with the help of a few BusyInternet staff and our "interns") and a focus group on DiSC at Ashesi with students.

Monday

Tuesday

Ronnie and I talked to the head organizer and negotiated free passes (White Express), but there were a lot of disgruntled attendees who had anticipated being allowed to attend and in at least one case traveled several hours to join in, only to be turned away as soon as any substantative discussion began. The invitations we received made no mention of the distinction between the conference and the expo.

Most of the presentations turned out to be vague policy discussions anyways, and we left the conference in the afternoon. I can't help but think that there was so little solid content in the presentations because all of the people who understand the issues were turned away at the door. I couldn't find the guy who traveled to come to the conference, so I left with an "Organizer" badge in my pocket.

On a more positive note, the technical director of the Senegal based Open Knowledge Network presented details on his organization. Basically, OKN provides a "broadcast" model by which web pages can be published via an email message, and slowly propagated to the whole network. They accomplish this by using a two tier network in which the upper teir receives and relays new publications and the broadcast them to the lower tier. The lower tier receives data via Internet or SneakerNet (aka floppy disk). I think that different OKN clients can subscribe to different subjects.

The goal of OKN is to encourage publication of local content, both an important goal, and a hard problem. This brings me to a little hypothesis which borders on obvious, but can not be taken for granted: Ghana does not have more local content because it is more expensive (in terms of time, money) to access locally hosted content if the person viewing the content does not get internet access from the same ISP that hosts the local content. That is, any locally hosted content must go out of the ISP to Europe via satelite, and back to Accra via satellite, just to get accross town, so the net benefit is higher if you host the content in Europe. My hypothesis is that this leads less locally generated content, because it is difficult to generate and it is slow to access, regardless of where it is hosted.

If, on the other hand, there were a fast local backbone in Ghana, there would be a hugh performance and hence economic benefit to consuming local content, and thus a high demand. Samir points out that high demand may not translate in to greater supply, but I personally think it would.

I have been thinking about this problem of using point to point links between different ISPs until all the ISPs here can decide how to build a local Internet Exchange point. I wonder if a more tangible monetary benefit would benefit or hurt the prospects of my idea. For example, if the two cooperating ISPs agreed to route each other's traffic to the Internet in the event of a WAN outage at one of them, then the both ISPs would receive an immediate financial benefit from the arangement. The ISP who's WAN link went down would benefit because they would not lose customers while the link was down (though they would be charged a higher fee than normal for their internet access, and thus have smaller margins). The ISP who's WAN link stayed up would receive a direct financial benefit by charging access fees to the other ISP. It would also encourage each ISP to maintain its network well.

Of course, as Gregg Zachary has pointed out to me, generating this type of cooperation is really the problem rather than the solution. Because of lack of strong contract law, and other sources of mistrust (to use Gregg's words, in the business community in Ghana "voluntary associations are rare, a legacy of colonialism and post-colonial authoritarianism,") it is difficult to get businesses to trust one another.

So the benefit to both ISPs must be very clear before they will be willing to take a risk on cooperation. I wonder if the benefits would be great enough:

1. Greater redundancy of the WAN link, so less downtime, so more revenue
2. Financial benefit to helping fellow ISPs recover from outages
3. Reduced WAN traffic (since traffic to the other ISP can be routed via the point to point link

Wednesday

Thursday

Afi lead us around Makola Market, the hugh (around 8-10 city blocks or so) market in central Accra, looking for cloth to make clothes, some supplies for Afi's store and other small items. I wanted to take some pictures, but people didn't want their picture taken, though I don't yet really understand why. On one of my previous trips to Ghana, I was told that people are afraid that I would use the pictures to make Ghana look bad in a magazine: not an unreasonable concern.

In lieu of a picture, I will try to describe Makola market: the streets on the perififery of the market swell with pedestrians walking on the sidewalk and spilling onto the roads, where cars ruthlessly push ahead, and girls with giant towers of goods perched on their heads narrowly and non-chalauntly avert serious injury at the hands of angry taxis. Entering the market, we are greeted by ladies sitting on the ground or small stools selling smoked fish, fresh fish, pig's feet, giant snails makeing a mad dash for the edge of the table, and other miscelaneous meat (Omar, remembering smoked sardines Saudi Arabia, wants to buy some before we leave).

"Kwasi, Aitiseng?" (Kwasi, how are you), the ladies enquire (Kwasi is a Ghanaian name which means you were born on Saturday. People sometimes call strangers Kwasi to make them feel welcome). Children say "Obroni Coco" (brown skinned forigner), and I am relieved to not be the center of attention. Entering the market proper, we pass by various food stuffs: giant yams, a row of hot peppers, tomatoes, ocra, more meat etc, as women weave through the narrow lanes between stall after stall, truely giant trays or towers of merchandice magically suspended above their heads. I am reminded of goldfish, who never stop moving and always continue foward, weaving between each other, intent on their goal.

We enter a structure perhaps 150 meters long and 50 meters wide filled, from corner to corner, floor to ceiling, with brilliantly colored clothes. This open giant, lofted ceiling building, is packed full of printed cloth: the other types of cloth, such as Shadda and linen are located elsewhere. Afi asks me to take the lead to pick the pieces of cloth that I want, but I can not: I want to buy everything! Both the variety and quality are truely amazing, overwhelming. One lady here wanted her picture taken, so I did get a picture of her in her stall, and you can get an idea of what one stall of about 500 in the building looks like.

We also bought some food supplies for Afi's store at a little storage room piled to the ceiling with boxes, and a man standing admidst the boxes selling to Afi, replentishing his merchandise via another lady, and keeping books with his partner. Slightly ladened with more goods, I put one of Afi's boxes on my head, to the great amusement of everyone around. After a death defying walk down the narrow street, I came to discover that typically only women put small things on their head: men reserve their head for large, akward and/or heavy items. In any event, I'm not sure how much of their amusement was because an Obroni tried to balance things on his head, and how much was because a man tried to balance a little box on his head...

Unlike the Art Center, where salespeople assault forigners with an aggression that borders on dangerous, we are largely ignored here, and this is the Ghana that I love. After a late lunch, we head back home and I work on my laptop while Ronnie and Samir work with the interns back at Legon to enter the data from the surveys into a spreadsheet.

In the evening, Afi cooks us dinner, we finally meet Guido Sohne, and I finally get to see my longtime friend Albert, who's wife is pregnant, and thus has had trouble connecting with us. Guido is a mostly self taught and rather sucessful programmer who everyone here seems to know, and has been featured in more than one article about technology in Ghana. We have a broad discussion about his interests, our project and other things, such as tracking social networks and implementing ubiquitous storage.

Friday

Omar, Samir and Ronnie and the interns are over at Legon, still entering data from our surveys. I head over to Ashesi for a focus group with eight students. Omar shoots over from Legon with a tape recorder (oops, I forgot it... thanks a million Omar!), and we both agree that the focus group gave some valuable feedback. Here are some of the highlights:

• They all agreed that they would like to see data get pre-fetched based on what site you have visited. If you visit a particular web site, they want to prefetch links off of that page.
• Because their lab is closed from 8pm to 6am, there is 10 hour/day in which the network is basically idle. This is a good time for prefetching
• Privacy is a concern for several of the members. Most of the students wanted to make sure that they couldn't be associated with a particular document, but didn't care if a controversial document was stored and indexed in the cache. One of the students felt the opposite. He didn't mind being tracable with a particular file, as long as he could remove things out of the cache that he didn't want to be indexed.
• If given a choice between a search result that downloads fast, but is not from a "trusted source," (say, Lansing State Journal instead of BBC.com), but is on the correct topic on the one hand, and a result that downloads slowly but is from the trusted source on the other, the students would probably just download the "trusted source" in a seperate window and browse the fast results while they are waiting.
• The students tend to multitask a lot. This is probably at least in part due to the high latency of web requests.
• The students would use the cache even if there was no bandwidth/latency problem, because they would like to see what other students are viewing, and also because they don't trust that the network will stay up, even if its fast.