Monday, September 28, 2009

Harnessing spam bots for cyber warfare

Disclaimer

I have not researched internet criminal law and I won't speculate on the legality of this idea. I'm not advocating any type of internet warfare or vandalism. Don't implement this idea unless you determine that it is completely legal.

Summary

With that disclaimer out of the way, I'd like to explain one of my most ambitious and long-planned idea: redirect spam bots to launch a DDoS attack on a website. Even small websites have to guard their forms against bot spam. Why shouldn't this enormous source of resource-wasting power be put to good use?

Background

Spam bots work on the economics of scale. Unscrupulous companies and criminal organizations pay spammers to use their servers to crawl the internet, looking for comment forms and public email addresses. When a spam server (or bot) finds a web form, it fills it out with a mix of garbage and spam links and moves on. If the data is posted on the website in some way (blog comment, forum posting, wiki entry), the bot has succeeded in exposing the link to more people. If even a minute percentage of the people who see the link click on it, the hiring organization can make money by infecting the unwary user's computer with malware and selling his personal information.

However, if the spam is detected by any part of the system, it is blocked and the bot has failed. Unfortunately, it has still consumed the bandwidth and computing power of the victim web server. In addition, the victim server's organization has to use its resources to harden its website against spam bots. On a low-traffic website, the spam bot traffic is negligible, but on larger sites, the cost of spam bots is significant. One only needs to examine the measures taken against bot spam to realize its power: reCaptcha, image rotation tests, and a few more esoteric schemes. Despite all these barriers, spammers can make money with spam bots.

Plan of action

Why should all those spam bot processing cycles be used for nefarious purposes? Right now, when a website detects a spam bot, it has several options: it can block the IP address of the bot, preventing the bot from coming back; it can simply reroute the bot to a dead-end page; or it can attempt to waste the spam bot's processing cycles by rerouting the bot to a bot trap. Typically, bot traps work by enticing the bot to fill out a never-ending line of forms or follow a web of garbage links.

What if this stream of spam bots was instead pointed at link farms or phishing sites? Web forms could reroute the spam bots they detect to a spam bot portal site that would reroute the spam bots to known nefarious sites. If a significant number of websites used the portal as a bot trap, the effect on the nefarious sites could be devastating. The spurious sites would be crushed by the traffic from their own advertising bots.

Eventually, the portal site could be automated to find and destroy targets on its own. It could pick its targets from the list of sites ejected from Google's index for phishing or link farming. Once it had chosen a site, it would redirect its traffic to that site, checking each minute to see if the site was still functioning. Perhaps it could crush multiple sites simultaneously by evaluating their stability and pointing just enough traffic at each to overwhelm it.

Obviously, this idea would be far more difficult to implement than to describe. How would the portal handle the massive amounts of traffic? Who would want to shoulder the cost of this plan? How would the predator portal get the list of sites rejected from Google's index?

Do you think this plan is viable?

Sunday, September 20, 2009

Scribblenauts: Accomplishment and potential

Although Scribblenauts is a terrifically fun and shockingly innovative game, its concept is even more compelling than its gameplay. Scribblenauts is a new game for the Nintendo DS in which you solve puzzles by summoning and using items. If your goal is encased in a block of ice, you could break the ice with a hammer, melt it with a flamethrower, or detonate it with a grenade. The game's hook is that you can summon almost any item imaginable.

Hidden machinery

Of course, the game is phenomenal, but the technology behind it is even more impressive. 5th Cell, the developers of Scribblenauts, managed to cram tens of thousands of interactive, animated characters and items on a tiny DS cartridge. On top of the actual pictures, they programmed in complex interactions between the items. For example, a toaster will turn bread into toast; monsters scare and attack people; cops shoot criminals and chase donuts; and beavers gnaw down trees. This game's backend is completely unprecedented.

Unlimited Possibilities

Imagine if the contents of the Scribblenauts asset database were accessible to all game developers. Designing a traditional game would be a breeze: lay out a level, populate it with items and characters, and show the player an objective. However, giving developers (and players) instantaneous access to a massive library of items would create entirely new gaming possibilities. An RPG could allow players to equip their teams with typical household items with different strengths and weaknesses. A point-and-click adventure could use the sprites and interactions to let the player pick up literally any item in a room and use it in the game. A platformer could use the vast selection of items and characters to make each level a unique experience. A universal resource database could fundamentally change process of creating and playing games.

Making the concept a reality

Thinking about the possibilities if the Scribblenauts database was open is entertaining but unproductive. In order to give developers and users the power of instant item creation, steps must be taken.

  1. A lightweight format for interactively animated sprites must be established. Whether it's a particular arrangement of sprites on a sheet or an XML dialect for defining how a character's parts fit together and interact, there must be a standard for people to follow.
  2. A central database must be created to systematically store and retrieve the standardized sprites. It must be easy for people to contribute, but impossible for someone to damage or corrupt. The sprites and their interactions must be version-controlled.
  3. Developers must have access to the database from their games. Web-based technologies could access the database directly, but pc-based games should be able to use a copy of the database. It could be optimized and compressed, or developers could simply "check out" the portion of the database they intend to use. Perhaps, when a game using the entire database is started, it could check to see if any new items have been added to the database.

Conclusion

The game Scribblenauts is an enormous accomplishment, but it could be so much more. The technology behind Scribblenauts could revolutionize the resource management of game development, tearing down barriers for both designers and developers to create experiences. Nevertheless, in order for that to happen, 5th Cell must release the Scribblenauts' resource database for non-commercial use, or the independent developer community must unite to create their own asset database.

If you'd like to support the developers of Scribblenauts (and me), you can buy Scribblenauts for yourself.

Thursday, August 27, 2009

Adobe, Flash is unsustainable

Flash doesn't work well on Linux. It's a well-known fact in both the Flash-development and Linux communities. When I try to load a Flash-based video in Firefox 3.1 on Ubuntu, the video has about a fifty-percent chance of working and a twenty-five-percent chance of freezing my browser. Any games I attempt to play are likely to never load, and the ones that load are likely to run slowly, look ugly, or freeze.

Not just Linux users are experiencing these problems. As Flash increases its compatibility and pervasiveness, backlash against its instability and slowness grows. Here's an excerpt from Engadget's review of HTC's new Flash-compatible smartphone:

Unfortunately, in our testing, we found the inclusion actually hurts operation of the phone more than it helps. When browsing to a site heavy on Flash (there are many), the browser loading times were abysmal. Furthermore, trying to view videos in-window produced choppy, nearly unwatchable results. You may have a better experience with lighter kinds of content, but in our opinion the main reason to introduce Flash into a mobile environment is to allow for broader media viewing options, and in the current state of this Flash player, you're not really going to get much mileage out of it.

The push against Flash can also be seen in PC market. Savvy users, tired of Flash slowing their browser and displaying obnoxious ads, are saying no to the Flash plug-in. Cutting-edge web developers, taking advantage of the major advancements in browser technology, are abandoning Flash in favor of embedded fonts and slick Javascript animations that are now compatible with modern browsers.

Adobe could repair these problems with more time or more money, but it has neither. If Adobe slows development on new Flash versions to focus on increasing its compatibility and stability, Microsoft's Silverlight will gain ground with new features. Adobe simply doesn't have enough money and resources to ensure good performance across the numerous platforms to which Flash is spreading.

I only see one way for Adobe to ensure acceptable performance on all platforms while expanding Flash's reach: release the Flash Player and compiler source code under an open-source license. Open-sourcing Flash would benefit Adobe in three ways:

  1. The cost to maintain Flash would go down. The developer community would help to identify and patch bugs, freeing up Adobe's Flash developers to work on the next version.
  2. Flash's performance and stability would improve. Many people would report bugs and contribute patches, and a few developers would spend hundreds of hours fixing Flash problems and making improvements. Additionally, users on non-major platforms such as Linux distros and smartphone OSs would help to test Flash more thoroughly.
  3. More people would use Flash and more people would develop Flash apps. Developers know that open-source projects are more long-lasting than proprietary platforms. Even if Adobe goes away, Flash will remain a viable platforms for interactive web apps. More users would trust the stability and security of Flash if they knew that thousands of people were working to make Flash stable and secure.

Adobe, Flash is too expensive to be maintained and developed by a single company. If you truly want Flash to be the universal platforms for highly-interactive web apps, you'll give it to its rightful owners: the users.

I'm not the first one to suggest that Adobe should release the Flash source code. See this open letter to Adobe for another take on the matter.

Friday, August 21, 2009

Help people regain speech

Every year, thousands of people around the world lose their ability to speak. Some are struck dumb by strokes, while others' voices are stolen by degenerative diseases such as ALS or muscular dystrophy. These people can regain their ability to communicate by using speech augmentation software or devices to speak for them. Unfortunately, the obscene prices of both the software and hardware prevent many people from communicating.

I want to solve this problem by developing an open-source speech-augmentation program. My goal is to create a cross-platform speaking program with both a symbol-based and text-based interface. I plan to write the program in Python with an emphasis on flexibility and extensibility. I plan to abstract both the input and output interfaces, allowing the program to be independent of both its text-to-speech engine and its GUI framework.

Consider this post a project announcement and a call to action. If you'd like to help people regain the ability to speak and you have experience with Python, interface design, text-to-speech, or accessibility, send me an email. I need all the help I can get.

Thursday, August 20, 2009

Pygame2 example game finished

Early this summer, when my Google Summer of Code proposal to the Pygame project was rejected, I told the coordinator that I would nonetheless contribute to the project over the summer. I planned on writing examples and tutorials for the Pygame2 multimedia library. Unfortunately, events conspired to prevent me from writing the examples. I had tremendous difficulty installing Pygame2 properly and no time to find the problems: I finally overcame my installation problems only two weeks ago.

Composite screenshot of all four skins Four graphical skins for your gaming pleasure!

However, I spent a great deal of time over the last week writing my first example for Pygame2. It's based on the oldalien.py demo distributed with the original Pygame, but I greatly expanded its scope. I tried to demonstrate as much of the library as I could, but I ended up using only a sliver of the library's functionality. The finished example demonstrates Surface handling, vector drawing with pygame2.sdlext.draw, text rendering with pygame2.sdlttf, image loading with pygame2.sdlimage, and sound effects with pygame2.sdlmixer. It's not an exhaustive demonstration of those concepts, but I think it's a good starting point for exploration of the library.

In addition to showing off the Pygame2 library, I tried to demonstrate best practices. I encapsulated the object instantiation and sprite drawing in external factories, allowing users with incomplete installations of Pygame2 to still see parts of the example. I only have to switch factories to drastically change the look of the game. Most importantly, I expunged the global variables from all the classes. The game still uses a couple long-life variables, but I eliminated the unnecessary ones.

If you'd like to see my uber-example in action, you can get it from its SVN repository or download it as a .tar.gz archive. If you have any trouble with the game, let me know.

Friday, August 14, 2009

Need app developers? Give them respect.

According to Engadget, both Microsoft and Sony are attempting to lure iPhone developers away from the platform to develop apps for the Zune HD and the PSPgo, respectively. Allegedly, Sony is just asking developers to convert, but Microsoft is offering cash to defecting developers.

However, I don't anticipate that either of these strategies will attract many high-quality developers. Switching platforms is a major change that requires a great deal of time and effort. More importantly, the developer has to believe that there's a future for the platform.

Nonetheless, I believe that both companies could foster development environments as fertile as the Apple's by dropping development barriers and treating developers with respect.

Break down development barriers

Many great app developers have never written an app. Some don't have the cash to buy Macs and pay $99 to get their developer certificates from Apple. Some don't consider the investment worth the money. Some don't know how to develop an app. Whatever the reason, there is a large pool of potential developers who have never made it over Apple's speed bumps.

Sony and Microsoft need to make it free and easy to develop an app for their platforms. The SDK should be free, open-source, and cross-platform. It should be completely self-contained, but it should also be compatible with a major IDE. It should contain a thorough tutorial that takes the new developer from project creation to final compilation.

When it comes time to publish an app, the developer should pay less than fifty dollars to make the app available for purchase and absolutely nothing from that point on. Developers should be able to set their own prices and the platform should charge < 30%. If an app is free, the developer should not be charged when it is downloaded.

Respect developers

Both Sony and Microsoft should capitalize on the current uproar about Apple's opaque and draconian app screening process by making their app screening as transparent and inclusive as possible. The application rules should be straightforward, comprehensive, and accessible to developers. Before the developer even downloads the SDK, he should be shown a summary of the rules.

Every developer whose app is rejected should receive a thorough and specific explanation. The app reviewer should say which rule the app broke, which part of the app broke the rule, and what can be done to fix the app. If the app store doesn't have the time to write these explanations, it's obviously rejecting too many apps.

Apple's iPhone development environment has its weaknesses, but to focus on luring away its developers with cash would be short-sighted. Sony and Microsoft must build their developer base with App Store cast-offs and new developers before they can offer the massive install base of the App Store. They need more than a just few money-motivated iPhone developers to launch their platforms.

Tuesday, August 11, 2009

Suggested reading for elected officials

Despite the politically charged name of this post, Welcome to Obscurity will remain a non-partisan blog. It just happens that my most recent idea relates to my elected officials.

How many times have you shared an article with your friends or followers? If you're like me, you've probably forwarded plenty of interesting stuff to your social network. On the other hand, how many times have you contacted one of your congressman? I've sent less than ten messages to my elected officials, and I consider myself relatively politically active.

Despite the massive improvement in communications technology, few people bother to communicate with their elected officials. And why should they? One person's suggestion is unlikely even to be considered. It takes a coordinated effort to make an impact on a powerful politician, and few people have the time or resources to gather support.

To solve this problem, I propose a website for people to suggest articles for their elected officials to read. A worried elector could use a website, bookmarklet, or browser extension to mark articles they believe should be read by their elected officials. He could send the article to a specific politician or to every politician responsible to the elector.

The first time an article is suggested to a politician, a page is created for him that contains a link to the article. As more and more articles are suggested, the system sorts the articles by their popularity and specificity. An article's popularity will be based on the number of people who have suggested it. An article's specificity, on the other hand, will be determined by the number of other people to whom the article has been suggested. If a voter suggests an article to every one of his politicians, that article will have a lower specificity than one that he forwarded to a specific politician.

A politician will be able to confirm his identity with the service and respond directly to the suggested articles or subscribe anonymously to the RSS feed of his reading material. The politician could set his suggestions to expire quickly if he has a large volume of them or stay on his page until he responds. Normal people could also subscribe to the RSS feeds of politicians to learn which political articles are popular.

What do you think? Is this idea a potential world-changer or an insignificant pebble thrown against a Mack truck?