You can never have too much documentation!?

A picture is worth a thousand words...

Ubuntu, the dumbing down of Linux?

I've spent a couple of days now getting an Ubuntu installation into shape for some development work, so I thought I'd share my experience.

I like Ubuntu, because I'm pretty dumb when it comes to Linux, and Ubuntu lets me get things up and running pretty quickly without having to remember all the configuration 'tricks' needed to get a basic system working. It just works...
However, this time I needed a development environment, and I was surprised that the install of Breezy did not include gcc. Or lynx or ncftp, or pretty much anything that would allow you to do a compile or use cpan (supplied) on the machine.
Maybe I'm old school, but a Linux that doesn't allow you to recompile the kernel (without installing additional stuff) just isn't Linux.
Anyway, my requirements were very basic, I wanted to run Perl to connect to an Oracle 9i database at work, using the fantastic DBI by Tim Bunce.
I had done installations of Oracle on Linux before, and it stretched my limited Linux skills to the limit (patching libraries in the kernel in order to get the Oracle Java (read: portable, run anywhere?!) installer to actually start!!). I was not looking forward to a re-run of that.
Fortunately Oracle have released Oracle Instantclient at version 10, and it looked promising (especially the 33Mb download, instead of 650Mb for the normal one).
So, first things first, let's get Perl working with cpan. Hmmm, half of the tools are missing; no 'make', no 'gcc', no 'lynx', no 'ncftp' etc.
The machine was also behind a firewall, so apt-get didn't work until I configured the http_proxy variable (as 'http://NN.NN.NN.NN:8080'). Then I had to configure apt-get to look at the broader repositories as most of the above apps were not in the standard lists!? Finally I got the applications necessary for cpan to run and then decided to do an install of a Perl package to test out the compile process before attempting the Oracle DBD compile.
XML::Parser would be useful, so I ran 'install XML::Parser' from cpan and watched. Serious compile problems! What the?! 'cannot find Expat.h'.
After some searching, it appears that Expat (by James Clark) is no longer provided with the package and must already be on your machine. Ok, 'apt-get install expat'; nope, package not found. Again some searching and I find libexpat1-dev (?!) install that, after some problems with apt-get repositories require me to rebuild the source.list, and then run a perl Makefile.PL; make; make test...
Everything works! sudo make install...job done (oh, except I haven't done the Oracle install yet!).
Get the instantclient basic (I got the 10.2.0.2), sqlplus and sdk zip files from Oracle (you need an OTN username and password - request lost password!). Unpack the zip files somewhere and create a directory structure for Oracle (I used /opt/oracle). Put sqlplus and glogin in a bin directory under /opt/oracle, create /opt/oracle/network/admin and put the tnsnames.ora file there. You can create a lib directory for the *.so* files, but I didn't bother, I just made LD_LIBRARY_PATH use /opt/oracle. Leave /opt/oracle/sdk as it is, even though it's tempting to rearrange.
Amend your .bashrc (or equivalent) to add the oracle environment vars including ORACLE_USERID and TWO_TASK. ORACLE_BASE and ORACLE_HOME are /opt/oracle. Add /opt/oracle/bin to your PATH.
Now try running 'sqlplus $ORACLE_USERID', this should connect to the SID defined in TWO_TASK. Assuming that works, then go to cpan again.
'install DBI' and 'get DBD::Oracle' (you could just install DBD::Oracle, but it's easier to see what's going on doing it manually)
Change to .cpan/build/DBD_Oracle*, run 'perl Makefile.PL'.
Assuming no problems, run 'make'
If everything compiles then you're almost home...
run 'make test' (make sure your ORACLE_USERID and TWO_TASK are defined correctly)
If tests are ok, then 'sudo make install' and head off for the pub (or a coffee).

Ubuntu, it just works (terms and conditions apply, please see pack for details)!

It ain't worth a thing,...

Sometimes I think that Mobile phone manufacturers want to add a kitchen sink to the device we carry around. Cameras, MP3 players, rfid readers, Key functionality etc.
People prefer a single device for everything, they say.

Ok, let's see them try to integrate this into a mobile phone!

They can use this other link as a guide-point?

Perhaps it takes something as un-integrateable as the Nike example to show us the benefits of separation of function; I can only hope...

How much is a customer worth?

Vonage IPO raised a few eyebrows with its value per subscriber, and it got me thinking about how we value our customers.
It occurred to me that telcos really only know 3 things about their customers (if they're lucky); their name, their address and their ARPU. They know some other stuff that confirms identity and they may know a bank account number, but essentially a customer is packaged in a little box that says "J. Public, $25/month, credit rating: Good".
This tells me very little about the customers potential, or why they are my customer, or how 'sticky' they are likely to be. What are the things that concern them, what do they do, who do they call?
Ok, we're getting into a grey area here where the telcos could get a lot of statistics together and mathematically calculate some reasonable answers from their data; I'm just not sure they would ask the right questions...

Vonage has 1.6 million subscribers that it knows very little about. Each of those subscribers was valued at around $1600, based on growth potential. For what? More VoIP?
Let me take another company with a growing customer base, significantly lower revenues, but, imho, with a far higher customer value. LinkedIn
The reason the value is higher is because there is community and trust built into the system.
When I first signed up for Linkedin back in 2003, or thereabouts, I was excited at its potential as a directory service, rather than the recruitmentfest that it has become.
Combine LinkedIn with Iotum and you would have a pretty powerful way of screening calls, without much in the way of 'relevance' logic.

But LinkedIn is failing to see that its growth is hurting its value. The value of their customers is being diluted as their service expands.
Why? Because of Ron Bates! Now Ron might be a nice guy, but he is one prolific example of the disease affecting social networking sites. Ron has OVER 25,000 connections. I can connect to hundreds of thousands of people, but that name always pops up! He breaks the trusted network. I would remove him from my list, but in LinkedIn that is not an easy thing to do, and perhaps he wouldn't like it and bring all his 25,000+++ buddies down on my profile, changing my name to Mud!

In calculating the worth of a customer, the links we have in the social network have value, but that value is not uniform. The social networks, like MSN, Yahoo, Skype, AIM etc MUST allow us to reflect the value of our connections within the social fabric. If they don't, they will be compared with the shoebox telcos like Vonage and they will miss out on a whole lot of opportunity.

It's not what you know, it's who you know that counts...

G'data, mate!

I feel a new breeze on the web, and I like the way things are looking...
Gdata extends Atom into search/query capabilities, Apache releases for S60 mobiles, identity is getting to be something like your MySpace pages.
Mashups are getting personal, dynamic and portable...

Directory is not DNS, communication is not just voice, contacts are not one-dimensional, attention is money, fame has a long tail (15 minutes can last a lifetime), when -> if, where -> how, brands -> communities, served -> server (or subscribe -> publish), QoS is ease of use.

The thing about all this is that nearly all of our communication can happen on the Edge, a community doesn't need any operator to authorise, serve and manage the community.
It can be a Naked Community on Naked Broadband.

Now that's the kind of nudity we need more of!

Quality vs Value vs Utility

I could have written about this at length but I don't have time and I'm feeling the time to say something is now.
I've argued often that the best example of Quality as a differentiator is the mobile industry. Quality: low, price: high, usage growth: exponential!
If you ask the Comms manager what they want from a mobile operator, they'll put quality high up on the list, but it's a red herring. They want quality of service so they get less complaints, in the end they will choose the cheapest option with the most utility (not necessarily equal to features)
You could equate this to low-cost airlines where a businessman will fly the cheaper option if the times suit approximately and it doesn't dump them in a hangar in the middle of nowhere (and they get some credit for choosing the cheaper option!).

Quality only has an impact if all other factors are equal, especially cost, or if utility is severely impacted by a lack of quality. E.g. 6 hour flight delay followed by cancellation and abandonment (n.b. this has happened to me only on 'quality' airlines!)

In services, if Quality >= sufficient then Price/quality -> 0 as quality increases. i.e. doubling the quality will not double the price paid.

The quality of mobile calls shows how low the bar can be in terms of sufficient quality (though perhaps living in thailand where mobile completion rates hovver substantially below 50% might have polluted my memories of mobile use in Europe)

Lack of quality can quickly reduce the price to zero too, if quality is not sufficient. So quality, value and utility are a balancing act, with the only safe way to increase the price paid being to add utility.

Not exactly rocket science, is it?

We are family, (I got all my sisters with me...)

As the telco market gets more and more crowded, and the mechanisms to attract customers become more and more cutthroat, it's worth the telcos considering a more loose definition of family and friends.
I reckon it's about time telcos started making on-net voice calls free. I can't think of a better way to get a group of customers onto your network, providing their information and giving you a good shot at the money they do spend on communication services. Telcos already do this for big customers (based on incoming call revenue share), now we're all family, right?!
How do you think your Auntie Julie got on Skype? Sure, viral marketing, but did she sign up because it gave her cheap calls to Ecuador? No, it was because her son/daughter wanted to call her for free...

Is the world flat?

There are a number of posts around the blogosphere taking issue with the pricing strategies of telcos. In general, the view seems to be that usage based pricing leads to customer dissatisfaction unless it is done just right.
Dean Bubley writes a seminal piece on Value-based pricing that every operator should pin in their marketing departments (though I'm not sure about his examples, for me, airport food is definitely an example of Resentment Pricing!)

The question is; is usage based pricing on the way out?

For high value content, there is an argument for pay-per-use, but for marginal applications like SMS and Voice, flat rate, all-you-can-eat is becoming the only practical option. How much does it cost to collect and bill the events if it's not flat rate?! How many telcos know the answer to that question!?
The customer needs predictability in their monthly or daily charges, especially as the services diverge into TV, information services and a whole host of other stuff.
From a company valuation perspective, the customer is king, and the more customers you have the more the company is worth.

I believe that we are moving to a model similar to cable, where the customer buys a package of services and potentially has some pay-per-use capability in there for high value events, but the era of tracking every message sent between users is nearly over (unless government regulation prevents its demise).
P2P communication is flat-rate, and possibly free (if you believe Carphone Warehouse advertising!)

Anything else is just Ignorance-based pricing, tending towards Resentment-based pricing.