Why would anyone do this? I think the reasons people invest such insane amounts of time in such a specialized skill are shifting. And I think that shift is healthy. It’s a shift in what it means to be a technologist. And the culture of our technical communities is shifting with it.

Back then
I learned to program in high school, early 90s. Looking back, I think my formative experiences as a technologist were pretty typical for my generation of programmers. I had three or four close friends in high school who also liked computers. They were all male. This was the dawn of the internet, around the time of the very first web browsers, and long before the first tech bubble made geeks into rich visionaries. We were not remotely cool. Technical information was somewhat harder to come by than today, but not excessively so. My first C++ compiler shipped in a big box which included a thick paper reference manual and a really nice language tutorial. We subscribed to Byte and Doctor Dobbs’ Journal. We hacked on stuff at lunch time and after school and weekends, and traded shareware on floppies. The technology was different, but the substance of the experience was much the same as today. We spent a lot of time at the computer, and we were well-connected into a community of like minded people. The community provided technical help but also motivation and inspiration.

We weren’t trying to change the world.  We were driven by an intense curiosity about the inner workings of machines, and we wanted to be admired for being good at something. I wrote the Windows port of Netrek, one of the very first multiplayer online games, and the local geeks knew who I was when I arrived at the University of Toronto. This kind of experience persisted through my undergraduate years studying computer science. Long nights in the computer lab; cool hacks. There’s a wonderful book which captures this culture as it evolved starting in the late 1950s.

Enter women
There were no women in the communities where I learned to program. Or, almost none. I did a head count in one of my classes: four out of 150 students. Sadly, this kind of ratio persists today in many technical fields. I didn’t really know why this was. Us nerdy boys would have welcomed geeky girls. For all sorts of the right and wrong reasons.

It’s only in the last few years that I’ve started to understand why the dominant nerd culture drove women away in droves. Simply put: it was a club full of very poorly socialized boys, and our peer-based motivation was all about status. We all wanted to be the alpha geek. We would jump all over each other to point out errors. We would never miss a chance to demonstrate our superior, elegant technical minds. We were completely insufferable to anyone else.

Fortunately, there are now more women in tech. And they’re starting to tell their tale. While I don’t want to generalize too much from the experiences of a single person, I found the account of Rebekah Cox to be really enlightening (there are lots more great stories in the same thread):

So, if you enter this environment as a woman without any sort of agenda or understanding of this culture the first thing you find is that if you actually say something the most likely reaction is for a guy to verbally hit you directly in the face. To the guys this is perfectly normal, expected and encouraged behavior but to women this is completely out of nowhere and extremely discouraging.

As a technical woman, this is your introduction and the first thing you have to learn is how to get back up and walk right back into a situation where the likelihood of getting punished for participating is one. How you choose to react to this determines the rest of your career in technology.

Now, I don’t want to give the wrong impression. It wasn’t all one-upmanship and verbal assaults. These geek scenes could also be wonderfully supportive, and often served as social groups too. You have to remember that this was before computers were cool, and it was an awkward adolescence when you were interested in things you couldn’t begin explain to anyone else. Also, it was a great learning environment. Cox again:

Even the aforementioned nerd trash talk is actually a useful tool that can help you. The reason that culture exists is to make everyone in the group better. The fact that you are getting hit in the face means that someone is either wrong and you can hit back with a correct answer or that you are wrong and someone is letting you know that directly. Sticking that out means you are learning in an accelerated environment with instant correction.

Furthermore, if you stick around long enough, you can find people who aren’t completely insecure and are confident enough to not resort to insults to assert themselves. Those people make the tough environment actually tolerable. If you can help each other then you can establish a safer zone to talk through ideas. And since those more secure people are typically so secure because they are really, really good, you can find yourself in an informational jet-stream.

In this artificial high-pressure environment we got good fast. But it was certainly off-putting to women, and not just women. Lots and lots of people wanted no part of this, and for good reason. Yet for quite a long time it was these sorts of socially dysfunctional communities that produced the lion’s share of the best technologists.

Why program?
Learning to program is still ridiculously hard, and still requires a community of practice. And it still requires an absurd focus and motivation. But the sources of that motivation are broadening. I’ve been watching this shift for a while. The notion of programming for the social good has even crystalized into institutions such as Random Hacks of Kindness (for international development), Hacks/Hackers (for journalists), and Code for Amercia (civic platforms.) For that matter, there’s Wikipedia. There are services and data all over the web. We don’t have to wonder whether software can change the world — it already has!

So by my old-school standards, the burgeoning hackers of today are very applied. I grew up desperately curious about the insides of things. Many of the programmers getting started now are far more extroverted than that. Here’s MIT Media Lab fellow Lisa Williams:

I want to learn to code because a lot of things piss me off.  

I believe a program can stand in opposition to Things That Suck, just like a documentary, a work of art, or a protest march.

…

I wanna code because SHIT IS BROKEN.  I want to code because corruption is real, because people are getting thrown out of their houses, because veterans aren’t getting what they deserve, because racism is real and has real effects, because yes it does matter when you cancel a bus line, because it’s really hard to shut a computer program up, because you can’t say it’s an isolated incident when there’s a bigass Google Map in your face showing you it’s not.

This is Lisa demanding “computational journalism.” But pretty much every field of human endeavor uses lots and lots of software now. Software not only determines what is possible, in many ways, but what is not possible: code-as-law. It’s part of the system, and if you want to hack the system, well, at some point someone has to hack the code. That person could be you.

Today
At the Online News Association conference last week, I ran into Michelle Minkoff and Heather Billings standing in front of couple dozen enthusiastic young journalists who had gathered in the hallway to hear about programming. Michelle works with me in the Interactive department at the AP, while Heather just started at the Chicago Tribune. Both are fearsome technologists, though I don’t think either would be offended if I said they are still near the beginning of their journey. That makes them the perfect people to talk to about learning to program.

Most of the people attending had some programming experience, but not much. There were 24 people listening to Michelle and Heather, 9 of whom were female. A great improvement. I sat in on this conversation for a while. It wasn’t what I was expecting. No code. Very little technical discussion at all actually.  One woman said she knew enough Python to write a Hangman game. “Great!” said Michelle. “You’re ready to learn Django!”

I guess I’m surprised anyone has to be told that they are ready to learn to program. But inclusion and connection was a major theme in the discussion. Here are some of the snippets of conversation I wrote down:

“You can make an anonymous account on StackOverflow and ask stupid questions.”

“Connect in person, build that mentor relationship.”

“But that documentation is for developers!”

This was a group of people who needed to be told that they could learn to program. That they could be one of them. This is understandable. When you can’t begin decipher the supposed “instructions,” technology can seem like an occult priesthood. But you don’t need them. You just need to want to do it, really badly, and you need to find some other people who want to do it badly too (and obviously, expect to meet these people online.) Then one of them becomes one of us. Of course you can learn to program. It just takes a while, and a stupid amount of practice.

In fact it’s probably necessary to devote a few years of your life to it full time. That’s one of the advantages of a computer science degree — time to focus. Also, a CS degree is a fast track to the deep theory of computation; if you find yourself looking at programming languages and asking why they are the way they are, or staring hungrily across the awesome gap between your web apps and a search engine, you probably want to learn computer science, and formal education is one way to do that. But CS theory won’t make you a programmer. Only programming will do that.

Every truly good programmer I have known had some period of their life where they thought of nothing but code. Something around a year or two. It’s got to get under your skin at some point. I call this the hacker gestation period. You’ll know you’ve reached the other side of it, because software will stop being mysterious. Eventually code becomes clay.

And this formative period is why it’s so important to have a community. You’re going to need friends who are interested in talking about geeky stuff. You’ll be so excited about it for a while that you won’t be able to talk about much else. (Really. If this is not the case, go do something else. Programming takes so much soul that you’re going to hate your life if you don’t genuinely enjoy it.) Your community will help you when you get stuck, and they will help you develop your sense of style. Code is the most obscure art, because only another programmer can see all the layers of beauty in a truly lovely piece of code. But it’s very hard to become an artist alone, without influences and critics.

So it takes a village to make a programmer. I won’t say that our technical villages are now inhabited by “normal” people, by any stretch of the imagination, but the communities where programmers are now growing up seem far more diverse, supportive, and extroverted than in years past.

Dutch journalist/coder Stijn Debrouwere has written a very thorough post describing the ways in which standard tags, like the ones on this blog or on Flickr, fall short when applied to news articles. There are lots of things we might like to know about a story, such as where and when it happened and who was involved. This additional information, sort of like the index to a book, is known as “metadata”, and there is within the online journalism community a great call for its use, including by Debrouwere:

Each story could function as part of a web of knowledge around a certain topic, but it doesn’t.

So here’s a well-intentioned idea you’ve heard before: journalists should start tagging. Jay Rosen insists that “Getting disciplined and strategic about tagging” may be one way professional journalism separates itself from the flood of cheap content online.” Tags can show how a news article relates to broader themes and topics. Just the ticket.

News metadata is a major topic, and many people have speculated deeply about the value of creating news metadata at the time of reporting, such as the ever-sarcastic Xark and the thoughtful Martin Belam who writes about why “linked data” is good for journalism. But I’m going to respond to Debrouwere because I read him today, because he has lovely diagrams that explain his good ideas, and because, in criticizing “tags” as a form of metadata, I think he misses some very important points.

And he’s not alone. My sense is that many of the coder-journalists of today have not learned from the mistakes of generations of technically-minded people who wished to talk about the world in more precise ways.

Moving forward from simple tagging, Debrouwere imagines more sophisticated annotation schemes that start to pick up on what the tags actually mean. For starters, the tags could be drawn from separate “vocabularies.” Does a tag refer to a person, or a place, or perhaps an event? Debrouwere uses the following picture, which I’m going to borrow here because it explains the idea so nicely:

But, he says, we can get even more sophisticated. What did the story actually say? If it mentioned a person, what did it say about them? Was it an interview? A profile? Did it criticize them? Here’s the diagram he draws:

He imagines using this information to perform chains of inferences, like so:

Barack Obama belongs to the Democratic Party and he’s from Chicago. If we tag an article with Barack Obama, it’s likely that the article also has something to do with the Democratic Party. If we’ve specified that the article is about Obama, and we’ve specified that Obama is part of the DP, the system now has all the necessary information to suggest our article about Obama as a possibly interesting related read on the topical page for the democratic party, even if we didn’t explicitly indicate that link.

First of all, note that this sort of thing is already possible, quite often, using tags as they exist today. Simple analysis of co-tagging information will tell us that Obama is related to the Democratic party, because many articles will be tagged with both. Which is not to say that encoding such relationships explicitly isn’t a good idea. We can do this sort of thing using “triples,” which are fundamental to the nascent evolution of the internet into a web of “linked data”:

• <Barack Obama> belongs-to-party <Democratic Party>

Here, “Barack Obama” is an object from, say, the “people” vocabulary, and “Democratic Party” is from, perhaps, the “political party” vocabulary, or maybe just from “groups.” Essentially, these are tags that have been pre-categorized. The relationship between the two is expressed by the “belongs-to-party” predicate.

But I argue that this is a rigged example. The world is normally much more messy.

“Are you now or have you ever been a member of the communist party?” was a killer question in its day, with complex answers like “I only attended one meeting.” And if parsing politician’s statements was easy, then Politifact wouldn’t devote entire articles to the question of whether a single sentence was true or false. Further, they distinguish between different “grades” of truth, like “mostly true.” Mathematical logic — which is what the sort of news inferences that Debrouwere and others discuss is based on — doesn’t deal with “mostly true.”

The problem is that the world is not neatly categorizable.

Don’t get me wrong — vocabularies and relationships (ala linked data triples) are surely a good idea. But they have some serious drawbacks that relate to very deep issues in knowledge representation.

Debrouwere says, “Events happen at a certain place and at a certain time.” Sometimes. For a house fire or a shooting, maybe, but how “long” were the post-election protests in Iran last summer? They continued at varying intensity for several days, then flared up weeks later. Was that one protest or two? And what about a Facebook protest that gathered supporters over the course of a week? “When” and “where” did that happen?

Or, take the example of describing what an article says about someone. How do we decide when a story “criticizes” someone? There will always be boundary cases — lots of them in professional reporting. How do we ensure inter-rater reliability? Can we extract any real data from analyses of this tag if we have no other reference points with which to interpret it?

Something is always lost in categorization. That is the point! To say that two things are like one another is to ignore their differences, for the purposes of the present discussion. Unfortunately, what can safely be ignored depends on the discussion. Simple date and place notations work for some purposes, and fail miserably for others. They are not very rich, and even worse, we don’t know exactly how much has been lost in each case. Knowledge of that error is sometimes critical, especially when trying to make chains of inferences, where errors multiply.

The reason we use text for reporting is that it’s good at representing these sorts of ambiguities. Strict adherence to the religion of finite relationship vocabularies leads one to believe that the world can be modeled in first-order logic (predicate logic), and this just isn’t true. Chains of automatic inference fail very quickly when applied even to very simple “real world” situations. The Artificial Intelligence research community went down that road for decades and found it really problematic, which is why we’re now seeing the rise of “statistical” AI techniques, such as statistical machine translation. This approach tries to find patterns in vast amounts of data rather than working out hard underlying rules; the categorization comes after you look at all available data, not before.

And therein lies the great virtue of tags: they are just about the simplest possible way of saying something, and don’t imply or require any particular inferential framework. They’re much harder to get wrong than more complex associations, and they make sense only in aggregate, and this makes them much more robust than predicate sentences. A tag says only, “there’s some association.” Full stop. I find this ambiguity a virtue. The meaning comes out of the relationships between the tags, articles, and users. Meaning is always relative, and tags force us to understand this, because there’s nothing else to go on.

Tags allow (or force) what we might call the “Google solution”: let humans describe it in a way that makes sense to them, then sort it all out later algorithmically. There are limits to this, of course, which is why metadata has value. But ultimately, computers serve humans, so the Google solution will always be a win when it is possible.

Linked data will be valuable because of the links. I predict that its main use will be as a sort of “super tagging” system: we still have “tags” in the linked data world,  it’s just that they’re now all “uniform resource identifiers” that are visible to everyone on the web. This means that tags can be shared between systems and maintained by communities, which only makes them more powerful. In fact, this is exactly what we’re already seeing, with the Wikipedia-derived DBPepdia at the center of all those linked data “bubble diagrams.”

Linked data also supports predicates that  say what the relationship between the tags is, like the “Barack Obama is a member of the Democratic Party” example. But I predict that these will  be much less useful, offering almost none of the “machine understanding” that’s supposed to come with the semantic web. I don’t know what “understanding” means if not the ability to draw inferences of some sort, and predicates are just too fragile, too subject to mis-categorization, too limited to capture the rich relationships of the real world. I do believe that we’ll see amazing new “artificial intelligence”-like applications built on top of linked data, but they’ll be built statistically: they’ll ignore the predicates or use them only in special cases, or only in aggregate.

Having said all this, I am fully in support of adding better metadata to news stories. I believe the  ”entity recognition” performed by OpenCalais is valuable, and that carefully managed tag vocabularies are essential. Often “location” will be a genuinely useful tag, and I can see the possibility for some wonderful news mashups.

But please, let’s not imagine that we can capture even the “essential” details of real journalism with any fixed vocabulary. And let’s not oversell the potential of machine reasoning or data-mining based on carefully-annotated news metadata.

We’re a very long way from understanding how to represent reality in machine-readable form.

For more on this topic, I recommend:

• “Ontology is overrated” by Clay Shirky
• “Metacrap” by Cory Doctorow
• “What is a knowledge representation?” by Davis et al. at MIT

If you didn’t follow that, consider yourself fortunate. You’ve never needed to wonder about such things.

It gets better. When I reset my phone it lost the WiFi password to my home network. I couldn’t find it written down. I couldn’t remember the password to log into my router to look it up. The internet told me how to reset the router at the hardware level, but to reconfigure the wireless I’d need to connect my laptop to it with a cable. Which I didn’t have. Luckily, I eventually remembered the router password.

I started drinking.

Password problem solved, every ten minutes I’d open the freezer door, reset the WiFi on my phone, wait for Cydia to download its package list, then tell it to download the mere 50kb of Ultrasn0w and hope to hell the radio didn’t blink out in the middle of the tiny transfer. Now I know exactly how many bars I get in the back of the freezer.

After eight or nine tries, I opened the freezer door to find my phone on the 3G network. Success!

Actually, it was way more involved than this. I left out a bunch of steps, all the things I tried that didn’t work. And of course the firmware upgrade did not fix the WiFi, so this experiment put me right back were I started and wasted six hours of my life and two tumblers of rather nice whiskey. At least I didn’t have to go out of my way to retrieve the ice.

This video shows NearestWiki, which tags nearby landmarks/objects and guides you to them. I am aware of a few other AR apps, as this post on Mashable and this AP story discuss. Many of these apps do building/object recognition, and one even recognizes faces and displays a sort of business card. We’re already seeing annotation with data from Wikipedia, Twitter and Yelp, and I suspect that we’re going to see these tools get very deep in the very near future, with Wikipedia-style tagging of  the entire history and context of any object.

Just a moment while I get over the fact that the future is already here.

Ok, I’m properly jaded again. Yeah, it’s an app platform, and that’s cool — but imagine the possibilities for art. Bets on who’s going to make the first “alternate reality spyglass” piece? Bets on how much Matthew Barney will sell it for in the app store?

Theoretically, desktop manufacturing technology then spreads exponentially, until everyone can make whatever material objects they need from downloaded plans, for only the cost of feed plastic.

The dream is best explained in this excellent little video:

It’s hard to overstate the fundamental shift that would come with truly widespread desktop manufacturing. Right now all of the objects we use are manufactured somewhere far away and shipped to us, and the designs are expensive and slow to  change. Instead, imagine if everyone had a household appliance, perhaps fed by spools of plastic and metal wire, that could manufacture just about any object from plans downloaded from the internet. It’s hard to see how private designs could compete with millions of amateur object designers geeking out over their widgets for the benefit of humanity, which means that designs for all the basic desirable objects would be freely available.

Want a new phone? Download the latest Android phone plan from the Open Handset Foundation. That’s cool, but the really cool thing is this: everyone in the world could have one for the price of plastic. More to the point, everyone in the world could have e.g. irrigation pumps, car parts, light switches, medical devices, essentially all the trappings of modern technology.

It is of course debatable whether or not an increase in humanity’s use of energy-consuming technology is a good idea at this time. However, it seems to me unconscionable to deny it to the world’s poor just because we got there first. Further, one could also replicate the parts for home biomass reactors, electric cars, and other advanced energy devices — regardless of whether or not anyone can make a profit selling such items commercially.

New versions of the replicator with enhanced production capabilities (now with integrated circuits!) would be designed to be manufacturable using existing models. This means that manufacturing technology would itself spread virally. To bootstrap this, all you need are a few basic self-replicating machines, then the technology passes from friend to friend until the whole world is saturated and capable of producing all future upgrades.

But we are nowhere near that dream. There’s a lot of promise to desktop manufacturing, but I’ve come to believe that the RepRap approach is probably not the right one. And I’m going to try to explain why.

Back to reality. Today the RepRap team has succeeded in designing and building a cheap 3D printer which prints in plastic only and can produce about 50% of  its own parts. This is a historic event, and should not be underestimated. However, producing a new machine still requires a lot of basic hardware such as metal rods and screws, and also more exotic components such as specific integrated circuits and stepper motors. In RepRap’s evolutionary analogy, these raw parts (as well as plastic filament feedstock) are the naturally produced “vitamins” that the RepRap consumes from its environment in order to reproduce. As time goes on, the team hopes to produce designs for upgraded RepRap machines that can manufacture more of their own parts, and not incidentally the parts for more complex objects too. For example, they hope to be able to deposit metal films with the next generation machine, which would allow the RepRap to produce electrical wiring and basic circuits.

All very lovely, but it’s time to examine the reality of this technology. Today we have a prototype design, and a vibrant community of people experimenting with and working on self-replicating desktop manufacturing. Good. But we are not by any means on the threshold of a viral explosion of manufacturing capability, because the machines are not self-replicating exponentially as hoped. A very insightful post from a site called The Clanking Replicator explains the situation:

By the second quarter of 2008 Vik Olliver had managed to print a full parts set for a Darwin with his own Darwin machine. Then a very curious and totally unexpected thing happened. Fully 6 months went by before a Darwin replicated again, this time in Canada. By that time, however, by Dr. Bowyer’s estimate of the population of Darwins was in the low thousands. What had happened?

Basically, Darwin morphed into a fully industrial product. It began with the controller boards being outsourced for production by the Reprap foundation and has culminated with a shippable kit purchasable for US$1,100  requiring little more than the sort of assembly you’d be expected to apply to something bought from Ikea. What is getting built out there in its thousands, to use Dr. Bowyer’s metaphor, is 100% vitamins – 0% replicated parts.

The initial self-replicating manufacturing machine, code named Darwin, has so far completely failed to spread virally. People are building this prototype device, but not by using another 3D printer. I want to examine why, and what this means for the future of desktop manufacturing.

First of all, the machine is far from “self replicating” from the user’s point of view. What you get when the existing design “replicates” is a set of plastic parts for a new machine. To this must be added metal hardware, integrated circuits and electronics, and stepper motors. Then the whole must be assembled by someone already skilled in making machines (here I must disagree with the Clanking Replicator post to say that assembling a RepRap still seems to me quite a complex undertaking, the sort of thing you wouldn’t want to attempt without jeweler’s screwdrivers and a multimeter.) This is hardly a consumer item. To press the evolutionary analogy, its niche is limited to hardcore geeks. This might still allow exponential growth to saturation of that niche, but desktop manufacturing is not going to transform the world until it goes solidly mainstream.

This means a consumer product. The RepRep must be no harder to reproduce and assemble than Ikea furniture; if the directions are longer than a page, you’re going to lose 95% of your market immediately. Further, the parts that cannot be desktop manufactured must be ridiculously common. You need to be able to get them at the hardware store, even in places where hardware stores are very limited (especially in such places, if we’re seriously going to consider transforming Africa.) Stepper motors are far too difficult to obtain, even in developed countries.

The RepRap team recognizes this, and is trying hard to make the machine simpler to obtain in at least two ways. First, the second generation system promises to be simpler, with fewer parts, a more robust mechanical design, and easier assembly. Second, they are looking into ways to expand the types of parts that the RepRap can print. Metal film deposition is an obvious way to go, because then the RepRap could print its own wiring and circuit boards. With time, the team hopes to further simplify the replication process.

Except, why not just buy the parts as a kit? Or even fully assembled as an industrial product? Although the exponential replication story is a beautiful solution to the problem of distributing desktop manufacturing technology, do we really care? Modern civilization is already extremely good at getting an object into the hands of absolutely everyone, everywhere. The towns ringing the Sahara desert may not have electricity, but they sure as hell have Coca-Cola, and usually motorcycles too. No one not a geek is going to care about trying to self-replicate a machine until that process is easier (and cheaper) than buying a finished model at Wal-Mart.

Which brings us to the second major problem I see with the RepRap concept: none of it matters at all until the thing is actually useful. This means making thing that people want other than parts for more RepRaps. Under the heading What Can It Make? the RepRap website shows us a fly swatter, a pair of child’s plastic sandals, a coat hook, an iPhone-to-dashboard mounting bracket, a strainer, a plastic ring, various brackets, a couple of gears, and a crappy martini glass. I understand that this is first generation technology, and it is very cool to make this stuff at home, but is there really any non-geek demand for a machine that can make these sorts of objects?

In my opinion, the RepRap community has so far focussed far too much on the coolness of a self-replicator, and not nearly enough on what it could be good for. For a research project, this is fine. However, the RepRap will never achieve viral status among the general public unless it’s actually useful to people who don’t care about technology. Which, for the purposes of this discussion, is everyone.

Now, 3D printers are just the thing for low-volume runs and prototypes, which is why such machines have traditionally fallen under the category of  ”rapid prototyping.” It’s a great technology, and I am very glad to see someone attempting to bring the price down. But it will not be a widely adopted consumer technology until it’s a better way to get stuff than going to the store or ordering it online. This means designing lots of useful things that are cheaper or easier to manufacture on the desktop than they are to obtain through the usual channels. Unfortunately, those places where consumer object distribution is most limited (it’s hard to get a martini glass in the Sahara) are exactly those places where it would be hardest to get parts for a home-built desktop manufacturing machine. Put bluntly, the RepRap team does not yet have a product that someone wants.

I have in mind a detailed analysis of common consumer items and their availability. Small plastic widgets are manufactured by the millions in China, and so are of interest only if specialized and currently difficult to obtain. But this is the market niche that high-end rapid prototyping machines already occupy, so no go. We won’t be desktop-manufacturing plastic spoons any time soon. More promising are assemblies of several parts, such as toys and small machines. Maybe spoons aren’t interesting, but I wonder if an entire suite of kitchen utensils on demand would be, including egg beater blades and corkscrews, or an entire set of model cars lovingly 3D modeled by online enthusiasts. After that, we rapidly get into objects that a RepRap cannot hope to produce. No one will be manufacturing their own light bulbs or microwaves any time soon.

Except that asking about the manufacture of existing objects is a little deceptive. Current consumer goods have been designed to optimize the cost per unit when thousands or millions are produced in a single factory. Desktop manufacturing imposes a different economics: the relevant parameters are raw materials cost, printing time, manual assembly time, and of course the requirement for parts that cannot be fabricated on the current generation of printers. An analysis of what could be made on the desktop must also be an analysis of how existing classes of objects could be redesigned to be amenable to desktop manufacturing. It’s not just about building a printer, but about redesigning the entire manufacturing supply chain and inventing entirely new fabrication methods. Want to reduce the external parts count? Perhaps new types of printable plastic fasteners can replace screws. Is the thing too complex for the user to assemble after printing? Integrated monocoque designs might be the answer. As far as I know, no one has yet tried to design a monocoque toaster.

This sort of research will also answer questions about what capabilities are most sorely missing in the current generation of printers. The ability to fabricate electronics is obviously desirable, and metal film deposition seems like a good first step. But is it really? The key criterion when evaluating any new capability must be how many useful objects could be fabricated. Asking what fraction of the parts in a RepRap could be made by a RepRap is only interesting when desktop manufacturing technology starts to become competitive with standard manufacturing techniques for complex electro-mechanical objects.

Which brings us to cost. I love the idea of downloading the design for whatever I want, but I still don’t think I’m going to print the vast majority of things I need. For plastic spoons, an injection moulding machine plus international shipping is going to be cheaper than desktop manufacturing for a long, long time. This gives desktop manufacturing a potential advantage for complex objects or short runs, especially if the labor of assembly can be avoided by automated production. Unfortunately, complex objects and minimal manual assembly are precisely what current desktop manufacturing technology is worst at. I can imagine a very advanced machine that can make its own integrated circuits and assemble them too, perhaps with its own little robot arm that is controlled by the downloaded manufacturing program. Awesome, and one day potentially a cheaper way to get one’s hands on that hot new laptop design, but that capability is a very long way away. Instead, the desktop manufacturing research program needs to ask itself what sorts of objects are not only useful and possible in the near term, but expensive to manufacture or distribute by conventional techniques. As an analogy, anyone can now print a book on their laser printer, yet we still buy bulk-printed books.

Finally, it seems to me that the RepRap team is trying to solve two problems at once, and is unclear about their separation:

1. Developing a useful desktop manufacturing capability
2. Getting that capability into the hands of everyone

The self-replicating exponential growth ecological analogy is a beautiful conceptual solution to #2, and it is also a useful technological driver for #1 because a 3d printer is a pretty complex thing to print. The RepRap team is also proud to be distributed, open-source, etc. and this is an admirable approach to #1.  However, these are far from the only solutions to these two problems.

Starting with the distribution problem, a moment’s thought reveals that it’s already extremely well solved! Getting a physical object to whoever wants it wherever they are is all but trivial at this point in history. We do already this with everything from soft drinks to mobile phones (and believe me, everybody has both.) The problem is not getting the object to people, it’s making it as cheap as possible to do so. This means that driving down the cost of desktop manufacture is the key goal; self-replication is only interesting if it’s cheaper than assembling it in China and shipping it.

As for the technology development problem, there are lots of approaches other than distributed and open source. Many basic technologies have come out of government research programs (such as jet engines and the internet) and private enterprise is of course reknowned for efficiently producing and distributing innovation. So rather than distributed global self-replication, how about this plan: figure out a way that someone can make money off of the idea, at least for a little while. Do the research discussed above and write a business plan. Take the core RepRap team and add to them the best mechanical engineers, manufacturing specialists, and consumer product designers that the world has to offer. Take the plan and the team to your favorite venture capital firm, and ask for a few hundred million dollars. I’m willing to bet that a well-funded team of crack personnel could solve the daunting technical problems of useful desktop replication much faster than the current distributed organization, and the debt to the VC would provide a strong incentive to build something that people actually wanted. This doesn’t immediately imply monopoly: open-source the design if you like, and make money off being the first to get there. There is no shortage of potential ethical business plans.

If the goal is to develop desktop replication technology to a useful state and get it into the hands of as many people as possible as quickly as possible, then I am not at all sure that self-replication is a useful near-term design goal. I do love the idea, and I believe that, eventually, self-replication will become a useful manufacturing strategy. I also really like the upgrade bootstrapping concept, where each new generation of manufacturing machine is designed to be fabricated on the last. This is an approach that will allow advances in manufacturing technology to spread at the speed of information. But we’re not there yet. Those working on desktop manufacturing technology today will need to concentrate on cost and usefulness, probably for many years to come.

Elegant, beautiful, and strangely sad.

The only substantial thing I can find on this Ulik character is this video. In it, Ulik performs with some of his contraptions such as a home-made jet-engine backpack (used with skis or rollerblades), a life-sized puppet who holds a camera and interviews him, and the front half of a car. It’s all wonderfully creative stuff, and it makes me wonder why we haven’t seen more hi-tech in circus.

For the potential is ample. We could use modern control-system technology to perform previously impossible man-machine feats of daring. I wonder about automatically balancing Segways 30 feet high than one could dance on top of, harnesses connected to a crane that cancels out its own friction and inertia and modulates the effective gravity under performer control, a ridiculously precise robotic juggling partner, or powered jumping stilts with built in balance and timing systems. This is not mere robotic circus; at their best, such machines become something between costume and vehicle, an extension of the performer’s body that makes them, taller, stronger, faster, or able to move excitingly inhuman ways.

Given that such wide artistic and technological possibilities exist, I find it hard to believe that they won’t be developed. We may currently be witnessing the last generation of aerial circus that does not make heavy use of technology.

It is now possible to see what a person is looking at by scanning their brain. The technique, published last November by a team of Japanese neuroscientists, uses FMRI to reconstruct a digital image of the picture entering the eye, albeit at very low resolution and only after hundreds of training runs. Still, it’s an awesome development, and many articles covering this research have called it “mind reading” (1, 2, 3, 4, 5). But it really isn’t, and it’s fun to explore what real “mind reading” would imply.

When I hear “mind reading” I want psychic abilities. I want to be able to know what number you’re thinking of, where you were on the night of March 4th, and what you actually think of my souffle. This is the sort of technology that could be badly misused, as the comments on one blog note:

Am I the only one finding this DEEPLY disturbing? It opens the doors to some of the scariest 1984-style total-control future predictions. Imagine you can’t hide your f#&%!ng MIND!

Fortunately, we’re not there yet. Morover, if we did have the technology to read minds, we’d have much bigger societal issues than privacy to deal with. The existence of “mind reading machines” would imply that we possessed good formal models of the human mind, and that is a can of worms.

But back to today. The paper by Yoichi Miyawaki and colleagues describes a technique for exploiting retinotopy, the fact that certain areas of the visual cortex are direct “maps” of the retina. First, a series of 10×10 black and white test images are shown to a someone while their neural activation is recorded by FMRI. The responses to these test images are used to ascertain which areas of the visual cortex correspond to which areas of the subject’s field of vision. When the neural map is complete, it can be read “backwards,” going from neural scanner results to a low resolution representation of whatever the subject is currently looking at.

This is a long way from a tool for the thought-police. First, the algorithm requires training on each new person. Also, an MRI machine is a huge, expensive, complicated piece of machinery which requires the subject to stay very still over a period of minutes — widespread brain scanning is, for the moment, completely out of the question. But most fundamentally, the information recovered is nothing more than what the eye is currently looking at. You might as well just tape a digital camera to the subject’s head. The pictures would be a lot better.

What is it that we imagine for a mind reading machine? Perhaps a printout, in words, of every thought that goes through someone’s mind. But do people really think exclusively in words? What about their emotions, or their unconscious responses, or even the complete set of minor joint aches and temperature sensations all over their body? Or how about a video playback of the events of yesterday evening? Impossible, because that’s not how human memory works. When we think about it carefully, we realize that we have an extremely poor conception of what is actually “in someone’s head.”

Compounding this problem is the fact we can’t even say what’s in our own heads. We think we can, but we can’t. Decades of psychological experiments show that access to the contents of our own minds and the working of our own thought processes is very limited. Consequently, we cannot answer the question “what would a mind-reader read?” through introspection.

This is why, before we could build a mind-reading machine, we would first need formal models of a “mind.” We need the sort of mathematical models that one can manipulate with a computer, because computers will surely be intensely involved in any mind reading technology. If recent developments in linguistics and artificial intelligence research are any guide, these models will be huge, associative, and statistical in nature, nothing like the structured logic we think we possess. For example, Google translates web pages between different languages without using anything like formal grammar models.

In other words, we cannot “read minds” because we have very little idea of how minds might be stored on a computer. This problem is known in AI as “knowledge representation,” and we still know very little about it.

Good formal models of the mind, if possible, are the technological precursor to entire fields of information engineering, and this is why I’m not worried about mind-reading technology per se. We’ll get beneficial things like accurate machine translation and computers that respond to voice queries — no more fighting with software that just doesn’t understand what you want. (Think also of the possibilities for art and expression.) We’ll also get uncomfortable technologies like sickeningly effective advertisements that exploit behavioral quirks we didn’t know we had, and NSA-funded conversation snooping programs that make existing keyword scanners look like the toys that they are. Finally, it would be possible to use accurate human mind models for pure evil: imagine a computer virus that was designed to read your personal files and figure out how best to convince you that the Dictator was beneficent. All of this may sound very far-fetched, but we’re going to build these things if we possibly can: think of how much money Google makes from each percentage point of improvement in ad clickthroughs.

If the Japanese FMRI technique seems positively simplistic in this light, that’s because it is. They have read retinas, not minds. They are extracting a representation we already have abundant experience with: images. Saying that we’ve made a step towards reading minds is ridiculous; Thomas Edison might just as well have claimed to “record thoughts” when he announced the phonograph.

I bother with all of this both because I think science journalism is often done badly, and because I believe that it’s important to get hysterical about the right thiings. One comment posted to a video of the research reads, “this is the beginning of the end of free thought.” Perhaps the continuation of this type of FMRI research really will one day lead to the ability to determine what someone is thinking without invoking their consent, but torture already does that. To me, the ability to represent someone’s thoughts in electronic form has far greater implications than mind-reading per se, and this sort of FMRI research — as impressive as it is — contributes little to that enterprise.

How much overlap is there between the web in different languages, and what sites act as gateways for information between them? Many people have constructed partial maps of the web (such as the  blogosphere map by Matthew Hurst, above) but as far as I know, the entire web has never been systematically mapped in terms of language.

Of course, what I actually want to know is, how connected are the different cultures of the world, really? We live in an age where the world seems small, and in a strictly technological sense it is. I have at my command this very instant not one but several enormous international communications networks; I could email, IM, text message, or call someone in any country in the world. And yet I very rarely do.

Similarly, it’s easy to feel like we’re surrounded by all the international information we could possibly want, including direct access to foreign news services, but I can only read articles and watch reports in English. As a result, information is firewalled between cultures; there are questions that could very easily be answered by any one of tens or hundreds of millions of native speakers, yet are very difficult for me to answer personally. For example, what is the journalistic slant of al-Jazeera, the original one in Arabic, not the English version which is produced by a completely different staff?  Or, suppose I wanted to know what the average citizen of Indonesia thinks of the sweatshops there, or what is on the front page of the Shanghai Times today– and does such a newspaper even exist? What is written on the 70% of web pages that are not in English?

We all live on the same physical planet, but the information worlds we inhabit must be vastly different. This are many reasons for this other than language, but language alone is enough to isolate humanity from itself.

And so, my question: how many islands are there in our multi-cultural information space, and how are they connected? I am willing to bet that a full-scale web map would show several large networks in the main languages of the web — English, Chinese, Spanish, Japanese, German, etc. — but few connections between them, web sites frequented by bilingual or bi-cultural individuals, who after all are the true gateways between cultures. The structure of the interconnections might tell us something about the relationships between cultures, and the actual number of links might provide some measure of how close or how far apart we actually are. The individual URLs themselves would also be extremely valuable information, representing high-bandwidth links between cultures, the trans-occeanic fiber between continents in the infosphere.

There is a second geography to the world that we’ve never seen. I don’t even know what I’m missing.

Creating such a map would be a trick, but by no means out of the reach of an academic project or a small company. Google says there are currently over one trillion (10^12) unique web pages (for their particular definition of “unique”, which is more complex than it might seem.) Unlike a search engine, a language-based web map does not require the full contents of every page, merely the outgoing URLs and a discrete categorization of the language (which can be automatically determined even without any document meta-data.)  Assuming that each URL  is assigned a unique 32 bit ID, another 32 bits for language and other info, and then links to an average of 20 other pages (estimates vary), this is about 100 terrabytes of data — or perhaps $15000 worth of storage at current prices. This index could be created from a fresh crawl, or by parsing an existing one, such as from the folks at the brand new and very awesome DotBot open index of the web.

The next step would be to generate the visualization of such a massive data set. The complete graph could be laid out in two or three dimensions using existing clustering methods. The resulting map could be traversed using GPU-accelerated rendering techniques for very large data sets, probably after some sort of hierarchical pre-processing that produces proxies for zoomed-out views of the network. A usuable UI would be crucial; the entire map needs to be navigable at multiple scales and composed of live, hyperlinked objects. The right visualization also depends on what you are trying to discover;  ultimately, there can be no single map because the choice of visualization is dependent upon usability and aesthetics, as the huge variety of beautiful maps at Visual Complexity demonstrate.

The analysis could go much deeper with more computing power. Machine translation is currently poor, but it is probably good enough to detect whether one document is a translation of another. With this capability, we would actually be able to quantify the percentage of (public) textual information that makes it from one language into another and identify the key organizations that act as conduits. Further study might reveal fascinating things, such as selection biases in the types of news or information that get translated. The implications for differences in belief between cultures are obvious.

Yet even  a “links only” data set could still answer some highly revealing questions,  such as “what percentage of web sites are visited by people from multiple cultures?” or even “what is the best gateway between Polish and English film reviews?” This could be done without visualization, but it would be a mistake not to draw the actual maps.  Not only do pictures engage our spatial reasoning in a way that raw bits never can, but such a map would re-make an obvious point that is too often lost: in terms of communication between cultures, the world is not nearly as small or interconnected as we’d like to think it is.

The English text on this page reads: “Access to this web site is banned by ‘TELEKOMÜNİKASYON İLETİŞİM BAŞKANLIĞI’ according to the order of: Ankara 1. Sulh Ceza Mahkemesi, 05.05.2008 of 2008/402″

Just to complete the irony, I was looking for a video of the Oscar Grant shooting when I first discovered this “blocked site” page.

I have previously reported on internet censorship in the United Arab Emirates. Turkey’s “you can’t see this” page is not nearly as flashy, and the censorship may be less severe: I can reach Flickr from here, for example. However, it is not possible to read the website of Richard Dawkins in Turkey; there even appears to be a more specific (and forthright?) banner page.

(Sadly, Google Translate does not support Turkish — dear lazywebs, can anyone out there give an exact translation?)

This suggests that Turkey’s censorship attempts — all of which can be easily circumvented with tools like Tor — are more concerned with social and religious mores of various sorts, as opposed to the efforts of countries like China where there is a clear political motive underlying the censorship pattern (for example, the Tiananmen Square killings never happened, according to Google China.)

For more, please see the fabulous Open Net Initiative, which tracks and reports on internet censorship worldwide, and has an excellent review article on the Turkish situation. Unsurprisingly, Turkey also has had some recent problems with freedom of expression.

The e-voting fiasco has illustrated that paper ballots are a better system than they might at first seem. Paper preserves voter secrecy, it is auditable after the fact, and it is even reasonably transparent, if one also allows election observers. But paper ballots must be closely guarded and cannot be directly counted by members of the general public, who in the end have no choice but to trust election officials,  observers, counting equipment, and the entire chain of custody. Rather than simply duplicating paper ballots electronically, we should strive to improve upon them.

This seems to be possible. Modern cryptography suggests the possibility of a new kind of incredibly transparent and fair election, where ordinary citizens can verify the soundness of the election for themselves, without ever needing to trust blindly that a huge array of machines and people have acted correctly. This represents a fundamentally new ability: for the first time, it may be possible to hold truly “open” elections.

What are we trying to accomplish?

Ideally, a democratic voting system would satisfy the following criteria:

• Secrecy: to prevent coercion or vote-buying, each person’s vote must be secret in perpetuity.
• Transparency: all voting procedures must be public and understandable by everyone.
• Verifiability: it must be possible to independently audit or validate the election results.
• Usability: it must be easy to vote, and cheap to deploy the system for hundreds of millions of voters.

In this way, each person would vote freely, while the entire society could have confidence in the outcome. The difficultly with these criteria is that they conflict: it is hard to preserve both secrecy and verifiability in a simple, transparent way.

Paper ballots fall short of the these ideals in many ways. They are nicely secret, and the process is reasonably transparent as there are public laws describing exactly how the votes are to be tallied, regulations providing for election observers, etc. However, independent audits are not really possible, because they require access to a large quantity of fragile and politically sensitive paper. In principle, we would like it to be possible for any regular citizen with sufficient time on their hands to perform a complete audit of the election results.

Elections results could be openly verified by publishing copies of every ballot cast, but only if there was some way to ensure that these copies were accurate. This could be done by issuing to each citizen to some sort of receipt of their vote which could be checked against the public list, but then votes would not be secret: they could be coerced or bought by offering a clandestine cash reward for receipts.

While paper ballots leave much to be desired, current electronic voting systems are worse. All e-voting machines are essentially “black boxes” that transform the voter’s choice into a final tally by some complex and unknown process. This makes them completely non-transparent. In the worst case, paperless direct recording electronic (DRE) voting machines are not at all verifiable, which makes them subject to both invisible malfunction and deliberate hacking  (either in the voting booth or at the tally station.) There has never been a convincingly documented case of miscount or fraud with DRE machines, but that may only be because such machines leave absolutely no record of the election process!

Because of this, many American states now require a paper record even for otherwise electronic machines, but even paper audit trails are problematic: when is an audit performed? Will all ballots be routinely audited or just a sample? What is the right sample size for confidence in the results? What happens if a discrepancy is discovered? Meanwhile, other states have gone back to paper ballots entirely and a number of electronic voting machines have been de-certified.

Cryptographic Hope

Enter cryptography, the discipline that has brought us such miracles as secure communication between two parties who have never exchanged any information in secret (public-key cryptography), tamper-proof  electronic documents (digital signatures), and the ability to prove that one knows a secret without giving it away (zero-knowledge proofs.) In the wake of these achievements, there has been some hope that proper cryptographic protocols will simultaneously solve the secrecy, transparency, and verifiability issues.

Voting might still be electronic in a cryptographic system, but the security of an election would rest on open cryptographic protocols rather than on trusted system implementations, the physical security of ballot boxes, or the honesty of certain people. Even better, the election results would be auditable at any time from public information, and each voter could verify that their own ballot was correctly recorded yet their personal vote would remain secret and unprovable.

Such a system is said to be end-to-end auditable, and represents a fundamental shift: for the first time, it may be possible to hold completely “open” elections in the sense that governments and election officials would have no more authority or power than ordinary citizens. This is unprecedented in human history, and it is exciting.

It is also quite a trick, and has never been demonstrated in practice. Aside from secrecy, transparency, and verifiability, any proposed cryptographic voting system must guard against many different kinds of attacks. These include tampering and “denial of service” attacks against the election, such as the ability to spoil the election through some sort of interference (as might suit an opposition group) or to arbitrarily declare that it was spoiled in some non-disprovable way (as a dictator might wish to do.)

A Toy Example: ThreeBallot

Like much of modern cryptography, the simultaneous provision of both secrecy and verifiability seems counter-intuitive. To aid in the study and conceptualization of such systems, professor Ron Rivest of MIT (the “R” of RSA fame) invented a “toy” voting system in 2006 called ThreeBallot.

It works like this: each voter is given three identical ballots in the voting booth. To vote for a candidate, the voter writes a mark on two randomly chosen ballots; to vote against a candidate, only one randomly chosen ballot is marked.

A valid vote is one in which each candidate is marked on either one (against) or two (for) randomly selected ballots. This could be checked e.g. by an optical scanning machine, much as paper ballots are currently validated at polling stations.

Then the voter secretly chooses one of the ballots and makes a copy of it as a receipt; the others are dropped into the ballot box. Each of the three ballots has a unique serial number.

After the election, all ballots are published publicly, and anyone can tally the election results from these copies. Additionally, each voter can verify that their ballot was published accurately by looking up their receipt in the published list. Yet there is no way to use a receipt to determine who someone voted for, because the voter can arrange to have any particular set of markings on the receipt that they keep. The receipt also prevents tampering, because a would-be tamperer does not know which of the three ballots the voter has retained. Thus there is a 2-in-3 chance of getting away with tampering with (or deleting) any one vote, but only a (2/3)^N chance of getting away with tampering N votes — like tossing N heads in a row, these are very rapidly shrinking odds.

ThreeBallot was never meant to be a real election system, and in fact in a University class voting experiment ThreeBallot was found to have significant security and usability problems: a third of voters couldn’t produce a correct set of ballots the first time, and a student “attacker” was able to manipulate about 20% of the votes cast, enough to change the election result. He did this in part by clandestinely reading other people’s receipts, such as those left in voting booths or on desks. This reminds us once again that security is always about much, much more than good cryptography.

However, the basic ideas of ThreeBallot — randomness in the vote-casting process, voter receipts, published ballots, and probabilistic tampering detection — are found in virtually all cryptographic voting schemes.

Realistic Proposals

Serious proposals are somewhat more complex. Some are designed to be entirely electronic while others are additions to paper ballot systems. Major proposals include Punchscan (2006) by cryptographer David Chaum, Scantegrity (2007) by David Chaum and Ron Rivest, and Bingo Voting (2008) by a trio of German researchers. All of these systems are very cryptographically clever, but as always, security in the real world is about much more than cryptography. A 2005 paper considered how a real election system employing end-to-end auditable protocols might work, and proposed various non-cryptographic attacks including collection of receipts, social engineering of election workers, and denial-of-service attacks which could invalidate the entire election (such as hacking the voting machines to record spoiled ballots.)

Also, many problems just cannot be solved cryptographically. One major reason why we don’t have internet voting is that it is impossible to prevent coercion and vote buying if voters can mark their ballots at home. A physical polling booth can at least be secured against witnesses — though not against, say, someone who will pay for cell-phone camera pictures of a suitably marked ballot. A completely secure voting system is  probably completely impossible.

Nonetheless, there is hope for electronic voting systems, not because they would allow us to vote cheaper or faster or more conveniently, but because they hold the promise of more transparent elections. Would-be designers and implementers of voting systems must realize that the purpose of a voting system is not just to count votes, but to ensure that everyone believes that the process was fair, and to ensure that this fairness can be proved as easily and as widely as possible.