After a busy two weeks, we’re back with a double weeknote.

In week #209, we were onsite with a client working through some data mining efforts. It’s been an excellent week of searching. A few set backs were we had to backtrack a bit, but that has allowed us to get further with our exploration. Right now, we’re big on giving things names. This is especially important with data sets. Rather than refer to a cluster of data as hits above one standard deviation and in the top 10% of engagement, why not call them “super stars”. Names are loaded, so it is important to be flexible and let things organically come about. When the team, the company and potentially the customers are all talking the same language it is much easier, and friendlier, to use these artificial names.

This idea harks back to the old days working at a small start-up here in Iceland called CLARA. We did loads of text analysis for video game forums. As part of that we looked into the four types of gamers; Achievers, Explorers, Socializers and Killers. This is part of the Bartle test. While we were deep in natural language parsing and clustering types of people, we also drew inspiration from Claritas PRIZM  Market Segmentation by US zip code. They have 66 different groups which describe different people and lifestyles.

Working with our new client, we have been developing similar terms. For now, they are easier, more descriptive ways for us to talk about their customers and content. Since the underlying mathematics are still in flux, it is better to anchor the concepts to a name rather than an equation. This week we coined the terms “King of Wrong”, “Evil Twin” and “Right seems wrong” which describe the data histograms. We’ll continue to refine the datasets, some of these might merge, new ones appear, and the equations to file the data into these buckets will certainly be tweaked, but at least we’ll are building a consensus and language to discuss and fixed this issues. It is easier to explain why “Evil Twin” are bad rather than “complimentary bars of a histogram, both within a few percentages of each other and both above the 25% median”.

We also floated some questions on Amazon’s Mechanical Turk this week. We are working on a few new projects and ideas for Analog.is. We wanted to gather some ideas, suggestions and stories, but didn’t want to contact our previous customers or fan-base and ruin any surprises and bias our sample. So we asked 50 random strangers to tell us a little story. What came back a few days later has been amazing! For $5, we have loads of great pull-quotes, ideas and themes to work with for our next project.

In week #210, we spent some time working on student survey work, getting up to speed on recent code changes. It was also a week of meetings and getting caught-up on various other projects. We had a good long meeting with some of our Berlin contacts which are in town this week. There is a long running project which we have been wire framing and planning. Getting the chance to sit face-to-face and work on the project has been really good.

This week we also got to see the first prototype version of our Tyvek project. Back in week #206 we ordered some Tyvek and in week #208 it arrived. Now we have a basic prototype to begin testing. There are still a few more parts that need to be procured, but the basics are there. Without the extras, it is currently weighing in at around 33g. This is super important. Right now, we’re working with a matrix of four variables; price, recyclability, weight and strength. Finding the sweet-spot between all four of those will be difficult and will require trade-offs, but that’s why we’re prototyping.

Analog.is has also been keeping us busy. We’ve had several more notebook orders and we’ve ordered our next round of Analog Markers. These are for Null Island, a fictitious place at Lat, Lon 0,0. Over the weekend, the Analog.is team is meeting-up for a big planning and work session. Soon, we’ll be able to reveal what’s happening next.

We also spent some time this week lining-up some more groups to talk to about their ambassadors and volunteers. We’re learning how they interact, what their needs are and how much of the workload is distributed between employees and volunteers. We have a project in the works to help with many of these things, but want to get the data straight from the source rather than making any assumptions.

Bric-à-brac

This week, there has been a discussion about sending a probe to the moon Titan. It has lakes of liquid methane which work in very similar ways to water. The interesting thing about this proposed mission is that it wouldn’t be a rover, but a submarine.

Finally, we can watch a man strap a GoPro camera to his trombone. The subsequent movements of the slide jump the camera in and out of his face creating a bizarre world of a dancing Point-of-View, but that’s just how a trombone and sound waves work.