I know I’m late to the responses, things have been busy and I wanted to edit my ideas and think on them before posting.
I guess I have a totally different take on this article.
My overall impression is that I thought is was really cool with a few things I didn’t think would work for me but a lot more I totally resonated with. I’ll go into more explanations here.
When I read the article I found myself clicking to nearly every external link. One of the first ones was to his detailed personal data he’s been gathering and his analysis of it using mathematica. While I don’t have as big an archive as he does I really understand the idea of collecting data on personal performance. I track books read, save nearly all e-mails, do periodic reviews of my projects and actions both active and completed in my GTD system and so on. I track body parameters (not just weight but things like measurements and physical parameters) as well.
His walking with an AR display and a chorded keyboard is something I’ve wanted for LambTracker and for many other tasks. My husband actually has some AR patents from back in the 1990s and there was an interesting project I worked on to implement a coded way to send messages without moving hands from the proper position for holding a firearm. Combining a chorded keyboard with some form of shorthand, similar to a court reporter could be a very fast effective way to retrieve data. Or alternatively combine it with cryptographic knowledge of letter frequencies, a controlled vocabulary and sentence structures. Heads Up Displays, AR glasses and other input devices would be highly useful on a farm. One I’ve always wanted, if I had a HUD display and a small ultrasound when I’m trying to extract a lamb that’s tangled or stuck I could see where in the ewe the lambs’ body parts are and more easily decide what head goes with what feet and how best to manipulate the lamb(s) to get them out alive without rupturing something in the ewe.
When I’m holding a sheep if I could use voice or via a tongue keyboard to do a LambTracker lookup I could have the data on the sheep, drugs given, lambing records etc in front of me in a HUD like display.Yes at one point we tested some tongue keyboards for various applications. They were based on a full plate denture like device. They worked well but were too expensive as they had to be fitted to each person by a skilled dentist. It took a while to type on them and they worked better for doing a limited set of predefined comands than general stuff.
Another cool thing would be a combined camera looking forward with an AR display that could take a picture of a sheep. Extract out the ear tag by doing enhancement of the number and then look up the sheep. Or when we lamb we paint numbers on the side of the ewe and her lambs and it would be faster to lookup by just looking at the sheep and asking who she is.
His walking keyboard is a lot better than the chute side debugging we do on Lambtracker now and I liked the shelf structure to hold it.
I too would love a note/drawing app for the iPad that allows the canvas to be a floatable window that moves around. I can see that being very useful and paper saving. Yesterday, when I was trying to explain some of the SQLIte Queriy questions to a friend while we were at the Pub having beer. I went through 6 pages of paper from my ever present little notebook. If I’d had a drawing note app on my iPad I could have easily drawn the notes, saved them and moved the pages and added pages easily. His solution of a document camera is a very workable one but I’d like a tablet based version.
I carry a 3x5 top spiral bound pad and space pen with me all the time in a wallet I designed. I’m currently carrying the first prototype of the wallet and hope to get a better pattern put out somewhee so other people can sew their own. My wallet and iPhone live in a belt pouch on my left hip. I make these for myself. They usually survive about a year to year and a half before the zipper gives out. I carry my phone in the same pouch and a few other things. it balances the leatherman on the right hip along with some additional EDC tools.
A head mounted camera that I can control by voice or a head movement (think the head tossing from Heinlein’s Starship Troopers that changes the various weapon displays the Cap Trooper could see) would be helpful. I have perfected the ability to deploy and take pictures with my Phone one handed very fast. That skill was important to capture on data both a dog that threatened our sheep but also the license plate of the car and the people who threatened me when I told them to get the dog under control. Both were important with my call to the sheriff to report the incident. (She who calls 911 first is the victim.)
The Wolfram language is very powerful. I initially really wanted to use it for Lambtracker because some of the complex things, Wards clustering algorithm to determine bloodlines based on genetic relationship for a large population, for example are single line codes. But it’s VERY processor intensive and requires more horsepower in the computer than the budget would allow for the LambTracker handheld.
I love the idea of his pBase for people and the connections and in particular the source and info on what people know. It’s very much like another project I worked on for DARPA where as part of validation of what you send via electronic means is evaluation of natural language and your abilities and you can become a trusted source and then your data is given higher priority in certain applications. I’ve developed my own pBase in my Farley file implementation within Apple Contacts but I can really see adding a lot more to it.
His library organization for books is also interesting. Particularly now that I’ve got so many more books as electronic books. My current library catalog consists of long lists of books sorted by author last name. Kindle books alone are over 1000 now and I struggle mightily against the total lack of any way to organize them into logical folders that can be hidden or seen like we do with opening a folder to see the files and then closing it. My paper book library is not nearly as large as his, around 2000 books and I sort them into large logical categories. I remember books by size, shape and author. For fiction I sort by author alphabetically and within the authpr by series and within series by the order in which the events happened. Non-fiction is sorted by category and within that by level of info, beginner or basic books, specific books, in depth detailed books and so on. I also have about 50 CD’s that are filled with old scanned historical books. Which reminds me, I need to make a project to extract those files off and get them into a more transportable digital format soon before I lose the ability to read CDs.
His info on the blood results and the changing “normals” reported by various labs and, sad to say, as the politics of certain health measurements comes into play is also very personally interesting.
The idea of using Wolfram language String Position to locate SNPs is rally interesting. We take blood samples from all the lambs whose genotype I do not know by pedigree for testing for the genotype at a single locus that is related to the susceptibility of the animal to developing scrapie. The prion disease similar to Mad Cow disease, Chronic Wasting Disease in cervids, FTSE, feline, transmissible spongiofornenchphalopty) in felines, kuru and all the other prion diseases. As we start to get SNP data on individual sheep using the new SNP chips there are massive amounts of data to store and use. I looked at adding that in this iteration of the LambTracker database but the volumes of data per animal far exceed the capacity of an SQLite database. The Ovine SNP50 for example produces a data set on 54,241 SNP probes across the entire genotype. There are variants at each probe and the datasets get huge really, really fast. One of those test is not cheap but also not too bad. Currently a SNP analysis of a sheep runs about $100 per animal. That cost should be reduced over time. I spend $11/lamb just to get info on a single codon and the testing is only looking for 2 variants of the 4 alleles at that loci.
His issues with image identification are not new and sadly no good ways to solve the problems yet.
What I came away with is a person who has a number of creative way to actually use the big data he’s been collecting all his life. Somewhat fitting for a person whose entire business is based on the manipulation of giant datasets.
Thanks for sharing the info.