Graham Glass, etc.

EDU 2.0 for Mobile Devices

We've just started work on EDU 2.0 for mobile devices, and we're going to target the iPhone in version 1. I'm not sure whether it's going to be native or browser-based; PhoneGap might be a way to get the best of both worlds.

The first version will support four main areas of the site - classes, assignments, messages and calendar. Here's a sketch:

• "Classes" will list the classes you're teaching and taking.
• "Assignments" will show your due and past due assignments.
• "Messages" shows your inbox.
• "Calendar" shows today's calendar with a weekly view as well.

It's not clear how much data users will want to enter from a mobile device. For example, I can imagine a student browsing a lesson or taking a quiz, but it's hard to imagine a teacher creating a lesson or a student submitting a freeform essay question.

Once we get version 1 out, we'll get plenty of feedback on which additional areas of the site are most visited from a mobile device.

Code Stats Update: EDU 2.0

The last month has been particularly busy on the EDU 2.0 coding front. We decided to spend 2 weeks of the month doing some major behind-the-scenes re-architecture, especially on the database schema. Then we added a ton of new features, including team assignments, rubric banks, proficiencies, portfolios, question banks, proficiency banks, improved parent support, and a prototype e-commerce integration for the business version.

Thankfully, the amount of code only went up by 1,500 lines and the methods per class actually went down from 7 lines to 6.

Here's the summary:

32,000 -> 33,500 lines of code (+5%)
606 -> 656 classes (+8%)
4286 -> 4579 methods (+7%)
7 -> 6 methods per class on average (-14%)
5 -> 5 lines of code per method on average (+0%)

Although the amount of coding and re-architecture has been quite tiring and stressful, I think it's going to pay off quite quickly, especially when we release the business version of EDU 2.0 in September.

Sleep Simulator, Part 6

I thought it would be interesting and instructive to post some of the interim results as I tweak the algorithms to try and match a human sleep cycle.

This version goes to sleep too early and doesn't have the kind of cutoff I'm looking for. Ideally, the simulated brain would stay awake until the last moment and then suddenly go into deep sleep.

In addition, the simulated brain gets too rested in the early hours of dawn and completely wakes up at 2am and 4am.

I could make the simulation work better by increasing the weight of the daylight component, but that feels like faking it to me; I don't think that outside stimulation has that great an effect on, since our "inner world" can generate plenty of stimulation regardless of how much external stimulation is going on.

Nevertheless, I'm happy with progress so far since I am only spending a few hours a week on the simulation. I hope to post a better graph sometime next week!

LASIK Update: Two months later

Overall, I'm very happy with the results of my LASIK eye surgery. I went in for my checkup today and I have better than 20/20 vision in both eyes!. My vision is very stable now, and things always look crisp and sharp.

The only negative side-effect (which is temporary) is that I became very sensitive to light about 2 weeks ago. It got bad quite quickly, and I have to wear sunglasses even when typing at my computer terminal which is already turned down to minimum brightness. I've light-proofed all the windows in my bedroom with aluminum foil to keep the room dark. 

I particularly noticed how sensitive my eyes had become when recording my latest video blog entry. My eyes were squinting noticeably under the studio lights and it was harder to concentrate when making the video.

Apparently, the reason for this is that the healing process in my cornea is accelerating and this can sometimes trigger light sensitivity. So I was given some eye drops that I have to use once an hour for the next week or so. These drops slow down the healing a bit which in turn should eliminate the sensitivity.

I'll blog again in a few weeks and let you know how it's going!

Sleep Simulator, Part 5

Here's the next version of the simulation. In this version, the probability of a neuron being awake is based on the amount of waste it has, the awake state of the neurons it's connected to, and the amount of daylight.

The next step is to modulate the red line with the 90-minute recurring cycle (blue line). My hope is that the cycle will push the neurons into an almost-awake state every 90 minutes which will correspond to the REM sleep phase, and then allow the neurons to go back to a mostly-asleep state. 

Sleep Simulator, Part 4

To speed things up (because the simulator is not that fast), I'm focusing entirely on the various parameters during the 11pm to 7am period. I'm trying to figure out the equations that result in the sleep activity that we measure in humans, such as the following:

and this:

My hope is the my simulator will be able to reproduce these patterns using equations based on the amount of metabolic waste, the amount of daylight, the state of the 90-minute cycle, and the number of awake neurons.

I'm starting with the simplest equations and then will start to layer other factors on top to modulate them. For example, here's a graph where the probability of being awake is simply proportionate to the amount of metabolic waste:

The next step is to also modulate whether a neuron is awake based on whether the neurons it's connected to are awake. This will hopefully force sleeping neurons to stay asleep longer in the early phase of sleep.

Sleep Simulator, Part 3

The next stage was add a super-simple "resting" phase algorithm. Because I'm developing the software iteratively and incrementally, I just decided to put a neuron into a rest phase when its waste became bigger than a max threshold and then exit the rest phase when the waste became zero. 

In real life, this would be useless, because it would mean that you could not wake up during the night for any reason, even if your house was burning down!

The next phase is to improve the rest phase algorithm to take into account the activity of a cell's neighbors. That way, a cell can wake up if its neighbors are also awake, allowing a brain to quickly awaken if necessary even if the cells have a lot of accumulated metabolic waste.

Here's a plot from phase 3, with the number of awake neurons shown in red.

Sleep Simulator, Part 2

The next step in the sleep simulator is to tweak the axes to start at 7am (I'm assuming that the simulated brain will wake up then) and also to track the average amount of metabolic waste in each neuron.

Here's the chart that the simulator produced, with the metabolic waste showing up in green. Note that since I haven't added the concept of a neuron going into a rest state yet, the waste simply accumulates and passes a threshold at around 12pm.

The next iteration of the simulator will add a simple rest state for neurons.

Sleep Simulator, Part 1

I've had a theory about why we sleep and dream for many years, and recently recorded a video that explained my theory. I decided that the next step is to write a software simulation of the relevant aspects of a brain that will hopefully exhibit the various sleep patterns based purely on a few simple equations related to cell metabolism.

I've written most of the simulation already, with the main actors - Simulator, Brain and Neuron - written in Ruby and the web interface written using Rails. I use ProtoChart to display the results of the simulation in real time.

The simulation is of a 50x50x50 "cube" of neurons (125,000 total) that are locally interconnected. Each neuron accumulates metabolic waste when it's awake and disposes of it when it's asleep. The probability of a neuron switching between states is based on its metabolic waste levels and the state of the neurons it's connected to.

I'm also postulating (and there is some anecdotal evidence for this) that there's a continuous 90-minute cycle throughout the day that stimulates neurons according to the position in the cycle. In addition, I'm simulating light levels because they are an external source of stimulation.

Here's a graph that only shows the daylight levels (yellow) and the 90 minute cycle (blue). Next week I'm going to incorporate neuron activity and hopefully reproduce the typical sleep/REM cycles from basic principles. I'm keeping my fingers crossed!

p.s. I will open source the simulation if anyone is interested.

Ecademy California to adopt the EDU2.0 education platform

Here's some cool news!
 

EDU 2.0 is pleased to announce a partnership with Ecademy California in which Ecadamy will adopt the EDU 2.0 platform for its virtual K-8 school. EDU 2.0 and Ecademy California are passionate about the future of online schooling, and together they aim to vigorously prove the effectiveness of virtual schooling.

"EDU 2.0 has thought of everything with its robust and student friendly tools. Our entire school community can use EDU 2.0 to share work, ideas, and personal information. This fits perfectly with our mission to bring public school home to students - creating a community of learners that is entirely web-based but relationally centered." said David Damico, executive director of Ecademy. Most virtual schools text dump - using web tools to deliver work that students complete and submit online. Ecademy brings teachers and students together in a content rich, project based learning environment. EDU 2.0 gives us a way to bring the academic and social components of student life together in a safe, robust virtual environment.

"Ecademy is passionate about virtual schooling and we are passionate about providing the best platform for delivering virtual education, so it's a perfect fit" said Graham Glass, founder of EDU 2.0. "We look forward to a fruitful alliance, and to providing whatever additional features Ecademy needs to further its mission".

About the companies:

EDU 2.0 provides a free and easy-to-use web-hosted learning management system for schools. Founded in 2006 by Graham Glass, EDU 2.0 currently has 70,000 members and is growing at 10% a month. It is used by a wide range of schools, from home schoolers to Universities with thousands of students. Visit the site at http://www.edu20.org .

Ecademy is a K-8 virtual online public school serving students in San Diego, Imperial, Orange and Riverside Counties. It is completely tuition-free and designed to give parents and their children access to the education they deserve in a personalized,  flexible, and fully supported model. Visit the website at http://www.ecademyca.org or email: [email protected]