This article called Genetic Programming: A Novel Failure probably sums up my thoughts best. Genetic programming is only a slight variation on solution searching algorithms.  Based on my reading, most work in the field has focused on how to make systems converge on a solution more quickly, i.e.: improving efficiency.  This seems wrong, or at best premature.

We live in the 21st century. We have more CPU cycles than we know what do with.  Where are the systems that are wide but shallow? The ones that are really non-determinstic and will generate truly surprising results? We should be wasting cycles exploring new possibilities, not generating new solutions for known problems.

The free ebook, A Field Guide to Genetic Programming, is a great primer in the topic.  I read through most of it the other night.  My biggest frustration is that almost all genetic programing systems focus on evolving syntax trees, usually some form of Lisp or it’s semantic equivalent. I see why people would do this. Lisp code is easy to manipulate programmatically, so evolving it should be simple as well.  There are other kinds of systems using different gene encoding, such as image arrays and direct byte code. However, these appear to be far in the minority. The Field Guide has a chapter on the topic, listing several alternate systems. The fact that these systems receive a single slender chapter while the rest of the book covers syntax trees gives you an idea of how under-explored the topic is.

When I first heard of genetic programming I imagined having a sequence of simple instructions that could be mutated. The instructions would be extremely simple and limited, perhaps more simple than most assembly languages. Evolving syntax trees certainly does let you make progress quicker, but the generated solutions will be limited to the underlying tree language. Our genetic beasties will never evolve novel flow control systems, or invent a crazy kind of memory register. ASTs are great if we want to produce a human readable program as the result, but it still feels limiting.

I would like to see a system that is as open ended as possible.  Create a system of small instructions of uniform length that can only manipulate basic storage and do simple math.  Then give them as much freedom as possible. Let them live in a wide fitness landscape. A digital environment with a huge number of potential ecological niches. Ideally we could take this to the next step and give our digital creations physical bodies so that they may evolve targeting real world constraints.  Evolving a robot’s brain seems far more interesting than figuring out how to gain 10 milliseconds trading stocks.  Of course we lose rapid iterations by running them in the real world, so it my be better to run them in a simulation of the real world at many times normal speed.  I imagine we could build self driving cars this way, starting with bots that play racing games then upgrade them to working with real world footage from actual self driving cars.

Am I wrong? Is there cutting edge genetic programming that is truly open ended? What successes have they made?

Please send feedback and comments to my twitter account @joshmarinacci instead of on the blog. Thanks!

Lets step back a minute and consider what Adobe actually does.  They have some enterprise backend products for document management and (amazingly still used) a server side platform in Cold Fusion.  They have some cloud products (Acrobat.com) built on the the Flash platform. Then there is Flex, an enterprise application platform designed to steal Java developers from Sun, as well as a mobile advertising and analytics platform (Omniture). And of course Flash for mobile.

Oh yeah. And they also make Photoshop, Illustrator, Fireworks, and a bunch of other industry leading graphics and content creation tools so pervasive that some have called them the Adobe tax.

This really sounds like two different companies, doesn’t it?  I’m not exactly sure when, but at some point Adobe strayed from focusing on high quality content creation tools for designers and artists. They entered the “platform space”: trying to be an enterprise company, a software as a service company, and own the mobile content market.  That’s a whole lot for any company to do, especially one who traditionally focused on content creation tools.  In the end I think it just became too much for one company to do, and it takes away from the thing they are good at: killer tools for designers.  If killing of mobile flash lets them focus on their core competency then this is a good thing for everyone.

Adobe is known for tools used by professionals to create content. The Flash designer tool is used by professionals to create animated interactive content. Currently, the format of the final output is a SWF file. Do the purchasers of this tool care? Not really. Flash designers want to create content that is viewable by the most people. The audience wants great content accessible from the most devices. Neither of these two groups of people gives one whit about the actual format of the bits.  Flash, the runtime, was simply an end to a means.  With HTML 5 technologies becoming viable for interactive animated content, the Flash designer tool can simply output a new binary blob to be uploaded onto web servers.  The designers won’t care, the audience won’t care.  Everyone will get on with making/viewing their content and Flash Designer CS 22 will sell millions of copies.  This really isn’t a big deal.

Well, except for one group of people who really truly do care about mobile Flash: the makers of iPad competitors. Apple’s refusal to allow Flash onto the Safari mobile browser created a market opening for a device that *would* play Flash.  While it was never a big factor for webOS, it was a flagship feature for the BlackBerry Playbook and various Android Tablets.  They’ve now lost a checkbox in their feature war with the iPad.

No matter. The world will move on.  The mobile web is built on HTML 5 standards. And in 5 years the mobile web will simply be the web; which may foretell the end of the desktop Flash plugin as well, but the end result is the same. Adobe will continue to sell world class content creation tools. Tools which output whatever format the world actually wants. And now, finally, the world wants HTML 5.

Long Live the Web. Long Live Adobe.

Well… no. We definitely are going to see a huge shift in the industry over the next couple of years, but there will not be just one or two OSes controlling the market. And laptops won’t be obliterated by tablets any more than TV destroyed the movies and radio. We won’t see Mac vs PC again, or desktops vs Apple IIs. 2014 won’t be like 1984.

First a disclaimer. These are my opinions, not the opinions of my team or employer. I work in Developer Relations. I have no knowledge of long term HP strategy, nor do I have any influence on it. This is simply the ramblings of an long time computing observer.

Tablets are no substitute for ‘real’ computers

Let’s tackle these issues separately. First the claims: ‘tablets suck for real work’, or ‘I would never use one. They are too limiting’, or ‘They are only for content consumption’. What are we talking about?

By tablet computers I mean things like the TouchPad or the iPad. These are devices which run a non-general purpose OS. There is no exposed filesystem. Apps are sandboxed and safe. A PC is a desktop or laptop running a general purpose OS like Windows, Mac OSX, or full Linux distros. Regardless of the form factor (touch vs keyboard), these are fundamentally different kinds of devices. Now the claims:

Tablets are too limited compared to a real computer. Yes the current generation of hardware is limiting, but it’s going to get better; and fast. My top of the line computer less than 10 years ago had a 400mhz processor with 64MB of RAM, no GPU, and 8 GB of slow disk storage. Pretty much all tablet computers far exceed this already, and will soon support printing, directly controlling hardware in your house, and be first class citizens of the network (assuming Apple ever lets you jettison iTunes from your ‘real’ computer). They will all get better, and quickly. Especially when there is competition.

Tablets suck for real work: Yes, they are primarily designed for content consumption and tasks that require typing a paragraph or less. But guess what: that’s a lot of stuff. In fact that’s what 90% of people do 90% of the time on their computers. Most people don’t write more than a paragraph at a time on their real computers. Most people surf the web, check Facebook, place games, pay their bills, and write a few short emails. A tablet device that can do 90% of what they need with less fuss and less cost is a big deal. A really big deal. Half of people could never learn to drive before the automatic transmission was invented. Yes, it’s that big of a deal.

‘I would never use one. They are too limiting’. That’s very true. If you get nothing else from this essay, I hope you remember one thing:

They aren’t built for you!

These things are built for the 90% of people who don’t need everything a full PC does. By definition, if you are reading this blog, then these things aren’t built for you. You are a programmer or writer or artist. You need a ‘real’ computer. In 10 years (probably far less), you will own a tablet computer, but it won’t be your only computer.

In ten years I will still have a laptop with a real keyboard, possibly a disk drive, and most certainly an exposed filesystem with regular installable apps. It will still have a command line. (bash4eva!) I’ll certainly use a tablet computer as well, but it won’t be my only computer. However, for 90% of people, the tablet will do everything they need. It’s built for them, not us.

OS Wars

Now that we have the audience for a tablet out of the way, lets look at the OS wars. There’s a lot of talk that we’ll have just iOS and Android. That they have an insurmountable lead. That the market wants one boutique option and one mass market indistinguishable option, just like MacOS and Windows.

I really don’t think this is the case. I don’t think any OS will have more than 25% market share in 10 years. Despite the similarities, the mobile OS market is very different than the PC market. Why? Well, let’s compare the world of 1984 with the world of 2011.

• Hardware: In 80s and early 90s you bought a PC in a computer store or maybe a department store like Sears. In 2011 you buy a phone in a cellphone carrier store, or on the web. You buy a tablet computer in a cellphone store or a mass market retailer like Target. These stores really didn’t exist 30 years ago. Getting distribution for a device is very different now.
• Apps: In 1984 you bought apps on floppy disks, wrapped in boxes, sitting on a shelf in a computer store. Or maybe Sears and Toys’R'Us (for games at least). There were no ‘app stores’. Today, mobile apps are almost always bought in a store provided by the mobile platform itself. It doesn’t matter if a retail location wants to carry your apps or not. No one has to fight over physical shelf space. The economics are fundamentally different.
• Advertising: In 1984 computers were mainly advertised in computer magazines and newspapers. Remember those? Those things that no one my age reads anymore? (I’m 35 by the way.) Now we read and shop online. Or on our phones. Or get recommendations from friends on Twitter. And mobile devices are advertised on television. Advertising has changed. People find out about products in fundamentally different ways.
• Compatibility: In 1984 software compatibility mattered. Software was hard to write, required huge lead times, couldn’t be easily updated, and speed was of the utmost importance. Only the biggest apps would be on more than one platform, so getting apps on your OS was a big deal. IBM went to a lot of trouble making OS/2 Warp work with Windows 3.1 apps, for the sake of compatibility. Apple created expensive Apple II plugin boards for the Mac, all for the sake of app compatibility.
  
  Today most of the apps we run are backed by platform independent web services. Only the client app is different. And even that is easier thanks to open standards, modern programming languages, and the web. webOS has smaller market share than iOS and Android, but we still have Facebook, AngryBirds, and about 2 million Twitter apps. And you can view all the same websites. Compatibility simply isn’t an issue anymore.

The economics of mobile operating systems are fundamentally different than the desktop wars of olde. To say we are in for a repeat of Mac vs Win is like saying the two world wars were identical because they both involved Germany and had the word ‘World’ in their names. Well the world has changed.

I think there will not be any single OS winner. Instead it will be more like cars. Many different models and vendors to cater to different tastes. They each have their own colors, addons, and spare parts; but they all drive on the same roads (the internet) and all take the same gas (webservices). 2014 simply won’t be like 1984. And that’s a very good thing.

As always, I am writing this as Josh the blogger. These are my opinions alone and do not reflect the opinions of Palm Inc.

• webOS devices are part of the web, not tethered to a desktop computer.
  You install apps through the web. You your data is backed up to the web. OS updates come through the web. Your address book is a merged view of your contacts living in the web. You never have to sync to a desktop computer. I know some Pixi users who have never once plugged their phones into a computer, because their phone is already a part of the web.
• The webOS treats it’s users like grown ups: they can install any apps they want. What if the app duplicates a built in app? Fine. What if the app isn’t in the on device catalog? Fine: you can install apps from the web or beta feeds without any restrictions and do the marketing on your own. What if the app hasn’t been reviewed, came from my cool programmer friend, and might hose my device? Well, if you enter the developer code into your phone then you’ve accepted the risk and can install any app you want. There’s a whole community of people making cool but unauthorized apps. They are called the Homebrew community, and Palm encourages them. You’re an adult. You can make the decision of what to install on your phone.
• The webOS lets you use any language you want to develop apps.
  While Palm doesn’t provide tools for languages other than JavaScript, C, & C++, there are no restrictions against using any other language. Our new PDK gives you a clean POSIX layer with direct & standard access to input (SDL), the screen (OpenGL), and device services (API bridge). There’s nothing stopping you from porting a C# compiler or a Lua interpreter. Developers are free to use whatever tools they wish. The results are what matter. Good apps are good apps.
• The webOS doesn’t have fragmentation.
  All webOS devices run the same OS, regardless of form factor. They are all updated over the air, for free, in all countries and carriers. This means that 99% of webOS devices have the current version of the OS within a few weeks. There is no fragmentation of the operating system across devices or form factors. This lets developers focus on making great apps, not waste time supporting 18 versions of the OS.
• The webOS is built from the DNA of the web. Yes this includes using HTML, JavaScript and CSS as the primary application development layer, but it’s more than that. I can just start typing to have my question answered by wikipedia. The address book contains your contacts that live on the web. If my wife changes her Facebook profile photo, my phone is automatically updated. I can write an app that links to other apps through Javascript calls. The web is about connections to the people and services you care about, not just HTML pages. So is the webOS.

At Palm we care greatly about the end customer experience. We are also developers, so we care greatly about the developer experience. And most importantly, we don’t see the two at odds. Happy developers create great apps that create happy customers. It’s a win, win. That’s why we are doing everything we can to make happy developers. We don’t always do everything perfectly, but when something is broken we do our best to fix it and be transparent. It’s how the web works and it’s how the webOS works.

So, as a developer, I hope you’ll think about the benefits and freedoms of the webOS, and consider it for your next mobile application.

references

• Palm’s Developer Center
• PDK: Plugin Development Kit
• Palm’s Hot Apps competition
]]> http://www.joshondesign.com/2010/04/15/webos-the-os-built-from-the-web/feed/ 6 http://www.joshondesign.com/2010/03/08/why-do-alarm-clocks-suck/ http://www.joshondesign.com/2010/03/08/why-do-alarm-clocks-suck/#comments Mon, 08 Mar 2010 16:42:05 +0000 Josh Marinacci http://www.joshondesign.com/?p=162 It’s been a month since I posted so I’d say it’s time for a rant. I’ve been traveling a lot lately so the object of my wrath this week is alarm clocks. Most specifically the alarm clocks in hotel rooms, but home use clocks don’t get off easy either.

Alarm clocks have one purpose in life. There’s only one thing they need to do to be considered a success. It’s not to ‘tell the time’. That’s a nice bonus, but the purpose of an alarm clock is to get you up in the morning. This is doubly so for alarm clocks in hotel rooms. If you are sleeping in a hotel it’s likely because you have traveled somewhere to do something, and it’s also likely that you want to do that something at a certain time; hence a clock to wake you up so that you may do that something without being late and getting fired. A device which cannot reliably complete this basic task is simply a failure, and not worth being made or purchased. End.

The Bad

How are they bad? Oh, please let me count the ways. (cue evil grin of glee).

First, most clocks tack on a ton of extra features, like iPod integration. Then they massively overload these features onto a small number of buttons by using modes. Many will have a single set of buttons to set both the time and the alarm, with a switch to toggle between the modes. Modes aren’t a great idea in desktop software when you have a huge screen. They are even worse on the limited user interface of a clock with a fixed LED readout. Which mode am I in?. Did I set the time or the alarm? After I’ve set the alarm how do I know which mode I’m in now? Did it switch back automatically somehow or do I need to press another button?

Some clocks use quasi-modes to get around these problems. A quasi-mode is like a shift key: a button you hold down to temporarily enter a new mode, then release when you are done. Not a bad idea for a computer with a full keyboard. Absolute madness on a clock where you must hold the mode button with one hand and try to set the time with the other.

Even worse, some clocks put the mode button on the front instead of the top. Clocks, typically being small, are lightweight. So pushing from the front will shove it right off the nightstand. Now you have to use your third hand to hold the clock on the table, while accomplishing the aforementioned gymnastics.

Now do all of the above right when you go to bed… when you are sleepy.. and it’s dark… No wonder so many people opt for a wakeup call or use the alarm on their phones. I always completely unplug the clock just in case the guy before me set it to ring at 3am. (yes, this actually happened to me)

Now suppose this is a clock you’ve never used before (very likely, since every hotel bought from a different supplier). Now you have to learn how to use this particular clock. Some come with their own instruction books. It’s madness, I tell you. Madness!

The Good

Only once in my life have I found a non-sucky hotel alarm clock. It was in my room at a very nice hotel in Tokyo. Here’s what it looked like.



Simply beautiful. Two buttons to change the alarm time up or down. One button to arm the alarm and snooze. AM/PM is indicated with real words.

How do you set the time? You don’t. I’m serious, there were no buttons anywhere to set the time or customize the alarm. Either the time is set via radio or there’s hidden controls locked inside somewhere. Maybe they use something wireless through that little transceiver on the front. The point is: I don’t have to care. The only thing I care about is setting the wakeup time, so that’s all the device lets me do. They also don’t put in a radio since that would require controls to change the station. If you are in an international hotel in Tokyo you probably wouldn’t understand the radio anyway, so jettison the feature. Simplify, simplify, simplify.

I don’t know why this clock isn’t used in the US. Perhaps it’s because everything in Tokyo is from the future. Perhaps in another five years will will all be enjoying alarm clock bliss. No wait.. I took that photo five years ago! Yeah, we’re doomed.

The Ugly

So why do hotel alarm clocks suck so badly? I think there is two reasons. First, these devices are mass produced overseas so they cut corners wherever they can to save costs and increase profits. If you can make a clock which uses two buttons to set the time instead of three buttons, then you might save five cents. Across millions of units that adds up to a lot of money.

Second is the buying decision. I’m not 100% sure, but I suspect that electronic clocks in general have become commodity products. The guts are a single chip that costs around 25 cents, and it’s probably the same chip used by everyone. So the various clock makers compete with each other on price, features, or by simply looking cool. Now, don’t get me wrong. There’s nothing wrong with competing on price, features, and visual design. But what suffers in this competition usability.

I think usability suffers because of how clocks are bought. When you go to the store you can see the price, features and design from outside the box. What you don’t see is how difficult it’s going to be to actually set the alarm on the damn thing. The buying decision doesn’t include usability. And since manufacturers optimize for the buying decision, usability gets dropped on the floor. C’est la vie.

The Lesson

I don’t know how to fix the economics of alarm clock design, but the moral of today’s story when applied to other products is simple: when you make a product you must design for the primary use case first and foremost. In this case that means setting the alarm reliably so you can wake up in time to not get fired. Everything else is secondary.

Okay kids. Time for work. I woke up late!

]]> http://www.joshondesign.com/2010/03/08/why-do-alarm-clocks-suck/feed/ 3 http://www.joshondesign.com/2009/11/27/the-success-of-multitouch-is-a-lie/ http://www.joshondesign.com/2009/11/27/the-success-of-multitouch-is-a-lie/#comments Fri, 27 Nov 2009 20:51:55 +0000 Josh Marinacci http://www.joshondesign.com/?p=77 Listening to some podcasts about mobile devices I heard over and over statements like “iPhone changed the world with multi-touch” and “Android could compete with Apple if it had multi-touch.” This simply isn’t true. Okay, while perhaps not a lie, the success and value of of multitouch is extremely overrated. In fact, the iPhone barely uses multi-touch!

Don’t believe me? Think back to the iPhone of 2007 when it launched. Or just look at the Apple provided apps in today’s iPhone (since not much has changed). How many of these built-in apps use multi-touch? I can only count 3: Maps, Photos, and Safari. All three of them use multi-touch in a single simple way: zooming in and out. You could make a non-multitouch iPhone by simply providing a zoom-out button for these three apps (and a quick-tap for zooming in).

Very little of what made the iPhone interface so revolutionary was multi-touch. What made it so great was the focus on using a single finger for virtually all interaction with the device. The designers at Apple decided from day one to make a device that was finger centric. This means UI controls that are large and require only a tap gesture to activate. Swipe gestures are used for navigation. And that’s it. Taps and swipes with a single finger. That’s what made the iPhone so great, not multi-touch (or the accelerometer, for that matter).

Oh, and one more thing made it great. The large screen with a capacitive touch sensor. Older touch enabled devices (like my old beloved Treo) used resistive screens which were far less accurate and required you to either push hard with your finger or use a stylus. A Treo with a capacitive screen could have supported an iPhone like interface with ease (even on the much slower hardware available to it at the time).

While the hard glass iPhone screen does support multiple touch-points at once, that’s not what made it a success. It’s designing an interface and device from the ground up for finger-based touch interaction that changed the mobile device playing field. Multi-touch is simply a red herring.

Finger photo used under Creative Commons from Flickr user bayat

]]> http://www.joshondesign.com/2009/11/27/the-success-of-multitouch-is-a-lie/feed/ 7 http://www.joshondesign.com/2009/10/19/in-defense-of-wires/ http://www.joshondesign.com/2009/10/19/in-defense-of-wires/#comments Mon, 19 Oct 2009 04:56:48 +0000 Josh Marinacci http://www.joshondesign.com/?p=37 An essay in which I expound upon the benefits of the lowly wire and resist the temptation to wireless-ize the world of personal gadetry.

This weekend, in a futile effort to preserve my back and wrists, I’ve retooled my home office by including picking up a new mouse and keyboard. The only thing available at the local store was wireless, either bluetooth or using a proprietary dongle. While reasonably nice to use from an ergonomic standpoint they immediately began having interference with my network, including dropping or repeating keystrokes and mouse clicks. After 2 days of frustration I returned them. Then, after searching 3 stores to find a decent wired keyboard, I gave up in frustration. Wireless all.

The Good

Wireless sounds like a good idea. The promise of “No wires!” means no tangles, no restriction of movement, and no ugly cords all over your desk. This is especially attractive when you use a laptop 100% as I do. The last thing I want on an airplane is a wire to get tangled up in the seat. The wireless devices all look quite sleek and futuristic. And the accuracy of modern laser trackers on virtually all surfaces is quite simply astounding. Wireless promises a trouble free computing experience.

The Bad

For all of the good, there’s actually a lot of problems with wireless devices.

First: all wireless gadgets must have batteries, which means one more thing to monitor, charge, and replace. Next, the minute you have an active network over the air you have to worry about eavesdropping. That means security layers, network protocols, and the bane of bluetooth: pairing. The act of connecting two devices which are a scant 2 feet from each other simply isn’t worth the pain. I returned three bluetooth headsets over the years due to pairing issues.

Once you get your device set up and authenticated you still must worry about interference. I found that my keyboard would drop key presses if I was doing my hourly backup over the network at the same time. Not doing two things at once is the opposite of progress. And finally, to add insult to injury, wireless devices cost more.

Now let’s consider the alternative.

Wires

Wires are simple to use. You attach a wired mouse to your computer by plugging it in. Through the magic of USB, the device is immediately detected and the driver installed. Plus you get power for free, so no more batteries to replace. Nothing goes over the air you don’t have to worry about eavesdropping, so no security system and no pairing. No RF transmission means no interference with the 4.8 billion other wireless devices in my house.

In addition to the security aspects wires are usually faster and cheaper. USB 2.0 is far faster than even the latest Wifi N standards, which I suspect is why Apple doesn’t sync the iPhone over wifi. And the cost of course is fantastic. No extra batteries and radio transmitters makes any gadget cheaper to produce.

The costs of wireless (financial, technical, and mental) are worth the benefits in some situations. But for lots of things: just go with a wire. It works.

]]> http://www.joshondesign.com/2009/10/19/in-defense-of-wires/feed/ 9