Missions to Mars

(Click to see full size)

Internet Speeds and Costs Worldwide

(Click to see full size)

Automatic Photo Montages

This is quite brilliant. It never would have occurred to me that such a thing as this should be built, let alone that it could be built. Nonetheless, once you see how it works and what it produces it’s quite impressive. By simply drawing shapes and associating simple keywords with those shapes the tool searches for images on Flickr, extracts parts of those images, and composites them into a final photo.

Sketch2Photo: Internet Image Montage from Tao Chen on Vimeo.

Honda’s One Wheel Cycle

Well it’s Crazy Japanese Invention Friday again:

Denial of Service Attack: People

Really excited to have spent 47 minutes and infinity seconds today forced to sit through an “online chat” with a sales person from Comcast. Instead of what should have taken 5 minutes to complete by filling out, oh I don’t know, a form, Comcast (seriously) forces you to load a Java chat applet, wait for someone to be elated to talk to you and take you step by step through the process of providing (and confirming twice) your home address, phone number, SSN, and crap tons of other bullshit.

Holy christ it was a massive waste of my time! But especially theirs — they must be replicating this across hundreds of reps. So it got me to thinking, I could DOS these guys by combining a few hundred Internet connections, some open source AI software (like ALICE), and a bit of programming.

From a link on the front page of Comcast.com you can immediately initialize a chat session with a human agent. Simply wire the output of this up to the human chat simulator and repeat. I decided to fire an instance of ALICE and see what would happen if acted as the proxy between her and, as it turns out, Comcast representative “Arnil:”

Comcast> Hello Guest_. Thank you for contacting Comcast Live Chat Support. My name is Arnil, please give me a moment to review your information.

Comcast> How are you today?

Guest_> Ah. Fair to partly cloudy.

Comcast> I understand that you are interested in Performance (up to 15Mpbs downloads with PowerBoost) $19.99 per month, is this correct?

Guest_> I never knew I had this correct

Comcast> Let me check if your area is eligible for Internet. Would that be OK?

Guest_> It’s OK to check if my area is eligible.

Comcast>Can you confirm if all these information are correct: 1211 E JOhn ST, Seattle WA 98102

Guest_> Let me think.

Comcast> I need your exact address

Guest_> Are you getting paid to ask me that? That information is confidential.

Comcast> I need your exact address to process your order.

Guest_> Well, what is your credit card number?

Comcast> Are you interesting in proceeding with this order?

Guest_> What is your credit card number?

Comcast> If you are not ready to proceed you can always chat us back.

Guest_> What would chatting you back involve?

Comcast> You can visit www.comcast.com and click the Chat Live with Customer Service button in the upper right.

Guest_> Have you ever clicked the Chat Live with Customer Service button in the upper right?

etc. etc

This actually went on for a while and probably could have kept going for a long time. With modest resources you could tie up their entire Army of sales people*.

Silence Anyone

Sometimes when people talk, it’s incredibly annoying. I’ve often felt as HAL apparently did in 2001: A Space Odyssey when he said to Dave: “Nothing further can be gained from this conversation, goodbye.” How I’d like sometimes to say this and back it up with some audio muscle. So, how about a anti-noise device, modulated to a specific human voice such that the moment the target spoke, the exact inverse waveform was produced and broadcast at just the right angle to completely and immediately cancel their voice?

I imagine this would be really difficult to pull off for several reasons. First, you’d need probably to get the angles just right and account for the speaker being able to move this way and that. If you broadcast the inverted waveform and they tipped their head up, the waves would be at cross angles and wouldn’t quite be synchronized.

Second, you’d need somehow to capture the voice output close to it’s source. If you waited until it reached the receiver you’re too late as it will have spread out in all directions and most of the message will have gotten through to anyone else listening. One idea might be that you could somehow model the persons voice and use computer vision to predict how the person would sound by lip reading. You could then calculate the correct wave form to cancel the sound at roughly the speed of light.

Also, I’m left wondering if you could produce a Cone of Silence using the same setup except allowing for a thin “tunnel” of sound between the source and the target. Anyone outside could never hear what was being said since no sound would escape.

Rad iPhone Navigation Concepts

Found via Mobile Cruch, Swedish design company Ocean Observations has developed two very rad ideas on how to improve the often difficult iPhone application navigation experience.

Google Wave

Well, my invitation to Google Wave finally came through. Without being able to include more people that I know it’s hard to see if it will be useful or not, but here is what it looks like:

New York Subway Augmented Reality

I expect to see a lot more of this type thing in the next 6 – 12 months.

Here is another very practical example using ARToolkit I believe. You can check to see if your physical object will fit in any of the USPS shipping boxes without actually having a box..

TED talks worth watching

I watch TED talks daily. It’s the best thing that ever happened to my AppleTV. Here are some of my recent favorites: