Slashdot Mirror
Google Programming Contest
AccordionGuy writes: "Google has just announced its first annual programming contest! The objective is to write a program that will do something "interesting" with the about 900,000 Web pages' worth data that's Google provides. In addition to writing the program, contestants also have to convince the judges why their program is interesting (or useful) and why it will scale (that is, handle a constantly increasing load of data that grows as the Web grows). The prize is US$10,000 in cash, a V.I.P. tour of the Google facility in Mountain View, California and possibly a chance to run their program on Google's complete billion-Web-page store."
I think I'll write a program that will delete pages as it finds them. This should scale pretty nicely and make the web faster in the process.
How about adding the option to have google understand what I *mean* to search for, not what I tell it to search for.
Oh, and the ability to find one non-fake Britney porn pic.
Much like the recent discovery of the average color of the universe, this would be a pointless, but fun, use of the data. Of course, I'm not sure exactly what to average. Do you take into account browser real-estate a particular color occupies? Do you simply average each color= and stylesheet instance?
Ideas?
All sweeping generalizations suck.
10K is nice along with the recognition and all, but... I'm sure that's a lot cheaper than paying a few Google staff coders to come up with the same thing in a few months.
Jus' being paranoid.
10000$/x hours of work we could get done for us...
Make sure we get a slashdot posting so a bunch of geeks with programming skills will enter.
The only thing I'd want is for google to stay just the way it is though, don't bloat. Great service, maybe I'm just pessimistic but sites rarely do everything well.
Kjella
Live today, because you never know what tomorrow brings
Of course this could be spammed, but as I said, a human could filter the results every day; besides, it would be hard to create a very large number of unique links from different servers pointing to a page. I'm sure Google is already doing some of this to prevent spamming their search-order algorithm anyway.
An automated Googlewhacking system.
Ingenius!
-Waldo Jaquith
They're going to (hopefully) get tons of interesting ideas and almost as much useful code for the price of $10,000. Sure beats hiring programmers.
That's assuming that any contest entries automatically become the property of Google.
Perhaps this is the evolution of a new buisness model... Either way, I don't really care as long as Google remains free, fast, and useful!
It's not evil, it's just business. Other companies have been doing it for years. Back in the day, car companies used to sponsor "car design" contest for little kids. The winner would get $50 and his car would be whisked away to the labs. Why pay a team of designers and engineers to do what a trained^H^H^H^H^H^H^H normal person would do for cheap? Maybe we'll get a spiffy new feature on google! Hurrah!
I'd go for a dictionary of every word ever used on the web. Complete with common usage examples.
I am the NUL and the DEL, the beginning and the end.
how about have google parse every page, and save the homepage as an image. then take the map of the internet, and make it using tiny thumbnails of the most heavily linked (popular) sites.
this would be just like those mosaic photos, only much nerdier. thinkgeek execs are drooling already....
MARIJUANA, SHROOMS, X: ONLINE?! - E
A few years back there was a game, I think it was called Virus or something like that. It would scan your directory structure and make a map for the FPS world based on that.
Looking at the web, I allways though it would be cool to make a game based on the same concept, but use web pages instead of your hard drive directory.
I'm just throwing out ideas.
Google Contest Winner Offers Better Porn Searches
Winner of the First annual Google Programming Contest creates greatest porn spider ever.
MOUNTAIN VIEW, Calif. - December 11, 2001 - Google Inc., developer of the award-winning Google search engine, today announced it's first winner of the Annual Google Programming Contest. Winner I. C. Porno has created a program to help catalog and organize google cache of the Internet, also refered to as the World Wide Web of Porn.
"This announcement is an important step in Google's ongoing effort to provide search services that are fast, easy to use, and that help people find the information they need," said Larry Page, Google's co-founder and president of Products. "To search our collection of 3 billion documents for porn by hand, it would take 5,707 years, searching twenty-four hours per day, at one minute per document. With I. C.'s new program, it takes less than a second."
World's Largest Collection of Porn
Google users now have the world's largest and most comprehensive collection of porn right at their fingertips and can immediately primal urges using the following services:
Google Web Porn Search: The company's newest search service now offers more than 2 billion documents - 25 percent of which are non-English language web pages. Google Web Search also offers users the ability to search for numerous non-HTML files such as PDF, Microsoft Office, and Corel documents. Google's powerful and scalable technology searches this comprehensive set of information and delivers a list of relevant porno in less than half-a-second.
Google Porn Groups: This 20-year archive of Usenet porn conversations is the largest of its kind and can serve as a powerful reference tool, while offering more porno than the Internet. Google Groups was released from beta today with 700 million postings in more than 35,000 topical porno categories.
Google Image Search: Comprising more than 330 million nude images, Google Image Search enables users to quickly and easily find porn images relevant to a wide variety of topics, including pictures of celebrities and popular travel destinations. Advanced features include search by image size, format (JPEG and/or GIF), coloration, and the ability to restrict searches to specific genre's of porn.
About Google Inc.
With the largest index of websites available on the World Wide Web and the industry's most advanced search technology, Google Inc. delivers the fastest and easiest way to find relevant information on the Internet. Google's technological innovations have earned the company numerous industry awards and citations, including two Webby Awards; two WIRED magazine Readers Raves Awards; Best Internet Innovation and Technical Excellence Award from PC Magazine; Best Search Engine on the Internet from Yahoo! Internet Life; Top Ten Best Cybertech from TIME magazine; and Editor's Pick from CNET. A growing number of companies worldwide, including Yahoo! and its international properties, Sony Corporation and its global affiliates, AOL/Netscape, and Cisco Systems, rely on Google to power search on their websites. A privately held company based in Mountain View, Calif., Google's investors include Kleiner Perkins Caufield & Byers and Sequoia Capital. More information about Google can be found on the Google site at http://www.google.com.
Th
Notice that they don't say exclusive license. You should be able to release it as GPL yourself.
i actually bugged the google guys a while ago about adding a spellchecking function to google. throw a URL or a set of pages at it, and it spits out a list of misspelled or questionable words - highlighted in the way they already do search terms in the cache...
anyway, someone there emailed me back basically saying it was an interesting idea, but not something on their agenda.
maybe someone out there can work up a scalable google spellchecker that i can run my big-ass database-driven website through (which is a major pain to spellcheck, considering the client simply refuses to do when they provide the content)
- Entertaining Bits from the Ancient Kernel Tree
What's evil about it? Smart maybe, but evil?
Anybody who would enter such a contest is primarily motivated by the challenge, I would think. Getting the $10K gives you bragging rights is all.
Sure, Google gets some value, but a lot of highly motivated programmers get a challenging problem.
If all good programmers were primarily motivated by money, there'd be no Linux, BSD, Apache, Emacs, Vim...
I reserve evil for things that actually hurt someone. This seems like a win-win to me.
Thing is, though that is a lot of money, what happens if you make them, say 20,000 USD with a great new compression/analysis algorithm.
If you're that good, they'll probably hire you to at least consult for them to maintain the code you wrote.
The idea is roughly to refuse to index sites which engage in keyword/description abuse.
- index keywords and description data
- Allow users to search with keywords on or off
- If users search with keywords on, provide a mechanism for users to nominate a site as engaging in
keyword abuse.
- semi-automatically, and then manusually review nominations.
- Refuse to index sites which have engaged in keyword abuse.
This isn't so much a system that meets the specs of the contest. And there is a scaling issue, but it is on my wish-list for google (and others) to do.Prime numbers are exactly what Alan Greenspan says they are -S. Minsky
How about a program that searches for the meta generator tags and looks for "Microsoft Frontpage X.X", deletes the page from the database, and commenses a DOS attack from the rest of the slashdot community?
Go Google! Get rid of the fake HTML goons!
Better yet, post a story to slashdot about a contest with a prize of 10k, read all the responses moded at 4 and above, spend a weekending coding a few of em up, and cash in!
Now that's evil!
Make a image-2-asciiart converter, so you could have a txt-only option on the google cache.
The Googlewhacking site lists reader-submitted Googlewhacks...which of course causes Google to pick up a second site for the search. And so the Googlewhack is whacked!
"Rub her feet." -- L.L.
Write an application to track keyword usage over time, when a keyword goes from only 10 hits to several thousand then flag it for jargon. The jargon can then be presented as a webpage of the top whatever with various statistics over popularity and suspected origin urls.
- MbM
The key word here is potential. I think that you would almost waste more money in evaluating a lot of the trash that comes in. The most valuable thing they probably will get from it are the ideas that people come up with. Notice how they made it as open ended as they could.
My Weblog
If someone can come up with a regular expression search engine that scales to billions of pages, that would be the killer app for Google. It would probably have to be a Deterministic Finite Automaton (DFA) regex engine, not the more powerful Nondeterministic Finite Automaton (NFA) engines like you have in Perl, Python, Emacs, and Tcl, but still, that would rock!
Fight Spammers!
Connect any two pages on the web to each other with the minimum number of hyperlinks.
I mean, how many contests have you seen on the back of a cereal box to "create a new slogan!" or "write an essay"? Just a cheap way to create some buzz and get your customers to write your advertising copy for you. Heck, the most blatant scams in memory are HBO's Project Greenlight (trolling for scripts - you don't even want to know what the Writers' Guild thought of this) and the Lego Film Contest (trolling for complete commercials).
Hardly new stuff. Remember Mark Twain's Tom Sawyer? There's a bit where he holds a "contest" to see which kid can whitewash the fence he's supposed to paint fastest. I'm sure that even as Twain wrote that bit, even he thought "I better be sure to give the fence painting thing a unique spin so it works. After all, it's an awfully old idea..."
"Prepare for the worst - hope for the best."
I personally think it'd be coolest to turn it into an art project.. imagine you had a repository of the consciousness of an entire race and could run a script on it. Things like the map of the internet. Or the web collage. Or use it to power some kind of AI chatterbot.
I dunno. Their webpage on it didn't seem to do much to promote being creative; they just want to pay someone 10k to develop a new way to make more relevent search results.
When did you last donate to Google? How many times have you used Google on your job, saving your self and your company money? Where is the friggin' "Do it for the love of coding" thinking now? I would be happy to enter (I just need the right idea ;)) and if Google gets better because of my code, so be it!
J.
It seems like it would be very easy to come up with something interesting, and only a small fraction of those interesting things are actually useful.
Examples of a few interesting non-useful things I can come up with just off the top of my head:
Google Poet: Generate rhyming poetry from randomly rhyming sentances on the webpages in the database.
Googlesaic: Input a picture and scavenge the webpages for pictures from which to create a large mosaic of the input picture.
Google Map: Create a picture/graph of all the website connections (links) in the webpage list, perhaps add 3d/naviations. Perhaps perform graph opererations and maybe find the longest path one can travel through the links and still stay within the Google search results/database.
These are just a few, I'm sure plenty of other people can find much more exciting/interesting things to do, but they won't always be useful to the google company.
Things you think are in the Constitution, but are not.
It's a party game. The basic idea is that a bunch of people are in the game, and it goes around in turns. On your turn, you type in a few words to search for. The game goes and queries google for the first hit on that search, and sends everyone's browser to that page. Then the other players get 100 seconds to guess which words you searched for. The first player to guess correctly gets points for the amount of time remaining.
It's written using BYOND, which you'll have to download if you want to play.
Say hello to zMac.
If you read the rules, you will see that you don't even have to assign copyrights to Google. You only have to give them a license. This means you can GPL your code or even BSD it. Sounds fair to me.
Make even shorter URLs - 8LN.org
Webcollage -- slowly builds a random collage of images from the net.
DadaDodo -- generates random sentences based on word probabilities in pages on the net.
-- The Hoss Man
Shayne
Today I didn't even have to use my AK; I got to say it was a good day -- Icecube
s/www\.microsoft\.com/www\.goatse\.cx/g
Six degrees of Google Bacon. How many links (and what's the path) to get from any page on the web to Kevin Bacon's personal homepage. Or more interesting from any page to any other page.
You just described open source exactly. Except the part about paying ANYTHING at all. Pretty slick!
Contrary to popular belief, coding is not all free blow-jobs and beer. Those things cost MONEY!
I wouldn't go for $10k. Perhaps $100k, or perhaps $20k plus some percentage of future revenue attributable to my invention.
Got to hand it to them, though, it's an innovative way to receive hundreds of ideas and get a working prototype. Only one person wins but they probably retain the rights to develop their own code that accomplishes the ideas submitted by everyone else.
Basically, they want a cool idea for something innovative but their brainstorming sessions haven't come up with anything new...
DFA and NFA are equivalently powerful. (It is a relatively simple proof to show transformations between them.)
It's true that Emacs et al. support a richer language than what's offered by traditional regular expressions (as can be implemented on DFA or NFA) but that's because the languages are *not regular*. It has nothing to do with the distinction between DFA and NFA.
$10,000. 8 weeks til deadline. 40 hours per week.
That's 10000/(8*40) = $31.25 per hour.
Annualized that would be a salary of $65,000.
Even in IT, that's nothing to sneeze at. But I'd say the benefits of winning a contest like this go beyond the money.
I do not have a signature
...something that looks through that data and finds the interesting bits based on a set of terms that the user provides?
Or has someone done that already?
I'm not sure if using USENET is such a great idea. While there are some areas where it has a great signal to noise ratio and intelligent commentary, there are a ton of places where it's simply awful. It's loaded with misinformation, flameage, and proof of the correctness of Godwin's Law. I doubt that I'd be very excited about chatting with a bot that learned to communicate by reading the USENET archives.
OTOH, you might be able to do some very clever work on using the page cache as a knowledge store for a chatbot. You'd just take the incoming message, try to find some keywords in it (probably using previous parts of the conversation to help) and use them to search Google for relevant information. Then you'd reformat the information you found into something like a conversational reply and send it.
There's no point in questioning authority if you aren't going to listen to the answers.
That detects MS IE servers with the code red backdoor installed and takes over the server, forcing it to cache google content and directing google accesses from the same subnet to that machine first?
I'm trying to teach myself to set people on fire with my mind... Is it hot in here?
Sure beats hiring programmers.
No, that's it!
According to this article Google is getting deluged by resumes, this is just a way for them to weed out the 600+ resumes they get a day.
The winner of this contest (and maybe a few of the runner ups) will most likely get a job offer as well. Beats having to weed through 4200 greatly exagerated CVs every week...
-Russ
Me
A friend of mine accidentally typed:
fat misgets fucking
into google....
Google knew exactly what he meant....
The secret of success is honesty and fair dealing. If you can fake those, you've got it made. (Marx)
Make a "find person" function. Write a name and Google figurs out what the facts are: e-mail, work, icq and interests. The problem today is that a lot of people are called the same, but with the corelation with email and other data. The program would be able to separate two persons with the same name. A great Big Brother function.
Unfortunately, all the comments at 4 and above are complaining about how Google intends to rip people's ideas off.
Personally, I'd like to see hits to pages marked, and the top 100 hits from each search are fed back in to be re-indexed. This would eliminated a lot of dead site material, I should think.
--John
This is also a good way to get a job at Google. They pay a lot of money.
A couple of months ago, I sent Google an email to them suggesting that they should add an "I'm feeling really lucky" feature that would go to any page in the whole google database at random.
:(
Maybe something like pressing I'm feeling lucky with no search string?
Haven't seen it yet
Believing something doesn't make it true. Not believing something doesn't make it false.
Hey, aren't Google breaching the copyright of at least some of those whose pages are included in the sample data being used -- especially the CDROM's worth that will be sent out?
As for the cost-savings involved in running such a contest, I expect the fact that they only have to pay $10,000 will be more than offset by the fact that they'll have to sort through a mountain of crappy submissions. That'll take a lot of people a lot of time.
So that pages that can properly be read by any browser comes first.
Then, maybe webmasters will stop doing IE-only pages.
{{.sig}}
I've got one:
Lets take all 900,000 pages, and look at the statistical distribution of the frequency of appearance of each letter of the alphabet. That way we could check to 10 decimal places that the letter values in scrabble are REALLY correct...
I hope Google reads these pages and gets some free ideas from it. At least take mine! Please. God knows that I don't have the coding chops to do it myself. I sent this same idea to Allaire (remember them) a long time ago and I had a couple of software engineers write me back, but nothing ever came of it. My guess is that this is a hard problem.
I want a browser control/plugin/whatever that harnesses a backend of web information to make my surfing more productive/predictive.
The gist would be to have a hover option for links which would give you information about what is behind the link without having to actually follow it. While browsing, the user would just hover over an link in a page and information pertaining to the page beyond the link would show up in a hovering menu or a sidebar (this would be great with mozilla, but I could see an activex control as well).
The types of information is where it gets useful. Using some of the more advanced summarization algorithms out there, it would pull up the summaries of those pages if they were in the offsite database (Allaire, Google, and the WayBack Machine being possible backends). Based on your preferences a short, medium or long summary would be displayed. If it wasn't in the cache, it could be summarized on the fly and then presented after some delay (the new summary now being cached).
It would also list, in an orderly way and subject to preferences, links from the page on the other side. That way the user could follow one of those if it turns out that she only needed the summary and a link. It would also list the elements of the page, like graphics, and give their specs (i.e. dimensions and estimated download times and ALT tag entries if present) and give the option to display them on a page by page basis. All of this would be nested, of course, so that a user could hover over links in the summary pages and get the same information all over again for that link (which is why I see it more as a "sidebar" feature). Theoretically a user could just surf by these summaries if they wanted.
Now, I realize that this would pose some problems like trusting the summaries and so forth. However, the nice thing about it would be features that could be built into the user's preferences. For instance, you could make it so that the user could have certain words or phrases set that would then be scanned for during the summarization process. You could then either relax the amount of summary for the entire page or, better yet, still pull the cached summary but also pull a user-definable number of lines before and after their keywords (best of both worlds).
Each summary could also list a numeric rank of where that page fits in "status" (like google's ranking system) based on the summary (generically) or the keywords of the user (specifically). Finally, it could pay for itself with text advertising (small and innocuous like the ones seen on Google).
If you start to think about it for a while, there are all sorts of things you could do with this and it would help cut through the "padding" that you usually go through while looking for informaition on a certain subject. I think it would be great! It is kind of based on the idea of the "magic spyglass" that was heralded almost a decade ago, but never implemented in any OS that I know of.
Like I said, I can't code it, but I would love to see it done. So have at it if you think it is good. Google's cache of pages and images and its ranking technology make it perfectly suited for this type of problem and they have enough PHD's that the summarization issue should prove an "interesting" problem to solve.
Then again, it might suck. If you do implement it, let me know. I would love to beta-test it. I called the whole thing the Clairvoyant Browser Plugin... but you could use what you want.