I'm currently splitting up some MindTouch Dream functionality into client and server assemblies and running into a consistent pattern where certain classes have augmented behavior when running in the server context instead of the client context. Since the client assembly is used by the server assembly, and these classes are part of the client assembly, they do not even know about the concept of the server context. Which leads me to the dilemma, how do I inject different behavior into these classes when they run under the server context?

Ok, the short answer to this is probably "you're doing it wrong", but let's just go with it and accept that this is what's happening. Let's also assume that the server context can be discovered by static means (i.e. a static singleton accessor) which also lives in the server assembly.

Here's what I've come up with: Extract the desired functionality into an interface and chain implementations together. Each class that needs this facility would have to create it's own interface and default implementation and looks something like this:

public interface IFooHandler {
  ushort Priority { get; }
  bool TryFoo(string input, out string output);
}

TryFoo gives the handler implementation a chance to look at the inputs and decide whether to handle it, or whether to pass on it. The usage of collection of handlers takes the following form:

public string Foo(string input) {
  string output = null;
  _handlers.Where(x => x.TryFoo(input, out output)).First();
  return output;
}

This assumes that _handlers is sorted by priority. It return the result of the first handler report true on invocation. Building up the _handlers happens in the static constructor:

static Bar() {
  _handlers = typeof(IUriTranslator)
    .DiscoverImplementors()
    .Instantiate()
    .Cast<IUriTranslator>()
    .OrderBy(x => x.Priority)
    .ToArray(); 
}

where DiscoverImplementors and Instantiate are extension methods i'll leave as an exercise to the reader.

Now the server assembly simply creates its implementation of IFooHandler, gives it a higher priority and on invocation checks its static accessor to see if it's running in the server context and if not, lets the chain fall through to the next (likely the default) implementation.

This works just fine. I don't really like the static discovery of handlers, and if it weren't for backwards compatibility, I'd move the injection of handlers into a factory class and leave all that wire-up to IoC. Since that's not an option, I think this is the most elegant and performant solution.

It still feels clunky, tho. Anyone have a better solutions for changing the behavior of a method in an existing class that doesn't required the change to be in a subclass (since the instance will be of the base type)? Is there a pattern i'm overlooking?

After writing "Maybe it’s time to stop pretending we buy software?", I talked to a couple of gamer friends about their purchasing habits. The concept of DLC that has a strong "withheld content" smell came up and whether this was a "first buyer bonus" or "selling crippled products" had no straight forward and agreed answer. But what did emerge is that pricing of games is a key factor in why used sale and purchase are considered a right. The sentiment that at ~$60 there is an expectation that the game has a residual value you can recoup should the game not justify itself as a keeper. Which, of course, itself is an indicator that our usage patterns of games are much closer aligned with a rental than purchase relationship. In particular, one friend uses Amazon's Trade-In store as a form of game rental. Depending on how many games you play, it is a better deal than Gamefly.

Now it turns out that arguing about used games and whether they are crippled or not may not even be an issue in the future. Ars Technica did a great summary called "No, you don't own it: Court upholds EULAs, threatens digital resale" of the US Court of Appeals for the Ninth Circuit ruling re: Vernor v. Autodesk. The gist is that EULAs are enforceable and that you may really only own a non-transferable license. In the light of keeping your upgrade while selling the old version, that makes sense to me. Of course fairness should dictate that you can sell your license. Then again fairness should also dictate that you don't make copies of the software for all your friends. So, given the unenforceability of fairness, software companies use draconian licensing EULAs and consumers have chosen to ignore them out of hand. This legal decision, however has the possibilities of escalating this conflict and if companies go after used game stores, used DVD stores, etc. I predict that piracy will really run rampant, as consumers will take the only option available to them in fighting rules that violate their sense of fairness.

Physical products engender First Sale Doctrine expectations

I personally have not bought a new Xbox game, relying on my amazon wishlist for those. Of all the games I've played on the Xbox, only GTA4 has felt justified of its full price. The ones I have bought were used and the ones that had no replay value I sold. After all, I had a box with a disc sitting there, so of course I can sell that box.

I have, however, bought plenty of games on Steam. It's a digital sale — I can install it on any computer when i want to play it but I can't ever sell it or even let someone borrow it. Yet I am happy about those purchases. Ok, what is wrong with me? The difference to me, if were to try to put a finger on it, lies in a combination of pricing and purchasing experience.

New PC games are ususally at least $10 cheaper. Whether you claim that this price difference is historical or because console's are recouping hardware costs, it makes a new game easier to digest. Add to that that Steam has mastered the art of the long tail, reducing prices for older games, frequently doing brief yet radical sales and even adding games not originally released on Steam along with patches and support for newer features such as cloud save game storage. Finally, with Steam (even if this is more Valve itself than anyone else), you usually get a longer life out of the game, with updates and free multiplayer.

The purchasing itself further severs the physical ownership bond you have with boxed games. Aside from a less painful price point, it's simple, immediate gratification, being able to buy a game at any time of day or night. You also generally don't run an installer, you just buy and wait until Steam tells you that the game is ready to play. In all respects the experience feels like a service not a product, which reduces the feeling that you own something that you can resell.

As a game dev, what seems like a better way to deal with the fact that only some percentage will buy your game at full price? Only make money at full sale and try to encourage new purchase by devaluing used games with exclusive DLC, etc? Or sell electronically and cut out used games entirely, but attract those not willing to pay full price by offering sales later? After all, 9 months after release $19.99 (like Left 4 Dead right now) still beats not seeing a dime.

While the Vernor v. Autodesk decision may embolden publishers to crack down on used sales, I sure hope more will follow Valve's model. After all, gamers generally don't talk fondly of publishers, but Valve is almost uniformly a hero to the community and that's while preventing gamers from selling their games. Sounds like they've got a good model going.

I'm going to go rather deeply off-topic and venture into tl;dr territory: Every time I drive through LA or am on the long 4-lane interstate corridors of Barstow-Las Vegas or the central valley, my mind spends a lot of time contemplating how highway driving is such an inefficient process. It's a perfect scenario of continuous lanes with defined entry/exit/merge points. You get on at one point and off at some other point. The whole having to drive the vehicle between the two points is not only a misapplication of resources but human nature seems to ensure that it'll always be slower than it has to be.

Why autonomous highway vehicles won't happen (anytime soon)

Before I make the case why and how autonomous highway travel could happen, let's just get the naysaying out of the way. Won't may be strong, but since my objections are based on people, not on technology, I can't forsee this change to happen in any near term. Long after the technological hurdles are crossed, the psychological ones, fear and self-determination, are likely to linger.

Fear of yielding control to machines is as old as machines. We're deeply suspicious of anything that wants to take over a task we believe requires our own skillset. Even if repeatedly proven wrong, we believe that a human being can do things better than any machine. Disregarding the thousands of people that die in car accidents due to their own failings (exceeding their or their cars reaction capabilities, driving impaired, etc.), we are willing to accept those deaths as the cost of driving. But should a single person die because of a computer malfunction, all computer controlled cars should immediately be suspended. We only have to look at the recent, false, accusation that Prius' we're running amok because of a faulty on-board computer and the public outcry as proof.

And even if we trusted cars to be better drivers, we still would not yield control because we want to be the ones that decide what to do next. This is more true in car cultures like the US, but the need for self-determination means that we want to be able to drive where and how we want at all times (ignoring that we already have agreed to a meriad of rules of the road). Maybe we want to cross three lanes to get off at that exit. Or we want to weave through traffic. After all, our superior cognitive skills will gets us there faster than flowing with the computer controlled pack, right?

How could it work?

There's just too much variability and unpredictability involved in driving for computers to take over. Well, not so fast. On surface streets that's certainly true. There are so many unexpected factors that require making decisions based not on hard rules, such as bikes, pedestrians, ambigious signage, bad directions, etc. that will keep daily driving out of the reach of autonmous vehicles reach for a while. But highways are different. 99% of all unexpected decision making on highways is due to humans driving in the first place. If you didn't have to react to the unpredictable cars around you, it's a simple set of rules: There's lanes, there's entrance and exit points, there's lane merges and splits and with communication at lightspeed, reacting to conditions created by another car would be far more reliable than the visual detection and reaction of a driver.

So let's say highways are a controlled environment that can be managed by today's technology, how would something like this come to pass, especially since we can't just set up a new, separate highway system and can't turn it on over night.

Autonomous vehicles

One fear and realistic obstacle in computer controlled cars is the central control of all traffic, that even with redundancy is seen as a single point of failure. Also extending trust in computers to trusting some massive government controlled computer is a special leap that's spawned a hundred dystopian sci-fi stories. For this system to have a chance, each car needs to be in control of itself. People will trust their own cars before they trust an outside entity.

You would pull onto the entrance ramp, determine where you want to get off and the car would take over, merge into the traffic flow and on exit at your destination, the car would hand control back over or stop if it sensed that you weren't acknowledging transfer of control. I'll cover how this is possible next, but the important concept is that it's really just an auto-pilot for your car.

Recognition of the static environment

In order for your car to work on auto-pilot, it needs to have a way to recognize entrances, exits, lanes, etc. This could be done with a combination of GPS markers and RFID. GPS for the layout of major features, such as interchanges, entrances and exits and RFID to determine boundaries, etc. This static environment can be built out and expanded one highway at a time and the combination of GPS and RFID means that there is a general expectation with a local verification of that expectation, i.e. a physical safe-guard to override outdated data.

Recognition of the dynamic environment

Just as important as recognizing the lanes is recognizing cars and other obstacles. By using RFID, radar and/or visual recognition and WIFI networking, cars would be able to detect surrounding cars as well as communicate speed changes and negotiate merges. This communication would also allow the forwarding of conditions far ahead without requiring a central traffic computer. It's basically peer-to-peer traffic control. Since the computers would lack the ego of drivers, merges would not require sudden stops that ripple for miles behind and cars could drive faster and closer while still being safer.

The awareness of all other autonomous vehicles and the propagation of information also allows the detection and avoidance of out-of-system obstacles, such as physical objects, cars with malfunctioning systems or rogue drivers who are controlling their cars manually. Once one of these conditions is detected, it might trigger manual control for everyone, which would just return us to the crappy situation we already have, but it still wouldn't be sudden since traffic ahead would warn our cars long before we'd encounter it.

Oh, right, then there's reality

All the technology to bring this about exists today. Mapping our highways for GPS is already done. Implanting RFID markers is no more complicated than existing highway maintenance. Converting the fleet will take a while, but we could easily start with HOV lanes as autonomous lanes and add more lanes until the entire highway and fleet is converted. Sorry, classic cars, you will be relegated to surface streets or require transport. But considering your polluting nature, that's a good thing.

But let's say the government did decide to undertake this, the implementation reality would be lobbying by large government contractors to create their proprietary systems, attach patents to the tech and create inferior technology (just look at voting machines). They'd create unreliable crap that would erode any trust in autonomous vehicles that people could muster. Maybe the government would require some standard but the development of a standard would be a pissing match between car conglomerates that ends up with something as useless as Cablecard and still lock out any innovative application. Finally, the hunger for data would mean that all this peer-to-peer communication and travel data would be an irresistible analytics goldmine for upselling car, travel, etc. products and services, turning the autonomous system into some kind of giant big brother of movement. Of course, considering present consumer behavior, the big brother scenario would probably not act as an obstacle.

I guess I'm going to continue to be stuck behind the guy in the left lane whose speed is the righteous amount over the limit and who only accelerates when his ego is threatened by me passing him on the right. And i'll continue to have to hit the brakes or react to someone else having to hit their brakes because someone decided that their lane change was of higher priority than the flow of the remaining traffic. All of which is completely uneccessary and counter-productive to everyone on the road and highway travel could be as simple as treating your car as your personal travel compartment in a massive compartment routing system. Well, a geek can dream.