“Ingenious ideas are simple.
Ingenious software is simple.
Simplicity is the heart of the Unix philosophy.
The more code lines you have removed,
the more progress you have made.
As the number of lines of code in your software shrinks,
the more skilled you have become
and the less your software sucks.”
“By avoiding complexity when possible,
and containing it when it is unavoidable,
we can maximize the probability
that the number of security vulnerabilities found
remains zero in the future.”
uniﬁcation via homoiconicity 1:
#:"Runs a program."
myeval:[lambda '[prgm:1 env:env]
'[if [isName? prgm]
else:[apply /prgm/0 prgm env]
uniﬁcation via homoiconicity 2:
deﬁne f == 7 b ! . (* a function *)
fb@. (* execute the function (7) *)
deﬁne a == [15 b !] . (* a quotation *)
ai (* execute the quotation *)
b@. (* 15 *)
37 a rest cons (* modify the quotation: [37 b !] *)
i (* execute the modiﬁed quotation *)
b@. (* 37 *)
Let&#x2019;s have a big round of applause for Alex Miller for putting together such a great event.
Introduce myself: Twitter, Down For Everyone Or Just Me, just finished writing a book about Scala (a decidedly un-minimal language).
This quote is from Eric S Raymond, &#x201C;The Art of Unix Programming&#x201D;.
I&#x2019;ve been thinking about minimalism in software for about two years. I maintain a tumblelog called Minima that captures example of what I call &#x201C;applied simplicity&#x201D;. I&#x2019;ve gathered dozens of examples, but it&#x2019;s something I&#x2019;m still thinking about. What I want to share today is the culmination of that thinking so far.
There&#x2019;s a difference between big-M and little-m minimalism that we need to establish straightaway. People have very different ideas of what constitutes minimalism, so first we&#x2019;re going to talk about what we&#x2019;re not talking about today.
This is little-m minimalism. It&#x2019;s beautiful, it&#x2019;s simple, it&#x2019;s marketable, but it&#x2019;s not what we&#x2019;re talking about today.
This is Steve Jobs in 1982. He hadn&#x2019;t just moved in. This is how he lived.
If you were hoping for a tutorial on how to be a minimalist, this isn&#x2019;t it. You know how to do that, in the back of your head: you get rid of everything you don&#x2019;t need, everything that makes you unhappy, and you just do it. You don&#x2019;t need someone to come up here and talk to you for an hour about that, just as you don&#x2019;t need blogs and self-help books to do it. That&#x2019;s the whole point. We&#x2019;re not talking about any of that today.
This screenshot was found under the headline, &#x201C;Beautifully Minimalist Conky Setup&#x201D; on the popular Lifehacker blog. This is the exact opposite of minimalism. It&#x2019;s fiddly, it&#x2019;s glossy, it completely unnecessary. It&#x2019;s also not what we&#x2019;re talking about today.
There&#x2019;s a quote that comes to mind when I see blog posts like this: &#x201C;The standing of Minimalist art has benefited from the mass of stupefying triviality that has followed it&#x201D;. It&#x2019;s from Edward Strickland in &#x201C;Minimalism: Origins&#x201D;.
This sort of thing is the definition of little-m minimalism. This is really more about simplicity than minimalism. They&#x2019;re closely related concepts, to be sure, but we&#x2019;re not talking about simplicity today.
A final counterexample: Google is currently experimenting with a super stripped-down home page.
This is pop minimalism, little-m minimalism. This is simplicity as a branding tactic. It doesn&#x2019;t mean anything more than that Google&#x2019;s brand has become so ubiquitous that they no longer need explain their purpose or how to make use of their search product.
So that&#x2019;s what we&#x2019;re not talking about. Now, I want to put big-M Minimalism in context.
We&#x2019;re going to start at the beginning of the 20th Century. In the world of art, we&#x2019;re exiting the successive dominance of Romanticism, Realism, Impressionism, and early Expressionism over the course of the previous century.
This Matisse&#x2019;s &#x201C;The Joy of Life&#x201D; from 1905, a quintessential representation of Fauvism, which some consider to be the first real modern movement in 20th Century art.
Fauvism was followed by Cubism. This is Picasso&#x2019;s &#x201C;Still Life With Chair Caning&#x201D; from 1912. Cubism, of course, is about turning our notion of perception on its head. Figuring out where said head is, then, becomes something of a challenge.
Following Cubism was Futurism, an artistic movement that&#x2019;s been hugely influential on modern culture not just in the visual arts, but one that has unfortunate associations with the rise of Italian fascism. This is Boccioni&#x2019;s &#x201C;Development of a Bottle In Space&#x201D; from 1913.
Dadaism followed Cubism and Futurism, and epitomized the chaos and nihilism felt by the Western intellectual elite after the senseless, unnecessary violence of World War I. Dadaism was a philosophy for a world that didn&#x2019;t make sense, one that produced as much graphic design as art.
Hausmann, &#x201C;ABCD&#x201D;, 1923-24.
The surrealists, active from 1924 - 1945 (right up to WWII) were the introspective, dreamy cousins of the Dadaists. Where the Dadaists explored chaos and nihilism, the Surrealists dove into the subtle and otherworldly influence of the subconscious.
Yves Tanguy - Divisibilit&#xE9; ind&#xE9;finie (1942)
This is the visual anarchy of Abstract Expressionism as hurled onto canvas by Pollock in 1947. In some ways, this is the quintessential post-WWII artistic movement: violent, rebellious, energetic, transformative. It&#x2019;s the expression of a world that&#x2019;s about to reinvent itself.
Finally: color field painting probably most closely precedes Minimalism, emerging in the 1950s. The style eschews the expressiveness of Abstract Expressionism for a focus on pure abstraction. Color field painting sees the art world exercising control, getting focused, exiting the chaos of the post-WWII era.
Barnett Newman - Vir Heroicus Sublimus, 1950-51
Then, we arrive in the late 1950s and early 1960s, and to Minimalism. Big-M Minimalism, though at the time it was alternating called Cool art, Rejective art, ABC art, and Primary Structures.
This a description of the style from one of its foremost biographers, Edward Strickland.
Frank Stella, Zambezi, 1959. One of the earliest Minimalist artworks, although Stella rejected the term. Part of his series of &#x201C;Black Paintings&#x201D;. Some critics at the time noted an almost religious quality when viewing the painting. At the time, it was utterly unlike anything else anyone had seen.
An emphasis on the geometric, the serial, but note the rough, hand-painted lines &#x2013;&#xA0;nothing more than a house-painter&#x2019;s brush used to create them.
This is &#x201C;Mountain&#x201D; by Agnes Martin, from 1960. It is again geometric and repetitive, but exudes a sense of peace and calm and, despite its rigidity, an organic, hand-drawn quality. This piece, too, typifies Minimalist art.
Another distillation of Minimalist ideals.
The idea that Minimalist sculpture, in particular, is as much about the object as its preparation and presentation. The object itself is art, but not in the sense of a &#x201C;found&#x201D; object being art. The object, bare, stripped of all context, is a complete artwork.
Donald Judd, Untitled
No frame, no pedestal, very little process, nothing that clearly demarcates the object as &#x201C;art&#x201D;. It could be functional - today, it could be an Ikea shelving unit. It is incidentally beautiful.
Donald Judd, Untitled. One of my favorite pieces of art from one of my favorite artists.
Minimalism is an expression of the sublime. It is the postwar aestheticism, the anti-dada. In a universe that doesn&#x2019;t make sense, Minimalism imposes order and beauty.
This statement, from John Perreault, a period art critic, is the bridge between the relevance of minimalism to art and to engineering. As engineers we, care about the end, but most often it is our clients and customers to whom the end is most relevant. We&#x2019;re here today, on a Friday, spending our day talking about programming because we care about the means.
What we are interested in is what this artistic approach can teach us an engineers. Because software engineering is such a young discipline, we need all the help we can get. What can other fields teach us?
Much of the way we think about Computer Science comes from physical architecture. The idea of patterns as applied to programming originates with architect Christopher Alexander&#x2019;s &#x201C;A Pattern Language&#x201D;. We often describe our trade and our works in terms of physical architecture, particularly when explaining it to non-technical folks.
Mies van der Rohe, German Pavilion of the Expo, 1929, Barcelona
There&#x2019;s a monograph from a committee at MIT entitled &#x201C;The Influence of Architecture in Engineering Systems&#x201D;. In it, the above are section headings. We look to architectural metaphors to understand, design, and manage the complexity of the systems we build.
These are the problems of designing for complexity, of building our entire discipline around the management and support of complexity.
John Graham-Cumming, author, programmer, recently secured an apology from the British government for the treatment of Alan Turing.
This is the philosophy of suckless.org, advocates for simple, reliable software. They are part of a whole generation of &#x201C;computing minimalists&#x201D; who eschew the complexity of modern software.
Absorb the previous quote, and think about what you can remove to great effect.
Robert Morris, &#x201C;Ring With Light&#x201D;, 1965 - 96
A quote from Colin Percival, security and cryptography expert, currently working on the Tarsnap secure online backup service. It reinforces the negative role of complexity in ensuring our safety and security in a digital world.
Chuck Moore, creator of the Forth programming language and a masterful engineer. Any of you in the audience running Macs? The firmware that booted you into your OS, Open Firmware, is written in Forth.
Moore said in the recent &#x201C;Masterminds of Programming&#x201D; that &#x201C;operating systems are dauntingly complex and totally unnecessary&#x201D;.
Can we work within a simpler way of thinking about the world? Are we always building buildings? Does every system have to be so complex? In systems architecture, we&#x2019;re encouraged to break systems down into their component parts. But can we produce our own works that are self-explanatory, self-contained, powerful, and unable to be further reduced?
Anne Truitt, Night Naiad, 1977.
The question is: if we had a blank canvas on which to inscribe a new way of thinking about the design of software, one informed by Minimalist principles, what would we end up with?
Untitled (To Dan Flavin) by Salvatore Panatteri, probably 2007.
We need guideposts with which to approach minimalism in software. Appropriately enough, there are just three broad strategies. We&#x2019;ll explore each strategy with some examples.
Recall the earlier discussion of Minimalist sculpture, the idea of the unification of object with materials. &#x201C;reveal materials as materials&#x201D;, as one critic puts it. This is the most powerful of the three concepts we&#x2019;ll explore today.
There are strong parallels to how we organize code and data in programs.
From homo meaning the same and icon meaning representation&#x201D;.
Wikipedia: &#x201C;a property of a language in which the primary representation of programs is also a data structure in a primitive type of the language itself.&#x201D; That is, the AST is the language, and the language is the AST.
Discussed primarily in the Lisp community, but present in Lua, Io, Ioke, Factor, Prolog, and many other languages, including machine code itself. This powerful concept enables metaprogramming, which can enable powerful, terse abstractions that reduce the number of necessary lines of code.
This is from the core of the MISC metacircular interpreter. The interpreter is written in terms of itself, a property enabled by homoiconicity. The entire example is 55 lines.
An example of homoiconicity in the purely functional Joy language, taken from the C2 wiki. The specification:
1. Create a "code" data structure (block) for assigning 15 to a variable.
2. Evaluate it and view the variable's contents, which should be 15.
3. Modify that data structure to assign 37 to the variable.
4. Evaluate it and view the variable's contents, which should now be 37.
Here&#x2019;s another way of thinking about unification as it applies to programming.
HTTP has become the lingua franca not just for the client-to-server web, but even for the web inside the server clusters. Having back-end processes adhere to a RESTful architecture encourages the reuse of known, trusted pieces of software like HTTP load balancers and caching proxies that fulfill requirements &#x201C;for free&#x201D;. It keeps architectures clean, simple, and manageable.
Unification is a powerful concept. In Minimalist art, the object, its materials, and its presentation form a unified aesthetic whole. In Minimalist software, unification gives us a simple basis from which to extrapolate powerful ideas.
Recall the earlier discussion of Minimalist sculpture, the idea of the unification of object with materials. &#x201C;reveal materials as materials&#x201D;, as one critic puts it.
There are strong parallels to how we organize code and data in programs.
37signals released their Ta-Da List application in 2005, it was a mere 600 lines of code. Of course, that&#x2019;s building on a framework well in excess of 100,000 LOC, so does it really count?
We need practical strategies for code reduction in real, non-trivial codebases. In Kevlin Henney&#x2019;s &#x201C;Minimalism: A Practical Guide to Writing Less Code&#x201D; he offers a number of such pointers.
This code snippet, from a paper entitled &#x201C;Squeak: A Minimalist Syntax&#x201D;, illustrates every part of Smalltalk method syntax except primitives.
This quote, from an interview with Ken Thompson, one of the fathers of Unix, illustrates the need for a few simple, powerful abstractions in Minimalist software.
Niklaus Wirth, developer of Pascal, Modula, and Oberon, in &#x201C;A Plea For Lean Software&#x201D;. He offers nine lessons from the success of the Oberon project, and this is one of the most important, along with the use of a strongly-typed programming language. Too much hierarchy causes needless complexity and confusion.
Reduction isn&#x2019;t about doing as little as possible. It&#x2019;s about distilling complex ideas down to their essentials, such that decorations and explanations are unnecessary.
The idea of constraint, the final of our three strategies, is somewhat more abstract. It&#x2019;s more about structuring your work such that it is inherently bounded. Sometimes we don&#x2019;t choose our constraints, but more often that not we have too much leeway at the start of a project.
Talk about Twitter.
&#x201C;Minimalism results in reasonable and maintainable goals&#x201D; - suckless.org
Constraint is really about ensuring your success, at the end of the day. There&#x2019;s no shame in saying, &#x201C;we can&#x2019;t do that&#x201D; or &#x201C;we shouldn&#x2019;t do that&#x201D;. Overcommitting and over-featuring is a recipe for disaster.
So, we&#x2019;ve talked about what Minimalism is and how we go about it in software. But what are some reasons to be skeptical about this approach?
If something appears simple on the outside (Apple products, Twitter, Actors in Scala or your favorite cleanly designed API) but has complex underpinnings, is it still simple?
A perpetual problem: it&#x2019;s rarely turtles all the way down, and nobody really knows everything about computing. Is there anyone in this room who can say with confidence he knows the lowest low-level and the highest high-level concepts?
C# author: &#x201C;It&#x2019;s all about &#x2018;do more with less&#x2019;. It&#x2019;s not about doing more with more with a simple layer on top.&#x201D;
Say we figure out how to build big, complex system reliably. Arguably, inside of companies like Google, they have. If that knowledge becomes widespread and institutionalized, is a Minimalist approach still valuable? If we can tame complexity, is Minimalism anything more than a stylistic preference?
This is probably the most worrying point.
James Rumbaugh, the designer of UML says that &#x201C;life and computing are too complicated for simple solutions&#x201D;.
We&#x2019;ve all the had experience of writing some beautifully simple piece of code, only to find out a few hours later that it doesn&#x2019;t really do the job. Likewise, it&#x2019;s exhausting work making a business of selling simplicity - it&#x2019;s not demonstrably the way to be a market leader in this industry.
Let&#x2019;s talk about a future of computing in which Minimalism can play a crucial role. In some ways, desktop and server computing are beyond help. They missed their chance to be Minimalist, and we&#x2019;re not reinventing that wheel anytime soon, realistically. But where can we apply these principles such that they&#x2019;ll make a big impact?
Mobile technologies offer us an opportunity to reinvent computing. Developing countries are experiencing information technologies for the first time, and there&#x2019;s no reason why they have to suffer the same computing mistakes that the developed world has. We can build simple, mobile, goal-oriented devices for the next generation of human communication.
One way of considering the problem of modern desktop computing is that we&#x2019;ve failed to break down the problem sufficiently. The promise of ubiquitous computing - intelligent devices everywhere around us - is a chance to reconsider computers as simple, single-purpose devices.
The artists all grouped together by art historians as the Minimalists never really considered themselves adherents of a common school or approach. They just did what they did, and most of them fought the Minimalist label for the rest of their careers.
If you liked what you heard here today, I&#x2019;m glad, but the point isn&#x2019;t that you should go out and brand yourself a Minimalist. The point is to take the ideas that work for you and make the world a better, simpler place.