Jared Tarbell’s Gallery of Computation »
A beautiful set of digital artworks, mostly created in Processing. Source code is included.
My favourite is Substrate, which I’ve been running as my screensaver. It looks like crystals growing, or an aerial view of a city developing.
Review: Programming Collective Intelligence
Programming Collective Intelligence is a book about applying data mining techniques to analyse collections of data. There is submerged information in Ebay prices, in Facebook profile networks, in collections of movie reviews, in news sites, in the stockmarket; this book by Toby Segaran shows ways to extract, visualise, understand, and predict that information.
Each chapter explains and explores a different data mining algorithm, and builds up a working example in Python, while presenting different methods and parameters of the implementation. I hadn’t really worked with Python before, but found the code easy to follow, and picked up some interesting Python idioms that I haven’t seen in other languages before. Chapters end with a set of exercises to follow that build your understanding.
As you follow the examples you build up a reasonably generic code base that allows you to swap in and out different implementations, and reuse previous code to add to new applications.
The examples use live examples from the web: sites like Ebay, Facebook, and Yahoo Finance, and this makes the book more interesting and the results more visceral than some other books on the subject which use more contrived or obscure examples. Even though there is a strong web (or web 2.0) focus on the examples, the methods and the understanding is useful for a whole range of applications.
Some of the topics covered:
- Bayesian classifiers to detect spam, or to file news articles into site sections
- Hierarchical and k-means clustering to discover groups of similar items in massive sets
- Euclidiean distance, Pearson Correlation Coefficient, Tanimoto Coefficient: ways to measure the distance (or difference) between items
- Neural networks to predict user behaviour and improve search result ordering
- Optimisation methods like hill climbing, simulated annealing, and genetic algorithms
- Non-negative matrix factorization
- Support vector machines and kernel methods to go where linear regression can’t
I found it exciting to read — it’s one of those books that give you a whole bunch of new ideas for things to build as you read it. The presentation is very good: no background is assumed, and it doesn’t talk down to those more experienced.
Recommended.
Game/Space: An interview with Daniel Dociu »
BLDBLOG interviews game world designer Daniel Dociu:
Are there specific architects, historical eras, or urban designers who have inspired Dociu’s work? What about vice versa: could Dociu’s own beautifully rendered take on the built environment, however fantastical it might be, have something to teach today’s architecture schools? How does the game design process differ from – or perhaps resemble – that of producing “real” cities and buildings?
Of course, there are many types of games, and many types of game environments.
There’s some really beautiful concept design artwork included in the interview.
150 clocks working as one »
The starting point with this project was a personal study about form & time. I put together more than 150 individual clockworks and made them work together to become one clock. I show the progress of time by letting the numbers be written in words by the clockworks. Reading clockwise, the time being is visible through a word and readable by the completeness of the word, 12 words from “one” to “twelve”.
Via Long Views
NSW to review freedom-of-information laws »
NSW’s FoI laws will get an independent review by the state ombudsman after the State Government repeatedly failed to act on recommendations that it conduct one. The laws haven’t had a review since the legislation was written in 1989, despite every state ombudsman since calling for a review.
Mr Barbour did little to hide his irritation with the NSW Government in a release yesterday. “For almost 14 years, each NSW Ombudsman, including myself, has called for an independent and comprehensive review of the FoI Act.”
“In the absence of the NSW Government initiating a review of the act, I have decided to conduct my own independent review.”
Last year’s NSW Ombudsman annual report highlighted the abysmal state of NSW’s freedom-of-information laws. The report found that applications released in full in NSW dropped from 81 per cent in 1995-96 to 52 per cent in 2005-06. This compared with the Commonwealth where 78 per cent of requests were granted in full.
Photos: The Wish Kin launch party
My sister’s first published novel, The Wish Kin, was released to bookstores at the start of the month, and the launch party was held last Tuesday at the Lord Dudley in Paddington. I’m so excited and happy for her. The book is great — I strongly encourage you to read it.
Fourteen-year-old Colm Bell and his 11-year-old sister Lydia wake to the smell of smoke and the sound of gunfire -– raiders are attacking their home. They grab their backpacks -– prepacked for such an emergency -– and head to the tunnel and the hills, following the escape route they have practised every day for their young lives. They don’t look back.
This is a futuristic novel set in Australia when the earth has been plundered and natural resources depleted. Society has broken down completely, and small groups of people live in suspicion, desperately hording their meagre supplies.
Colm and Lydia are part of a generation of children who have never seen rain. As they wander through this Mad Max landscape, moving north to where they hope to meet up with their father, they hear of a great underground fire that is slowly incinerating the earth. They learn of the Wish Kin – people who are said to have the power to heal the earth at an event called the Rekindling. And they are captured by the Clan – a sinister group determined to rule the land and its people.
This is a haunting, lyrical story –- of children abandoned to a planet that no long sustains life, their journey alone through incredible peril, and the realisation of the part they must play in this brave new world.
Joss Hedley grew up on the South Coast and is the recipient of two residential awards from Varuna, The Writers’ House, in Katoomba. She lives now in Sydney where she tutors in performance studies at university. The Wish Kin is her first novel.
Estimating salaries by comparing job ads »
This is clever: job classified aggregator Indeed estimates a job’s salary when it lacks that data by comparing the text to ads that do have a salary posted.
When people search for jobs, they want to put in a salary floor. They don’t want to see jobs that don’t at least pay a certain amount. Problem is most job listings on the Internet don’t include salaries.
What Indeed did was built a system that estimates salaries on all jobs.
We use a proprietary methodology based on an analysis of similar job listings that include salaries. We start by extracting salaries from all job listings containing this information - about a fifth of the total - and then estimate salaries for the rest.
Example: CFO jobs in NYC that pay more than $200k per year
Fred Wilson: Adding intelligence to search
Monocole design notes »
Dan Hill has written a detailed piece on the design and production of the Monocole website. It’s a great insight into the team’s creative process — lots of sketches and mockups, and a review of the project path and decision points.
Amazon adds persistent storage to EC2 »
Amazon is adding persistent storage as an option to EC2 — currently it’s in private beta.
Previously, disk storage on an EC2 was transient:- when the machine was shut down or crashed, it felt like a hard drive crash. (And you’d lose your IP address too, but Amazon added static IPs a little while ago too.) The path to reliability was to use S3, but that can’t be mounted as a native file system.
The persistent storage appears as a raw, mountable filesystem that needs to be formatted. You’ll be able to make a quick snapshot of the data, for backup. No word on pricing or its performance, but you’d expect it to be aligned with S3.
There’s been the option of mounting S3 in EC2 using davfs, which mounts with WebDAV, but that’s a bit of a hack and one wonders what the performance would be like.
Why are mouse sensors in the middle of the mouse?
(When they should be at the front)
In 1987, Microsoft commissioned Matrix Product Design to design a mouse. A mouse that would beat anything that had proceeded it. It was to be clearly different and superior to the clutter of beige mice that existed at the time.
To complete the brief, Matrix found partners in ID TWO and David Kelley Design.
In a GUI, the mouse controlling the cursor is the fundamental link between the user and the computer. A strong connection is created for the user when the mouse performs well. The emotional attachment created between operator and system is strongest when the system does precisely what the operator requests. Imprecise controls make the operator feel that they are not in control: and so there’s a compelling case to get the design of the mouse right.
As part of the design process, Matrix made a series of prototypes to find the right form for the mouse; a form that looked right, felt right, and above all, gave the user real control and precision of the mouse and cursor.
Matrix conducted trials of these prototypical forms, using a series of tests written as HyperCard stacks. They tested how quickly users could home in to a marked position, how precisely they could click on a specific target, and how accurately they could move a cursor through a maze.
They found that the form of the mouse — how it felt in the hand — had a direct impact on how well the user could control the mouse.
But even more important, particularly for the precision and maze tests, was the location of the mouse ball.
Matrix found that the best place for the ball was up front as far as possible between the users thumb and forefinger. The forefinger can be controlled very precisely — much more so than the wrist and forearm. Matrix found that users would move their wrist and arm to move the cursor are large distance, but for fine control relied on the thumb and forefinger.
At this time, mouse manufacturers tended to position their balls far to the back of the mouse. It hadn’t been placed there for the benefit of the users, but for the ease and cost-efficiency of manufacturing. The ball and motion sensors were large and took up a lot of room, and the front of the mouse is taken up with buttons and the cord: simpler by far to push it to the back.
And so despite the extra costs of manufacturing, the final design of the Microsoft Mouse had the ball quite far forward. Matrix published their findings, and soon all the other manufacturers followed suit.
Nowadays, however, all the mice I own have their sensors placed roughly in the middle of the longitudinal axis, and randomly on the transverse axis. Including mice from Microsoft.
At first blush, the position of the ball doesn’t seem to make much difference to this fine control. After all, the user can move the mouse with their thumb and forefinger wherever the ball is placed. Push up and the cursor moves up; left and it moves left.
But moving just the forefinger — not the whole hand, or bunching all the fingers, but just the forefinger — actually makes the mouse rotate, particularly for left- and rightward movements. If the sensor is up the front of the mouse, that rotation of the mouse will give the user the desired outcome: the cursor will accurately move to the left or right, or back and forth. But if the sensor is in the middle of the mouse, then all of that fine control is lost: no or little movement will be registered. Worse, if the sensor is towards the back of the mouse, the direction of the cursor’s movement will be reversed.
I don’t know why this change has happened. I feel that an important understanding has been lost: that we have taken a step back.
The cost basis doesn’t seem to be there: laser diodes and the associated componentry are much smaller now than the old rubber-covered steel balls and the sensors to track their movement. True, up at the front of the mouse there’s a scroll wheel to get in the way, but the sensor can easily be placed to the left of that.
It simply seems that designers and manufacturers have, over the years, forgotten about the benefit of putting the sensor up front, or have placed precision and control further down their list of priorities. I hope that this isn’t the case: that newer research has shown that the current placement is the correct placement, or that something else has changed over time. But if that’s not the case, then I hope that some design team will rediscover either the principle, or the findings — so that we can continue to strengthen the connection between the user and the computer.
Further reading
Bill Moggridge, Designing Interactions: a fantastic book that looks at how designers design how we use computers, systems, services, and gadgets.
The earlier development of the Macintosh Mouse; also by the partnership that would become IDEO.
This partnership would later gel to form the well known design company IDEO.
HyperCard was an Apple product that was a precursor to hypertext, which included an inbuilt scripting language.
And being one not to throw things out, I own a few.
See the dissected Mighty Mouse for an idea of how small these components are. The Mighty Mouse is one of the few mice available that have the cursor anywhere near the front; unfortunately other factors including the poor button design make this mouse difficult to use.
If you know of any such findings, do get in touch.