Geeky Peek

This Friday we are hosting New York’s iconic Jelly co-working event – http://wiki.workatjelly.com/JellyManhattan-2009-05-01

My wife is a freelance video editor and attended one of the Jelly’s a while back which got a few of us thinking about hosting the event at Peek. Jelly rotates around getting hosted by companies, people’s homes, co-working spaces, etc. I think Jelly is an amazing idea, there are tons of freelancers, bootstrap entrepreneurs & folks who work from home who look to get a chance to work in an office with others. Jelly has been the perfect answer for those folks and I think its great that it is catching on wildly across the US (and even globally).

So if you’re in NY and you want to stop in for a bit and work with the crew at Peek feel free to stop by. We’ve got wi-fi (I hope it holds up with the extra guests), whiteboards, and coffee. Seating is not unlimited (I think we can handle 15-20 people).

I will make one note – we do not have the usual start-up super trendy all open concept space loaded with sofa’s (cough cough drop.io). We have a bunch of separated rooms with walls and cheap Ikea desks (but nice chairs). Peek is a functional-driven company

It was a very cool night. Lots of Peeksters came in to get their devices upgraded and we got to talk to lots of customers which is always learning and helpful.

I showed lots of people how to use some of the apps we pre-load (weather@askpeek.com, news@askpeek.com, etc). There was lots of customer delight in getting crc’s incredible weather report with images and all (http://www.charleschilders.com/ent/).

The other moment of customer delight was the overall snapiness in using the device from before. I’d love to get reactions from some of you on earlier versions vs Pronto and usability.

We did a few video things at the event there that I’ll post later (some interesting, some silly).

For any of you in NYC, we are upgrading Peeks in Midtown at the Gallway Hooker. If you have an older version of the Peek software on your device, we are upgrading customers to the Pronto version.

And if you don’t yet own a Peek, feel free to run to Target and buy one of the older Peeks and we will still upgrade you to Pronto.

Also feel free to stop by and just say hello if you are a friend of Peek.

I’m live Yammering to the rest of the folks here at Peek so they know what a night in the life of a deployment platoon is like.

Some of you may have read this article in the New York Times about Bruce Sterling’s now infamous quip “connectivity = poverty”. He muses that there is a correlation between the usage of texting, phones, facebook, skype, IM, twitter, etc and poverty. The final capping quote “poor people sure love talking on their cell phones” is particularly lovely (sarcasm). At first glance, if you think about life from a more comfortable middle-class view it seems to ring some truth.

But let me make this clear, his statement is pure and utter bullcrap. Having helped setup and launch a phone company in Haiti (www.digicelgroup.com) and having lived a year and a half in Kingston, Jamaica, I find Bruce Sterling’s statement smells of the hypocrisy of one who has never experienced real poverty. His statement is in fact completely backwards, the opposite “Unconnectivity = Poverty” is a much more representative truism. Throughout the world the poorest countries show the lowest rates of adoption for internet or mobile technologies. And throughout the developing world there is study after study that shows “connectivity = decreased poverty”. I have posted a big pile of links to studies at the bottom.

Anecdotally, I can also share the case study of Haiti. In 2006, when Digicel launched in Haiti, GDP per capita was $1300… i.e. the amount of money a person had to spend was $1300. The average cost of a phone was $100, with a $100 activation fee and $40/month fee – that is $680, 50% of a Haitian’s income!!! We launched with a price of roughly $20 for the phone, no activation fee, with an average person spending $15-$20/month.

The net result, note that the population of Haiti is 10,000,000 people:

Year GDP Growth* Mobile Phone Users**
2006 1.80 % 60,000 – 80,000 subs
2007 2.50 % 1,000,000 subs
2008 3.20 % 3,000,000 subs

*Sourced CIA Factbook
**Sourced – my own knowledge of sales in Haiti

People in Haiti threw an impromptu parade in front of Digicel HQ one day when I was there. When I visited the extremely impoverished farmers in Haiti’s rural areas they were extremely excited to be able to have cell phones to arrange transportation of their goods and to understand market prices. These are words I heard from their own mouths (French immersion pays off sometimes).

In many poor countries, remittance from the global diaspora is a major portion of GDP. By way of example, Haitians living around the world send back about $1 billion each year to Haitians in Haiti. To put this in perspective, the overall GDP of Haiti is about $10 billion, so 10% of the economy is basically money sent back from family & friends living abroad!

Previously, 10,000,000 Haitians had to split communications across around 100,000 total phones (landline + mobile). 1 phone per 100 people. And of course, the owners of those phones used them as a for profit enterprise. They would get a really long cord and sit outside their home, with their phone, and charge people money to use it! As you can imagine that kind of communication system hampers the ability for people to be able to talk, especially abroad.

When communications increase, people can more readily phone abroad and get access to money. This is micro-lending at its rawest. More communications fosters more $$$ at the lowest income levels. And this example (remittance) is only one form of how increased communications uplift people out of poverty. There are plenty of other reasons – from efficiency gains to better knowledge of market prices – that also help prove the point. These points prove the time and trued recipe seen over and over again – “Connectivity = uplifting people out of poverty”.

So, Bruce Sterling next time, before you wax poetically, maybe you can do a bit of research. Or instead stick to science fiction and throwing large SXSW parties in your aristo-hedo lifestyle of the rich and quirky before lamenting about the life of the poor. Go visit the poor before you make veiled comments like “poor people sure love talking on their cellphone”.

References
1. http://www.independent.co.uk/news/business/sustainit/closing-the-digital-divide-1640433.html – study showing mobile penetration = GDP
2. http://www.egovonline.net/articles/article-details.asp?Title=Mobiles-Penetration-Boosts-GDP&ArticalID=2228&Type=FEATURES – study showing mobile penetration = GDP
3. http://www.huawei.com/publications/view.do?id=5771&cid=10665&pid=10664 – study showing mobile penetration = GDP
4. http://www.jamaica-gleaner.com/gleaner/20060526/business/business2.html – tells more about Haitians growth in telephone customers
5. ftp://ftp.zew.de/pub/zew-docs/dp/dp0431.pdf – about developing worlds and transfers of cash through remittance
6. http://isim.georgetown.edu/Publications/RCRCCPubs/Orozco/Understanding%20the%20remittance%20economy%20in%20Haiti.pdf – Haiti and remittance

Well with Pronto done and deployed, we decided to have a bit of a fun day here at Peek…. hack day.

A few of us have disappeared into a warehouse in Brooklyn (Red Hook) where we are going to build some zany, hacked out version of the Peek software. It’ll probably be really unstable and ultimatley do nothing… but if something cool comes out of our (mis)adventures, maybe I’ll put it somewhere so a few of you can play around with it.

Excuse the re-post.

Is the All-in-One Device Dream Doomed?
By Sarah Lacy
http://www.techcrunch.com/2009/04/13/are-super-smart-phones-still-too-dumbis-the-all-in-one-device-dream-doomed/

As some of you may be guessing, we’ve been making a lot of Go/No-Go decisions these days.

When your software is near ready, the Go/No-Go decision is probably one of the toughest decisions you’ll face. The trade-offs can be immense – time to market vs quality, having customers buying it vs having better customer satisfaction, overloading your ops team vs underwhelming your sales team… everybody is carefully balancing these major decisions to ensure that the best product is delivered in a timely way.

And on top of that you can have politics that cripple decisiveness if one isn’t careful. When it comes to a go you can have powerful people holding on to pet issues with the release team not agreeing on the criticality and need to solve them.

All of this can take an incredible opportunity to celebrate success together… “we delivered the product!!”… and turn it into an emotional tense moment.

So to account for that we try to have two values:

1. Team first, team second, team last – if we deliver the product but fall apart as a team, we did not succeed. The ends do not justify the means. One product release is a small step along the way of many steps that a lot of people have to take together.

2. Dissent, decide, move on – as a team we need to invite dissent and let people weigh in. But ultimately the owner of a project or a workstream needs to decide. Once the decision is made, everybody has to move on. You can’t hang on to ideas and be emotional about them.

The last point is actually really important for more senior members of the team. They have to avoid trying to pull authoritative rank or using other bully pulpit tactics to get their way across. I recently heard a great speech about decision making, it basically sums to this – “with the wrong culture in a company, everybody can undo decisions but nobody can make them”. That is what happens when senior guys try to pull rank, it kills the ability of the team to make decisions, and everything circulates upwards for decisiveness.

So in order to do embody these values, we have a very sage guy (Rob), who has been through many launches. He brought in a disciplined, analytical approach to the go/no-go decision.

Basically we have a number of criteria that make a “go” decision. Each one of those is scored “red, yellow, green”. We use Agile & iterative release processes, so we actually start charting this early on so that way the go/no-go decision is gradual and we all step into it over time… allowing us to slowly but surely understand tough issues and practice tough decision making as we go, while the pressure is less intense. I think this step is absolutely crucial. People who have not been thinking day in/day out about these issues for weeks can feel awkward trying to contribute to this process at the end. Its a note to ‘hero’ CEOs or others who feel the need to jump-in at the end, you can be an explosive, backfiring, inducing spark to a well-oiled release team. Weigh in with caution.

Underneath the red, yellow, green of a single criteria are a number of tests. Example “Performance” may be a criteria, but underneath that there are 2-3 complex test suites to show analytically that performance criteria is met. While one member of the team will own completion of the testing suite, we usually split the decision on status to somebody else to mark it ‘ready’ with a green (or not ready with a yellow or red). Its a separation of church and state if you will to help come to the right decision.

Our style and process works for us. We stay strong as a team through the emotional moments.

I’d love hear how others make these sort of crunch decisions? Google-style, perpetual beta-mode? Slam it out and let the customers test?

This morning we woke up to a full scale outage. There is a single point of failure in our system… only one across a very large system. It doesn’t do very much, it runs Apache & Linux, nothing else.

At some point last night the file system started corrupting. Our monitoring and stats started showing some funky stuff before the node just became completely unusable.

I can’t imagine how a really basic Apache/Linux setup gets corrupted, I mean this is one the most common and mature platforms in the world and while we have a pretty high traffic load there are other people in the world using Apache/Linux with obscene traffic.

So we’re digging around now to try and find out what happened. If any of you have seen similar issues in a Amazon AWS/Apache/Linux environment I’d love to hear from you.

March numbers are in and we hit our 99.9% target. We were a tad over 1 hour per month per customer.

The only outage on our system was on March 16th when we had a full outage for a bit over an hour due to our underlying network provider having a large data center failure.

There were two other outages of note in March which I have previously posted about:
1. Yahoo stopped accepting logins
2. Hotmail stopped accepting logins

A few other incidences and interesting notes:
1. We also had some issues on the weekend of March 20th with slow registrations and slow email receiving. Around 20% of registrations took from 20 minutes to an hour to receive your first email over a 36 hour period.
2. Yahoo registrations have been much slower.
3. We lose .02% of sent emails and most of those are around 10 accounts.