Computer Bootcamp Day 1: Raspberry Pi

This weekend we are holding a two-day computer bootcamp for our kids, some friends of theirs and a handful of adults. Here is a recap of the first day both as a reminder and reference for those who participated and as an outline for anybody else who wants to try this.

We started by watching the first minute or so of a clip showing NASA mission control. I then explained that for the Apollo 11 program NASA had five mainframes from IBM (System 360 Model 75 with 0.034 MIPS and 1 MB of main memory each). We then found that the Raspberry Pis in front of us come with 512 MB of main memory and can perform at 100s of MIPS. Put differently we were holding in front of us a computer that was 100 times more powerful than the entire computing infrastructure for the moonshot!

We put the RPis in this great case (purchased as part of the great RPi starter kit from Adafruit) and connected one of them to keyboard, mouse, monitor and then power. Of course nothing happened. Why? Because we had no operating system installed. I got everyone to search for how to install an operating system for the RPi. We then proceeded by downloading Occidentalis 0.2 (based on Raspbian Wheezy, which in turn is a variant of Debian, which is a type of Linux). We used the Raspberry Pi SD Installer to flash the Occidentalis disk image onto a 4GB SD card. We did this from the command line using terminal in OS X and I explained that the name terminal came from the original dumb terminals.

Since we had a few minutes while the SD cards were being written we played a little game. Two students paired up and one had to get the other to draw a letter on the whiteboard by giving precise commands about line strokes. After that we discussed how programming or computational thinking is something that we all do to some degree every time we explain to someone else how to do something.

We now had working RPIs and booted each of them up connected to a monitor, keyboard and mouse. I walked everybody through how to configure the keyboard layout, timezone, etc. and soon enough we had them running and fired up the Scratch environment to play around a bit. We then proceeded to install a wifi card and had our first encounter with the GNU nano editor. We also learned about why we need to use sudo if we want to edit a file such as /etc/network/interfaces which holds the wifi configuration information.

We then learned how to use sudo ifconfig to display the IP address of our RPIs (and write them all down on a big white board). I then drew a diagram that showed how all of our addresses were local to the network we were on and not reachable directly from the internet. We used ping and traceroute from the command line to examine the network between our RPis and also computers on the internet. Since there were 10 students in the group in total each with their own RPi but only 5 monitor and keyboard setups it was now the perfect time to learn about ssh and how to use it to connect to a headless RPi. In fact we took several of the RPis and distributed them to different locations around the room to really demonstrate that we were remotely connecting into the machines. For the two students with Windows machines we used Putty.

Now it was time to do a little bit of programming. We again used nano to write a small program in Python that adds the numbers from 1 to 1000 (yes, there is a simple formula for that but the point was to illustrate how fast the machines could do this brute force). We then also wrote the same program in C and used gcc to compile it. We noted that our C program executed significantly faster than our Python one. I explained that what we read and write is the source code which needs to be translated into machine code which are the instructions executed by the computer.

To show that we can extend our RPis we then all put together a very simple breadboard containing an 8x8 three-color LED matrix. We downloaded the sample code for lighting up the LEDs using git. There were a lot of screams of excitement when the LEDs connected to a headless RPi started to flash.

The plan for tomorrow is to write a short program to fetch some value from the internet (eg the weather) and then make the LEDs light up accordingly. We will then also set up a website on our RPis and finally do the same using a cloud server.

Posted: 21st June 2014Comments
Tags:  raspberry pi computer boot camp learning

Shapeways: Make Things With Code!

Shapeways is a partner for Google’s brand new MadeWithCode initiative aimed at getting more girls excited about coding and computational thinking. What is important about this initiative is that it will broaden the understanding of what can be done with code. Code is not just games or websites. Increasingly it is everything and everywhere (even in a long piece on computational thinking in MotherJones).

So what I am particularly excited about is how this initiative will introduce a many more people to the idea of constructing physical objects through code. This is a key enabler for mass customization and for creating objects on demand (rather than having wasteful warehouses full of stuff that may wind up in a landfill). Shapeways makes this possible and even easy by providing a powerful 3D Printing API and the incredible ShapeJS Javascript Library.

The MadeWithCode bracelet project is built on top of these. Anyone can use the Shapeways API and ShapeJS to create their own customizable or fully code generated objects. This opens the door to amazing possibilities, such as the automated creation of a perfect fit lightweight arm or leg brace instead of a traditional cast. I can’t wait to see what people come up with!

Posted: 19th June 2014Comments
Tags:  shapeways google code api shapejs

Unbundling of the Job

Last week I posted a draft outline for my planned book on “The Coming Information Age" with a proposed chapter on the "Unbundling of the Job." This is an idea that first occurred to me when I attended a workshop at MIT on Work and Value in the Digital Economy at MIT in late 2012. At the time here is what I wrote:

Do people need jobs or can we deliver what jobs provide some other way and in a potentially unbundled fashion? The “jobs of a job” include income, structure, social connections, meaning, and at least in the US, access to healthcare.

Let me elaborate on this idea here. Whenever I mention the idea of a Basic Income as a way of addressing the impact of technology on the labor market someone will reply “but people have to work!” (usually in an exasperated voice). When you then ask “why?” you get one of the following:

1. Because people need structure in their lives. The benign interpretation of this is a genuine concern about people being bored if they don’t have work. It is more likely though a secular variant of “idle hands make the devil’s work” — a longstanding suspicion that people will be up to no good if they aren’t working.

2. Because people need companionship and communication. It is absolutely the case that historically work was where we spent most of our waking time and was therefore the fulcrum for companionship and communication outside and possibly ahead of the family.  

3. Because people need meaning and they get meaning from work. In the US it is common to ask people “what do you do?” to find out what kind of work they do. This is often followed (implicitly) by a conclusion about what kind of person they are based on their work.

4. Because people need healthcare. Thankfully with the Affordable Care Act we have started to unbundle healthcare from the job. By doing so we have moved healthcare into category #5 below. 

5. Because people need to pay for food, housing, etc. That last objection is of course what the Basic Income is designed to address. I will write more about the math of that in a world of technological deflation.

None of these are really about work qua work. Rationales 1-3 completely ignore that other activities and networks can also meet those needs. More fundamentally they reflect a view that people are not capable of truly living in freedom, where freedom includes choosing what to spend one’s time on. We have come to hold this limited and pessimistic view as a result of centuries of systems that relied on the control of (most) people’s time and effort first for agrarian and then for industrial production.

In the upcoming posts I will write about alternative ways to address these needs without a traditional job. As a quick preview: people are creative (and will be more so if we change our education system) and will find interesting things to spend their time on. And importantly in an unbundling based on a Basic Income there can and will still be a market for labor — it is just that people will be in a very different bargaining position.

Posted: 17th June 2014Comments
Tags:  work labor jobs unbundling technology

The Coming Information Age (Possible Book Outline)

As I am blogging my way through the various aspects of a transition to the Information Age I am still keeping in mind that eventually I would like to gather up all the material into a book. So here is possible outline for such a book. Would love to get feedback on what’s missing (or too much), different possible orders and anything else that comes to mind. So please fire away with comments!

1. Introduction / Motivation

— fundamental premise: industrial -> information is a transition as important as hunter gatherer -> agriculture and agricultural -> industrial

— why now? because we can see the beginning of the changes, examples: self driving cars, face recognition, machine translation

— changes build slowly at first but then accelerate, example: up to 1900 almost 50% of people still worked in agriculture

— prior transitions did not go very well and we now have more destructive power at our hands than ever before (nuclear, biological, etc)

— much of the policy debate is stuck in the industrial age and amounts to re-arranging the deck chairs on the Titanic

2. Industrial Age Recap

— brought us extraordinary material progress - examples: air conditioners, cars, medicine, life span

— not yet equally distributed but have been making great strides in eliminating poverty globally

— fundamental engine of the industrial age: consumer demand -> products made with capital and labor -> wages to labor -> more demand

— capitalism became the winning model in the industrial age because entrepreneurs created innovative products to meet consumer needs

3. Industrial Age Breaking Down

— substituting machines for labor, so wages declining and eventually going away and with them the demand for more stuff

— initially counteracted this challenge through consumer debt

— already producing more stuff than we need to meet our material needs (hence the rise advertising)

— research shows that stuff doesn’t make us happy, experiences do, biggest growth need is spiritual (not material)

— facing species threats including environment, asteroids, disease that need to be addressed at global (not national) level

4. Information Age Arriving

— first time in human history that everyone on planet is connected instantly at no (marginal) cost

— information is non-rival - we should share as much of it as possible (eg how to build an electric car)

— combined with exponential improvements in compute power and storage

— makes many things possible that were previously impossible, examples: collectively creating and editing an encyclopedia, teaching a machine to translate between languages, find cures for diseases

5. Fundamental Change 1: From Hierarchies to Networks

— increased information flow allows for more cooperation and more motivation

— individuals become peers in networks

— networks are resilient and global

6. Fundamental Change 2: Technological Deflation

— automation, on-demand manufacturing

— asset utilization

— rise of consumer surplus

— importance of the information commons

— attacking healthcare and education

— energy?

7. Fundamental Change 3: Unbundling of the Job

— separating of income, meaning, daily structure, health insurance

— guaranteed basic income in a deflationary world

— educating for motivation

8. Fundamental Change 4: Moving Past the Nation State

— the fall of artificial boundaries

— the rise of cities

— from big government to information standards

9. Threats Along the Way

— incumbents against the Internet (eg re-imposing geographic boundaries, overextending copyright) 

— breakdown of attention and money in politics

— digital balkans and lack of empathy

— surveillance versus open sharing

— income and wealth inequality in the transition

10. How to Prepare

— individuals: fight for the Internet, pursue passion

— parents: motivation is paramount

— educators: make connections

— companies: empower networks

— governments: embrace transparency

11. Outlook: Towards Abundance

Posted: 12th June 2014Comments

Raising Money to Help African Girls Stay in School (Our Daughter’s Fundraiser)

One of the things we have been emphasizing in homeschooling our children is the importance of giving back. So I am incredibly thrilled that our daughter has launched a Crowdrise fundraiser to support girls at schools in the Lewa Wildlife Conservancy. Lewa was one of the places we visited on our trip to Africa earlier this year and we spent a day visiting one of the schools.

The funds will help with various initiatives at the school aimed at keeping girls in school. At present the number of girls enrolled in school starts to decline rapidly after 5th grade. Katie is selling photographs she took in Africa (you can see some of her pictures on the Crowdrise page and more here) and original artwork she created. We will have a show of her work at our house next Wednesday (email me if you want an invitation).

If you are not interested in buying photographs or artwork but are interested in going for a sail, you can donate to Katie’s fundraiser through me. I will take up to five of the donors out for a sail on our J100 in the Long Island Sound.

Posted: 11th June 2014Comments
Tags:  crowdrise fundraiser Africa girls

Technology, Time Preference and the Return on Capital (Companies)

Just as a quick recap. I have argued in Computers and the Return on Capital that having cheaper information flows will in the long run drive the risk free rate of return to the time preference. I then examined how technology is likely reducing time preference for individuals through a variety of different mechanisms. In reply to that post Marc tweeted 

And yet tech companies keep raising giant rounds of funding and spitting off huge gushers of cash :-).

That would suggest tech companies need more capital (increased time preference) and that they are producing large returns on capital. The second part of this is I have already addressed, writing that in the short run (some) tech companies will produce huge returns on capital.

Today though I want to dig into the first half of Marc’s reply and examine whether companies generally (and tech cos specifically) are needing more capital and how technology is impacting that in the long run. Here is the TL;DR I will conclude that overall technology acts to reduce time preference for companies as well but that in the short term (some) tech cos are raising larger sums due to potential winner-takes-most dynamics.

As a starting point it is useful to remind ourselves why companies need external capital at all. This was the favorite interview question of one of the first people I worked for after college. Most people answered something along the lines of buying equipment or paying expenses. But the answer he was looking for was more precise: paying for stuff before being paid by customers!

That’s important because one way technology reduces the need for external capital is through the pre-purchase of products by customers as happens on crowd funding sites like Kickstarter and Indiegogo. One way to look at this is that the company is effectively borrowing from its customers (ie getting capital in the form of loans rather than as equity). That, however, is only a semantic point. There is no separate opportunity here for a return on capital for someone who just wants to invest without buying the product.

Now obviously a bunch of companies that have done successfully avoided the need of capital initially through pre-purchase campaigns have gone on to raise lots of traditional equity capital, most notably Oculus VR. I will get back to that point later as it is part of the perceived winner-take-most dynamic. Before that I want to cover several other ways in which technology is reducing the need for capital for companies.

Companies, just like individuals, have less need for capital as technology is making (nearly) everything cheaper. I am typing this on a Macbook Air which gives me a ridiculous amount of compute power and memory by historic standards and yet costs less than $1,000. As I wrote previously, companies can also increasingly substitute technology for labor which as a result of this pressure is also becoming cheaper (at least at the unskilled end of things).

And, in another parallel to individuals, companies too are benefiting from assets becoming shareable and rentable by the minute or even second. Most startups today for instance don’t buy their own hardware. They rent it as they need it, which again reduces the amount of capital that they require. Companies that achieve profitability early on can then rent additional machines out of current income instead of having to raise debt or equity. Our portfolio company Science Exchange is helping to bring this model to lab equipment which demonstrates that this trend is not isolated to computer servers.

If these are the trends, then why are some companies, and in particular tech companies as per Marc’s tweet, raising literally billions of dollars in equity? For starters of course part of the answer is “because they can” meaning that investors are offering this money up even if companies could do without it. To dig a bit deeper though, there is a sense that more markets will have some level of winner-take-all characteristic where the leading company will be 10x, 100x or even more valuable than anyone else. And if that’s the case it would be a great source of time preference, meaning if you can buy your way into that position by spending ahead of revenues it will still be worth it.

I do believe that this argument has merit due to network effects. It is still worthwhile though to remind oneself that some formidable network effects businesses have been built on relatively little capital (eg. Google raised only $25 million of venture capital) and that networks too can be disrupted. That can happen for instance when there is a technological shift or if the network operator is trying to keep too much of the economics of the network for itself (or its investors).

In summary then, I believe that technology largely acts to reduce the time preference of companies, ie reduces their need to raise external capital. The exception to this appear to be winner-take-most dynamics which to the extent that they are real and sustained will lead companies to want to spend significantly ahead of revenues (and investors to want to provide that capital).

Posted: 10th June 2014Comments
Tags:  capital technology time preference

Reset The Net - With Care

Today a broad campaign called Reset The Net has launched with the tagline “Don’t ask for Your privacy. Take it back.” Today also marks the one year anniversary of the Edward Snowden leaks that showed the extent of the NSA data collection and surveillance activities. Reset The Net is meant to be a citizen and corporate response that uses technological means to thwart the NSA and other agencies like it. The basics of Reset the Net are indisputably good practices. Sites should use SSL to avoid session highjacking and eavesdropping. People should know that there are tools for encrypting private communications as well as tools for anonymizing their access to the web.

There are, however, two important ways in which we need to be careful about Reset the Net. First, technological means are not a substitute for political change including comprehensive reform of the NSA. As the heartbleed vulnerability showed and as Quinn Norton describes extensively in her post Everything is Broken, the complexity of existing systems is such that we cannot and should not rely on technology alone. There is a good chance that whatever system you are using has some kind of backdoor or exploitable flaw in it (and if it doesn’t today it might tomorrow as part of an upgrade).

Second, it would be a tragic loss if more and more of the net migrated into private, ephemeral and encrypted communications among small groups. Much of the net’s power to move us forward as a species comes from open collaboration and the potential of collective intelligence. Whether it is discovering great content from unknown writers or using patterns of queries to identify public health issues there are a myriad of ways we can all be better off when information is openly accessible. That is the fundamental promise of a network that connects every single human being on the planet and we should not lose sight of it.  


Posted: 5th June 2014Comments
Tags:  reset the net surveillance collective intelligence

The End of Categories

We like to put things in categories: public versus private company; for profit versus charity; professional versus amateur; creator versus consumer; journalist versus blogger; teacher versus student; and so much more. Of course not all categorizations are binary. In academia, for instance, we have the faculties (e.g., chemistry, biology, physics) and degrees (Bachelor, Master, Doctor). 

Categories are a way of aggregating information. If I tell you that a company is public that is a shorthand for communicating a lot of information about the company such that it is likely to have audited financials, file quarterly reports, shares that are traded on an exchange, and so on. At a time when communicating information was costly it made a lot of sense to organize the world around categories. 

Now, however, with the marginal costs of computation and communication nearly zero, we can reveal the underlying information and replace categories with continuums. For instance, companies can publish as much or as little of their financials and operating data as they want. Third parties can then publish scores on the basis of that information, including a transparency score, which replaces the previous binary breakdown between public and private companies. 

Similarly instead of using thee-tiered academic degrees as categories for how much someone has (supposedly) learned, we can publish our papers, analyses, projects, test results etc. That again allows for third parties to evaluate how much knowledge or mastery of a particular subject we have achieved on a fine grained scale.

Because we as humans still need to process and store some of this information in our heads we may super-impose existing or new categories on these scores. For instance, we might call someone a novice or an expert. But the more information is ultimately accessible the less these categories will be used by themselves for decision making.

As an important corollary we should rethink a lot of regulation to move away from categories established and maintained by regulators and towards standards of information publication. If we did this right it would let us get rid of the often arbitrary distinctions that exist today and result in many distortions. This would have the added benefit of vastly simplifying regulation.

Posted: 4th June 2014Comments
Tags:  categories information regulation

Technology, Time Preference and the Return on Capital (Individuals)

In my post on Computers and the Return on Capital I made the point that cheaper and more ubiquitous flows of information will drive the risk free return on capital towards the time preference (assuming that capital can also flow freely). But I didn’t really explain what that meant or how it was determined. Let’s start with an extreme case to illustrate the basic idea. Suppose that everyone everywhere was happy with how much they can spend out of their current income (that includes individuals, companies, governments). In that case the risk free return on capital would be zero as nobody would be interested in having debt or raising equity capital.

For there to be a positive market clearing rate of return on capital it has to be the case that some people or companies or governments want to spend faster today than their income allows and so need to attract capital from those who have capital to lend or invest. Time preference is a catch all for capturing this desire to spend now rather than in the future. The relationship between technology and time preference is complex but I think there is fairly strong reason to believe that over time technology is reducing time preference.

Let’s start with individuals. The major purchases for which individuals tend to borrow include education (biggest outstanding debt in the US today, greater than credit card debt), cars, homes and consumption more generally.  We are seeing the beginnings of technology being used to dramatically lower the cost of education, in the extreme by making it available for free as in the case of Khan Academy, EdX, etc. So one way technology can reduce the need for capital is simply by making something a lot less costly. But there is another effect at work here. If I need to get all my education up front I may have to borrow, but if we are moving to lifelong and on-demand education I may be able to pay for it out of current income.

Another way that technology can reduce the need for capital is by making assets shareable. That’s what we are seeing with cars which used to be incredibly inefficiently used assets (standing around idly 95% of the time). Whether it is through car sharing, ride sharing or ultimately driverless cars, technology can help shift from an ownership model (which may require capital) to an on-demand model which can again be paid for out of current income.

As far as general consumption is concerned, technology has been driving down the cost of many products so that more people can purchase them out of current income. Computers themselves are of course a great example of that with smartphones being a computer that almost everyone can afford now (not just here in the US but globally). Known deflation has another impact on time preference: unless you have an immediate need you are more likely to wait as products are getting cheaper and better over time.

The one place where for the time being technology has not yet had an impact is real estate. Many people borrow to buy a home and with urbanization still increasing the cost of homes has been on the rise. There is a great infographic showing the multiple of annual earnings required to buy a home in different parts of the UK with a breakout for London. Technology won’t solve this problem in the near term but (unlike many others) I suspect that we are close to peak urbanization and will in fact see people spreading out more in the future. Probably a subject for another post!

A final impact of technology that is relevant to individual time preference is longer expected lifespan. My desire to get something today rather than say next year or the year after is likely to be lower when I feel I have more time. None of this is saying much though about individual preferences. There will likely always be people who are patient and others who are not.

In an upcoming post I will consider the likely impact of technology on time preference for companies. There too I believe that technology is largely acting to reduce time preference in the long run. So we might get all excited about a future of lower returns on capital and how this might reduce the impact of wealth inequality. That, however, would neglect the potential for external shocks. The one I am thinking of in particular is climate change. There is a scenario in which a lot of capital is required very quickly to deal with say rising sea levels. Time preference would go way up, eg we need a sea wall now, not over the next hundred years. If that were to happen there would be a spike in the returns to capital *if* this is financed by borrowing and investment in privately owned companies and infrastructure.

Posted: 2nd June 2014Comments
Tags:  capital technology time preference

Consumer Surplus and Returns to Capital

Yesterday’s post on the impact of computers on the return of capital didn’t mention the shift from producer to consumer surplus. Tren Griffin rightly called this out on Twitter. I have written a lot about the rising importance of consumer surplus and should have mentioned it in this context as well. This shift will definitely reduce the return on capital in the long run as we move more information into the commons and as information becomes more important than physical capital or output. It is a good reminder that I also wanted to write about the role of information as a factor of production (and as an output).

In the near term though this very same shift is behind some massive returns to capital. The best way to rapidly build a large and defensible information network is by leaving a lot of the surplus to consumers. Making much of a service free speeds adoption and also makes it more difficult for competitors to come in. Even when lightly monetized, producer surplus in aggregate can still be a huge number as these networks are global. Facebook is a great example of that. And so the returns on capital can be staggering as the VCs who invested in Facebook early have proven (ditto for Twitter, Tumblr and many of the other services that Tren mentions — just because a service was sold early doesn’t change this argument).

In the extreme when everything becomes consumer surplus (eg Wikipedia or a future completely decentralized messaging service) this will no longer be true. But we are not quite there yet. Or to sum it all up: the shift to consumer surplus further strengthens my argument that we are seeing great returns to capital in the near term but that those will decline significantly in the long run.

In the same Twitter thread, Marc was skeptical about the long run decline in returns on capital. As I will write in a follow up post it is possible that something like climate change causes a shock to time preference. Meaning: we will suddenly want to spend massively to avoid imminent disaster which could hugely spike the returns to capital. More to come.

Posted: 28th May 2014Comments
Tags:  capital returns consumer surplus

Newer posts

Older posts