sencjw

Here’s my website. Some common things are on the navbar. I’ve also worked to preserve any locations that people would have linked to from elsewhere. If I’ve broken a link of yours that used to point to sencjw.com somewhere, let me know and I’ll put it back, cool URIs don’t change.

Recent Blog Posts

my gpl talk

2012-05-05

note to the reader, my speaker notes are reproduced below and all run together. Also, the editing isn’t top-notch.

notes

First a warning, if you thought I was going to talk about software licenses, I’m not. Not really. I’m going to talk mostly about people. Ideas are one thing, they are the compiled results of processes that people go through. I want to decompile some ideas. I want to talk about that.

To that end, let me tell you a story. Let’s go back to 1980, Richard Stallman is employed as a staff hacker at the MIT AI lab. This is the stuff of legend.

The lab had recently been given a new prototype printer from XEROX PARC. This was ten times faster than the previous printer, finishing a 20 minute job in 2 minutes and with more precise shapes to boot. This was the same sort of tech that a decade hence would touch off the desktop publishing revolution. But back at the AI lab, the printer was becoming the source of more headache than anything else. Stallman and others would send jobs to the printer only to show up later to find that the job had jammed four pages in. This was a minor annoyance, but it was multiplied by everyone at the lab. Stallman thought, “why should I have to babysit this machine when I can code?”

Stallman knew a way to attack this sort of problem. On a previous printer he had modified the source to insert some monitoring code in the printer driver. Periodically, Stallman’s code would check to see that the printer was proceeding in its assigned job, if it had stalled, the program would alert whoever’s print job was affected. You’d get a message like: “The printer is jammed. Please fix it.” It wasn’t perfect, but it informed those most interested in the problem.

The solution this time around would be similar. Stallman could grab his old code, tweak it for the new printer and voilá: jam notifications. So Stallman rolled up his sleeves, grabbed a coffee, and opened up the Xerox source code.

If you see where I’m going here, you’ll probably see what’s coming next. There was no source code. Stallman even spoke with the programmer that had worked on it and that programmer wasn’t allowed reveal the code to Stallman.

This is the moment where something happens. This is where an insight strikes, the apple falls on your head, the disparate pieces line up and you need to jump out of the tub and tell the world. Stallman decided, at that moment, that some fundamental wrong had been done: the wrong of not being allowed to help your neighbor by telling him how code works., This brings me to the main point of this talk. This is the thing that, even if everything else you hear is mangled or forgotten, I want to come through unchanged: RMS believes that software has moral implications, the choice of what kind of SOFTWARE you want is a choice about what kind of WORLD you want.

Note all the things that I didn’t say. It isn’t about what is technically superior. It isn’t about what is good for being able to sell. It isn’t about what the legal department says. It has no bearing on what various companies will tell you to be worried about. It isn’t about being good for playing games, or having flash support. It is nothing more and nothing less than a philosophical stance. You can agree with it, or disagree with it in exactly the same way as you would argue about Plato’s Forms.

I feel like this is the key misunderstanding in discussions surrounding the GPL and Free software. I’m taking a philosophical, an ethical, and maybe a moral stance. I haven’t brought anything else into it. Often, when I see discussions about software licenses, I feel like people are talking past one another from the very first sentence.

It is profoundly nonsensical to compare something like “justice” to something like a wrench., Philosophers begin by defining words, because if they don’t, we’ll get so mired in the muck of argument that no points are made, no progress is made.

The word “free” is a good place to start. Free can be taken to mean “no cost” but it can also be taken to mean “freedom”. This is sort of a fine point to make, but I think it could lead to lots of confusion.

Free software has to do with the “freedom” part. There are lots of really good objections at this point. The one that I have anticipated is “freedom for whom?” And that’s the core of the so-called permissive divide in the broader category of “open” software. The permissive people would respond to the “freedom for whom” question with something like “certainly not for me, you say I must share changes, that’s pretty restrictive.” And the answer to “freedom for whom?” that I want to present here is…

Well, that’s the rest of my talk., The GPL is a really a more general case of the Emacs license.

Now Emacs has a pretty storied history, wikipedia dates it back to the mid seventies, well before GNU or Emacs-as-GNU-project. But by the time of the release of Emacs 15, there was a sort of proto-GPL license attached. It served to give “users the right to make and distribute copies” and “the right to make modified versions, but not the right to claim sole ownership of those modified versions”. It was moving in a similar direction, but it was not as legalistically formal as the eventual GNU project would need it to be.

Stallman’s intellectual property attorney at the time viewed the GNU Emacs License pretty much as a simple contract, although one that stipulated a rather odd price. Rather than money, the license cost access to any changes. Users would have to share modified versions of the software. The attorney remarked: “I think asking other people to accept the price was, if not unique, highly unusual at that time”

In 1989 a 1.0 version of the GPL had emerged. The preamble read:

The General Public License is designed to make sure that you have the freedom to give away or sell copies of free software, that you receive source code or can get it if you want it, that you can change the software or use pieces of it in new free programs; and that you know you can do these things.

To protect your rights, we need to make restrictions that forbid anyone to deny you these rights or to ask you to surrender the rights. These restrictions translate to certain responsibilities for you if you distribute copies of the software, or if you modify it.

one notable change was that users were no longer required to share changes. You could make private in-house tweaks to the software without being forced to share these changes back to the community. , License agreements are not usually characterized by what they give you, rather, as we scan ever longer End User License Agreements, or plow through revision 271 of Facebook’s new much-better-we-assure-you privacy policy, we are looking for things that they are taking from us.

The GPL, in essence, tries to codify a very idealistic hacker ethic. It is for tinkering, changing, breaking, reassembling, and passing it on to your friend. It is software as mix-tape.

The main things that the GPL gives you are broken down into four parts, aka the “four freedoms”, Zero: You’re allowed to do what you want with the software. An author can’t proscribe the software’s use for something that they don’t approve of. This is pretty profound, I think.

one: if you are going to be able to do this, you’ll need the source code. You’ll also need whatever is required to actually end up with a working program. This can be a point of contention. A corner case of this is in embedded systems such as set-top boxes where the code may be GPL, busybox is a common example, but you can’t actually change the code due to things like code signing., two: I think it is interesting that two emphasizes the goal of the redistribution. It isn’t just for fun or for copying’s sake, it is because we view software as something that can help people.

freedom three, the final freedom. You’ll also need access to the source code to realize this one. You are allowed to make public changes to the code. The difference with freedom one is that you’re allowed to do this out in the open, rather than just in private and for your own reasons. You can fork. You can contribute back.

lurking in freedom three is also the core of my argument, which I promise I’m getting to really soon., I’m going to try and dispel a common myth about the GPL, one I’ve heard a lot. The general gist is that “the GPL is to copyright as anarchy is to government.” Something that is opposed to the very notion of it. This is where people get the idea that any business built on such shifting sand of self-destruction must be flawed in some way.

Opposition to copyright is an interesting subject, there’s lots of good debate. But it doesn’t really have anything to do with the GPL. The GPL has staked its efficacy IN copyright.

Far from being some sort of anti-copyright construct, the GPL’s EXISTENCE depends on copyright, if you didn’t have copyright, you couldn’t have the GPL (or lots of other stuff). You wouldn’t get any say in what people do with your stuff… but that’s another discussion entirely!

So for the rest of the talk, consider copyright to be a constant underpinning, a foundational necessity for everything else we’re talking about. It’s just that we’re going to use it for something that it wasn’t intended: we’re hacking it., BAM: flip it

copyleft: pay it forward.

As Stallman said: “see [the GPL] as a form of intellectual jujitsu, using the legal system that software hoarders have set up against them”, he actually lifted this from a similar sticker from a sci-fi convention which read: “Copyleft (L), All Rights Reversed.”, And this brings me around to what my thinking on the GPL is. I guess I’m kinda surprised by all the emphasis on virulence these days. The metaphor is broken. Metaphors are broken–but that’s another talk.

Casting aside any metaphors, the GPL is an inductive license. This is a term that I made up but I think it describes the nature of the GPL much better than saying that it is viral.

An initial case is established. You have the four freedoms: the freedom to run, freedom to change, freedom to redistribute, and the freedom to share those changes.

But for it to really be Free software, the person receiving the software must have these freedoms. So it is not good enough for us to leave it here. We only have the base case for a software license. We have to prove the general case, not me or you, but person N+1., So the person who’s freedom we’re talking about is person N+1, the inductive person.

This idea is the essential difference between being permissive software and being free software. Free software describes the case of that person N+1, inductively. It raises the “freedom for whom?” question and answers it with “the inductive person”.

So I’ll leave where, approximately, I started with a definition:

The word “induction” is the practice of deriving general laws from specific cases, it arises from a root word meaning “leading to” or “hypothetical”. Free software asks us to consider this hypothetical person on the assumption that it could someday be anyone, indeed everyone.

secret santa

2011-12-27

While I was sitting around and eating a ton of Christmas food, I got to thinking about the Secret Santa problem. In its most basic form, this is the same as something called a derangement. I mention it just because I think the name is cool, the concept is super simple, a derangement is a permutation of the elements of a list such that no element stays in the same place:

[1, 2, 3] would have a derangement:
[2, 3, 1]

notice that each element has moved. So this pertains to secret santas because if you are just not allowed to chose yourself then a derangement (like this) is all that you’d need, it would be a valid secret santa!

> zip [1, 2, 3] (derangement [1, 2, 3])
[(1,2),(2,3),(3,1)]

cool! person 1 gives to person 2, person 2 gives to person 3, and person 3 gives to person 1.

As my family could tell you, I thought that I could do better (in keeping with my motto “if it ain’t broke, fix it until it is”). Wouldn’t it be cool if in additon to just forbidding the case where you pick your own name (reflexive), you also can provide two more lists. One is a list of pairings which are disallowed and the second is a list of pairings which are to be discouraged (less likely).

I’ve implemented almost what I just described. In the code below, I don’t actually make a selection from some distribution where discouraged selections are less likely. Instead, I’ve added a bestSantas function that allows you to limit yourself to selections that are under a certain amount of badness (a selection has 1 point of badness for each discouraged pairing that it includes). I hadn’t decided how I wanted to select from among differing levels of badness yet. But anyway, enjoy!

sopa hearings

2011-12-15

These SOPA hearings are beyond awful. The fact that the hearings, at least as I have found them at 6pm CST, are about who has standing to sue in the US, nothing more. This feels like a group of people standing around trying to decide to roast hotdogs or marshmallows as the Library of Alexandria burns.

This letter, which carries the signatures of many the designers and builders of the Internet (TCP/IP, BIND, DNS, HTTP, MIME, etc.), clearly lays out the damage that this bill would cause.

What I find infuriating about the discussions that are happening right now regarding SOPA, is that the committee members are so glaringly ignorant of the Internet. There is no discussion of how it would work with DNS, what burden that would put on hosting companies, schools, organizations, etc. In short, they don’t appear to know how the Internet works.

I cannot help but worry that there are influencers promoting this bill who have a very narrow interest in the Internet. They see it as an economic threat to an old business model (movies, music) and it must be turned into a safe and effective mechanism to secure an income stream in the future.

strangeloop 2011 notes

2011-09-24

I got back from Strangeloop 2011 just this week and wanted to cover some of the interesting points from this really fascinating conference (it is on my must go list from now on)!

It was incredibly difficult to get to all the talks that I wanted to see because the conference was “seven talks wide” at most points. A common theme emerged where, as I finished up a talk in one room, I would see the stream of tweets start rolling in about some incredible talk that I had just missed; I can’t wait for those videos.

Here’s my recap of the stuff that I went to:

Sunday (workshop day)

• Haskell: Functional Programming, Solid Code, Big Data with Bryan O’Sullivan - this was a really nice intro to Haskell for someone that hadn’t ever seen it before. I’ve worked through about half of the “Real World Haskell” book so a lot of this was not new. But it was great to see one of the authors explain some points himself. There was also some interesting comments from Gerald Sussman about how haskell is the “most advanced of the obsolete languages” (more on that later).

Monday (first day of conference)

• Category Theory, Monads, and Duality in (Big) Data with Erik Meijer - This was a really cool opening keynote where Erik Meijer launched the new term CoSQL instead of NoSQL by showing how the two concepts are duals of one another (in the mathematical, category theory sense). This proved to be something of an overarching theme of the conference, things being different but mirrored versions of the same thing. see: A co-Relational Model of Data for Large Shared Data Banks.

• [I skipped this timeslot because I was on the hallway track listening to Erik Meijer talk about static typing with some scala folks; very interesting!]

• An Introduction to Doctor Who (and Neo4j) with Ian Robinson - I have to admit, I got sucked in because I’m a huge Doctor Who fan, but I had heard of graph databases before and Neo4j looked to be a really interesting one. In particular, I wanted to see if this could be used from Clojure (yes: borneo and clojure-neo4j). The talk concerend building a very complicated network of the relationships between several Doctor Who props (Daleks!) over time. It was pretty easy to see how these mapped nicely to nodes with arcs between them.

• Skynet: A Scalable, Distributed Service Mesh in Go with Brian Ketelsen - this was a cool talk about a lightweight framework written in go for writing distributed applications that are highly resilient. It uses Doozer for data storage (though it didn’t in this talk).

• Parser Combinators: How to Parse (nearly) Anything with Nate Young - This talk gave examples of writing parser combinators (where a parser here means a function that can consume a little input, and then returns another function that consumes input after it). The idea is to chain these parsers together with combinators (higher-order functions which take parsers and operate on them, like “oneOrMore” etc.). This talk reminded me of Bryan O’Sullivan’s funny phrase about how haskell’s “>>=” operator (read “bind”) is written in “moon language”.

• Getting Truth Out of the DOM with Yehuda Katz - This was a talk about the SproutCore framework. Katz had a lot of insight about how to keep the browser interaction abstract and event-based rather than mucking about (and then being mired) in the DOM.

• We Don’t Really Know How to Compute! with Gerald Sussman - This was a mind-blowing keynote. In fact, I had to develop a new unit of measure, the Eureka, which denotes having one’s mind blown once per minute. I think that in the 50-some minute talk that Sussman gave, I may have had more than about 50 mind-blowing thoughts. At one point Sussman asked how much time he had left and someone from the audience yelled out “who cares?”, which was pretty much the feeling in the room.

  Sussman started out the talk with a picture of a Kanizsa Triangle and mentioned that the brain can infer that there is a hidden triangle in just about 100 ms which is a few tens of “cycles” for the brain. With a computer, we don’t know how to even begin to solve this recognition problem in that few of cycles; we don’t really know how to compute. Sussman’s idea (which I can’t do justice to here), was that computing as we know it has to and will change in the near future. Computing will become massively distributed (“ambient”, but this term is from a later talk) and in disparate nodes that must collaborate to arrive at answers.

  His example, a Propagator was a program that can integrate more annd more data while keeping track of the provenance of that data. Or another way an “independent stateless machine connecting stateful cells”. Amazing!

Tuesday (second day of conference)

• Embedding Ruby and RubyGems Over RedBridge with Yoko Harada - This didn’t make that much sense to me until coworker (@devn) started doing some cool stuff with using ruby gems from clojure.

• Event Driven Programming in Clojure with Zach Tellman - This was a really cool talk. It looked to me to be an implementation of go-style concurrency (channels) in clojure. There was also a macro that would analyze data dependencies and do the correct async calls. The projects are called Lamina and Aleph and they’re one of those things that I want to find a project on which to use them.

• Teaching Code Literacy with Sarah Allen - This was a talk about how to give kids the opportunity to learn about programming at an early age (Allen says that programming is one of those things that you don’t know if you’ll like it until you’ve tried it.) She also had found that the ages that programming should be introduced is 5th-6th grade; earlier than I thought!

• Post-PC Computing is not a Vision with Allen Wirfs-Brock - This talk started with a breakdown of the eras of computing. First was a “coporate” era, then a “personal” era, and now we are entering the “ambient” era. Each era is defined by what ends computing resources are put toward. In the coporate era computing was used to solve problems that businesses had, then computing became more available generally, and finally it is becoming ubiquitous. This talk also covered the history of the browser and how it is, and will be, the platform for the forseeable future.

• Simple Made Easy with Rich Hickey - Rich’s talk was an argument for disentangling computing. It started with separating the notions of “simple”, “complex”, and “easy”. Easy is a subjective thing, things that I find easy you may not. Simple is objective, it derives from the notion of “a single fold”. Complex is just the opposite, it is “woven or braided”. We must avoid adding complexity to our software, or as Rich put it, we must not “complect” it (“to interweave or entwine”). Humans have a finite (and very limited) ability to handle many factors simultaneously, and so to have any hope of working with difficult problems, we must be rigorous in working toward simplicity.

Rich had a few words for TDD in his talk, and I think these were widely misinterpreted. His point was simply that tests have a cost and a thoughtless devotion to them will risk underestimating that cost. I think a lot of people took that to mean “you shouldn’t test” or that “tests are worthless”, but I think he was just pointing out that they’re not free. He introduced the term “guardrail programming” for a style that just bounces between the guardrails rather than proceeds to a destination by steering.

This talk drew a standing ovation from the crowd, including, I hear, Gerald Sussman. I’ll be looking for it on video when it comes out.

Strangeloop 2011+N is definitely on my must-attend list. The people that I met (which could be another couple of blog posts) were worth the admission all by themselves. The talks were fascinating and gave me a ton to read up on. The conference felt like it was well-run and organized. St. Louis was a cool city to hang out in (I wish we had the same open-container law in Madison!). I can’t wait for next year.

Madison Ruby Conference

2011-08-16

I’m going to be attending the Madison Ruby Conference this Friday and Saturday. I must confess that I’m not a Ruby-ista (does this reveal my pythonista heritage?) by training but I’m really excited to go and see who attends, chat with smart tech-folk, and generally have a good nerd-time yakking about code. Plus, it couldn’t hurt to see a little ruby code and see what all the fuss is about. I’m not going to confine myself to: C, Java, Python, Lisp, Haskell, and shell. Conference Ahoy.

For a complete list of posts, see the blog page.