AltSci Cell

Dec 24, 2012

AI3 is officially in beta. Hooray! Start testing when you get a free minute. Send any bug reports to me. I am so excited to work on them and respond to you.

What is AI3? The way that I usually describe it is like an inverse thesaurus. Instead of finding a synonym or antonym for a word, you find the most likely word to put afterward and the context of that word. You can also find sentences that use two words. If you're stuck on a word, check out a few hundred uses of that word. Unlike Google, most of the results are not sorted by popularity. They're sorted by position in the database.

Dec 13, 2012

Some of you may be excited to hear that I am finally on the brink of releasing a very interesting website. How interesting? It has kept my interest for weeks even though some of it is hard and some of it is annoying.

gzip -l ~/ai3_db_backup_20121211_ram2.gz
         compressed        uncompressed  ratio uncompressed_name
         1137167097           403791553 -181.6% /home/jvoss/ai3_db_backup_20121211_ram2

zcat ~/ai3_db_backup_20121211_ram2.gz | grep -e ^INSERT -e ^LOCK -e ^UNLOCK |wc
   4716  642299 4698740275

It's alright if you don't understand what the above means. The first one shows a bug I found in gzip where when you try to find out how large something is when it's bigger than 4GB uncompressed, it wraps and gives you a small number and a negative compression ratio. It's not a vulnerability so don't worry about it. The second one shows that there is about 4.7 GB of interesting data in that file.

Nov 2, 2010

You may be aware of my work on RockBand 2 Drum "drivers". If I had written my code as a kernel module (and it would likely be accepted into the mainline), it would be easy to call it a driver. However, I wrote it using libusb and ALSA, which makes it a userland program. I feel that I made the right design choice. ALSA libraries are low enough latency and libusb allows efficient use of interrupts, so it works quite well. Also, adding any complex code to the kernel that isn't totally necessary seems like a risky endeavor.

Anyway, that code is here: RockBand 2 Drum Instrument for Linux

June 10, 2009
Update July 23, 2009

Digg Diversity is a new project by AltSci Concepts. It uses the Digg API to calculate a more fair score for articles on Digg. Why is this algorithm necessary or preferable? Digg has an algorithm that is based entirely on profit, which is acceptable for a company like Digg. The more diggs that occur, the more profit that Digg makes, which means that they will accept, even encourage their users to game the system. The Digg front page algorithm which promotes articles to the front page with as few as 100 diggs means that a small number of people can control the front page of Digg by simply getting 100 like- minded people to digg their articles (and visa-versa). The company Digg benefits when corrupt users promote the same content repeatedly, but the overall community is diminished (especially those users who wish to see important non-repetitive content). This topic is extremely deep and deserves an essay but definitely not tonight on the night of the beta release of Digg Diversity. Many digg comments, blogs, and even a mashup that is currently offline have been written about this issue, but I hope to write the solution.

Digg Diversity is a entirely javascript mashup using the Digg API to retrieve important information about what data is found on Digg. The first set of results may be rather surprising. You will see a list of results quite similar to the front page of Digg. However, the order is by "divvs" which are a new calculated value based on timing and repetitiveness of the digger. The raw data can be found at the bottom of the page (there is a link that displays the data).