TechCrunch50, the industry’s American Idol-like startup spring board, wasn’t immune this year to the usual roundup of mediocre start-up ideas. Most of them pimped by overly optimistic business types who have convinced themselves their ambitious yet somewhat dubiously profitable start-up dreams are going to make it.
Of the miss-worthy companies, a website to allow children to be characters in their own stores (“Story Something” is the name of the company); a GuitarHero-inspired game to let you be any member of a band (“ToonsTunes.com”); a service that provides a “seal” to allow someone to have a image that represents them appear across various places of the web they might be represented like blogspot, tumblr, wordpress, facebook, myspace. (“Sealtale”); a hardware device that acts like an “invisible cable” – two usb keys that can be separated and plugged into any two computers (don’t even have to be on the same network), providing instant access to the remote computer’s drives (“iTwin”).
An interesting technical project which caught my attention is FluidHTML, a markup language derivative of HTML that allows you to write in HTML-like markup to create Flash applications. This solves three problems inherent with Flash, an Adobe technology which has received various adoption in the last few years. Once the foremost technology to build beautiful “rich internet” sites (the kind where things fly around, make noises, play movies, and the like), Flash is now experiencing a backlash as big players (most notably Apple with their refusal to include a Flash player in the iPhone) try to thwart the Adobe-dependance that would be created if Flash was the de-facto and only way to create animated sites (Many Web 2.0 site using animation libraries and AJAX come closer to providing rich internet experiences, are seen as competition for the Flash platform altogether).
FluidHTML, however, is not a Flash alternative, it is a markup language to make the creation of Flash embeds easier (and therefore cheaper) to code. It solves three inherent problems: 1) search engines can’t index Flash very well, 2) search engines can’t analyze Flash for deep links (the fundamental idea that pages link to other pages), and 3) browser history doesn’t work work because any “back” and “forth” within the Flash player isn’t part of the browser’s back & forth feature provided by the browser. Fluid HTML attempts to solve all of these problems by turning Flash into HTML-like content which can be search & indexed but is rendered to the user as a Flash movie. “We are making flash work the way the rest of the web works.”
One of the more popular companies was ToyBots and their forthcoming toy call Woozie. The device is basically a small electronic controller that goes inside of a stuffed animal, making the toy an interactive project between the child playing with it and other parents. In fact, ToyBots business plan to to build the platform that will facilitate all kinds of toys being made in this realm. Outlined in this slide, the device features thing a lot like an iPhone- GPS, accelerometer so it can react when the toy is shacked, etc.
The “Woozee” (toy) can wake up the child at a specified time, then tells the kid that she has a message waiting. “Using the accelerometer, I give Woozee a hop and he plays the message.” A message from ‘dad’ pays, saying that he has a great bedtime story to read the kid. The ‘dad’ is away on a business trip but using his iPhone has selected an audio book and remotely instructed the toy to play the audio book exactly at 8:30, bedtime. (Part of me thinks this is cool, and part of me thinks it’s a little creepy, but I don’t have kids.)
The judges response to the Woozie seemed to center around the porn industry and various adult uses of the toy, along the vein of whether or not this technology would be licensed to other people, if intellectual property is locked down and/or open to the public, etc.
“I’m amazed that a nation which can create this kind of sophisticated toy and send a man to the moon, cannot develop a toilet seat which doesn’t fall when you are using it.” – Yossi Vardi
In an impromptu calling up of a representative from Best Buy onto the stage, the judges wanted Best Buy’s reaction. For the most part, the representative was positive about exploring it more, but wouldn’t commit to numbers of how many he though Best Buy could move.
On the internet advertising front, two companies trying to create software to better tailor advertising (banner ads) – something which remains a problem for a lot of high traffic sites that run banner ads. Both attempt to eliminate the problem of inappropriate ads being run on sites that don’t want that kind of advertising. Pornography, or ads with an agenda clearly oppose the site’s because a computer matched keywords off the page but failed to realize that the ad’s agenda is opposite that of the sites, therefore degrading the brand of the site running the ad or the publisher showing the ad. (Like, a pro-Israel ad running on a news article that is politically pro-Palentine, for example)
First of the two was 5to1, a company which is competing in a space which is typically regarded as being dominated by Google: the reliability of targeted marketing advertising. Their claim is that many banner ads which run in the “remnant market” (when Google or their ad networks don’t sell ads at full price, they are sometimes sold to affiliates a lower prices).
Another company, DataXU, a start-up with a lot gusto (they claimed to have “several PhDs” as founders). “We create algorithms that figure out which are the best consumers, sites, and ad creatives are the most likely to drive sales.”
DataXU continually alluded to during the presentation (basically claiming that they had developed an algorithm better than any other for ad matching that was based on, yes, rocket science). When pressed by the judge for the “secret sauce” that he kept mentioning, the presenter offered no more than to say they were applying mathematical models of machine learning to get better ad delivery performance.
Finally, an interesting startup with the attractive charismatic Jack Ratzinger giving the presentation (always helps) describing SeetGeek.com, a concert & sporting events ticket brokerage site that tracks the prices of sales on the secondary-market (tickets bought & sold for an event on ebay, by scalpers, etc). The site attempts to track the fluctuation in price, then predict it as the event gets closer, allowing you to buy the ticket at lowest possible point.
According Ratzinger, the system crawls the internet and pulls in thousands of ticket sales. “We also pull in other external factors that we know that drives ticket prices: who’s playing, are they in a playoff, what’s the pitching match, is the ballpark holding a promotion, what’s the weather like, is there a Bruce Springstein concert right next door.”
As part of his demonstration, he showed the audience a slide with a line graph showing the ticket prices as the event got closer and asked by a show of hands how many people would buy a ticket today or wait (the room was evenly split). The second slide shows in red what actually happened – ticket prices when down.
Ratzinger claimed the software was currently testing at about 75-80% accuracy and improving every day.