July 04, 2006
Google Expansion & MENA - Market Interest in MENA
A rather quick note to draw attention to what may be a somewhat under noticed story, from FT: Google looks to expand in Middle East.
I noticed this when my usual robot searches brought up this on both the career angle and the news site searches. An interesting development.
First, on the somewhat less important direct question, the Google Arabic language searches, well, suck. My google robots in Arabic rarely if ever return anything remotely relevant or interesting. As a matter of business, this strikes me as a serious problem. The other issue that came to mind was their roll-out of the country locals is pan-Arab, but in the Maghreb the utiliser base is heavily baised to Francophone usage (although the users are bilingual, reader bias is Francophone. While expansion of users and official Arabisation policies in schools probably going to expand the Arabophone base, the heavy economic sphere bias is francophone). The Pan-Arab type approach strikes me as a tactical error, probably in part driven by having a Machreqi core staff. Happens a lot with MENA regional projects, they end up losing their Maghreb end.
Second, on the issue that struck me more than the above technical trivia was the sign that there is growing private sector interest in long-term MENA markets. Driven by the Gulf money image I am sure, but Google appears to have an idea of trying to lock in users in emerging markets and making a play off of their future growth opportunity.
One may look at this as an interesting potential early indicator of MENA market liberalisation paying off.
There are others, of course, I'm not an idiot to trumpet off of big name flashes like this, but it's a sexy angle.
TrackBack URL for this entry:
Arabic is doomed in terms of technology, if changes are not made to create a more native support for it.
Without getting into the technical details, Arabic in computers relies heavily on the Unicode bidirectional algorithm. That braindamaged algorithm prevents many basic uses of computers in Arabic and make quite a few things counter-intuitive. It is based on the assumptions that you're going to write significant parts of both Arabic and Latin in the same text and that your low-level system coordinates go left to right. There's a very clear bias towards latin based scripts users who will only occasionally write Arabic. And that definitely makes the choice of Latin based usage obvious against Arabic based usage when you master both to the same extent (like in the Maghreb), not to mention when you're more at ease with Latin scripts. Speaking of the specific problem of text searches, it definitely multiplies the cost of creating workable solutions because (grossly put) there's a discrepancy between the way the text is stored and the way it's displayed (humanly learnt and understood), unlike with Latin based languages. Unfortunately, the few Arab half-wits who have some technological knowledge think that since the Unicode Consortium made that algorithm a standard, then it's the Word Of God and It Must Be Right.
Another problem is there's no constant shape print glyph. Unlike Unicode's, it's a minor problem, but it adds to the complexity. Some square Arabic (think square Hebrew) script - one that would be readable by anyone who already reads Arabic today (reduces cost of learning and resistance to change) - needs to be created.
So there, fuckers, your language will fucking join the dinosaurs.
Posted by: Shaheen at July 5, 2006 05:00 AM
it is a bit sad to see them going for the Machreq over Maghreb market but, then again, the Maghreb market has been using google.fr for quite a while.
i can only imagine the fun that awaits google in locating server farms across the MENA. more fully penatrating many of the sub-markets will require server location within the country (and behind the country-wide filters) to keep the speed relativly fast. using Tunisia as an example, as of March of 05 (yes, my info is a bit dated now), the whole country was bottlenecked with a 5 megabit pipeline. data could flow just fine within the country, but with a centralized link to the outside world, rather than many decentralized (as in the USA and Europe), external traffic had a rough go of it. i've heard from a few people that said it's gotten better since the world internet conference back in November but i'm not holding my breath. the country's filter system will continue to hold Tunisian internet speed back even with larger pipes. why should google care? because google's products can become very bandwidth intensive. to get around that, google will have to locate server farms within MENA countries. to do that, google will have to let MENA countries into their inner workings to "properly filter things." google will most likely not do that, although i was surprized with their china approach, meaning google will have to locate the MENA server farms in Europe and elsewhere. well, you get the picture.
L: thinking of going to work at the GooglePlex?
PS: L: did you notice that Livejournal has been blocked by the Moroc filters?
No, but then I am still in exile.
Posted by: The Lounsbury at July 5, 2006 10:25 PM
well, i imagine you have a VPN setup (at work at least) anyway to get around such nasties as filters and censors. i dont know how i would have survived in tunisia without such a tool!
btw, since we mentioned Machreq, i give you http://www.sheikhmohammed.co.ae/
the personality cult continues!
Actually I never noticed any serious filtering effort. But then I hardly ever connect to politically suspect things.
Posted by: The Lounsbury at July 5, 2006 11:10 PM
come now, L! surely you must visit Mohammed VI's sister's "Marakesh Gone Wild" website from time to time!
i'm sure that moroc filters much less than certain countries farther east. some stuff put out (and i think covered by aqoul) during the world internet conference in tunis last november was quite interesting in that regard.