In CS 106B, there’s an assignment called “NGrams.” The program you build opens a text file and then generates a certain number of words based on the probability of each word appearing after the previous set of words. The result is a stream of text that is generated “randomly.” After completing the program, I tested it using a large input file filled with Facebook status updates from a friend’s profile. It was fantastic. So I decided to compile a file filled with Stanford Daily opinion columns. The following opinions column is printed randomly (including the title), with a few punctuation errors corrected.
***
Health issues are liable to promote this stigma too. I don’t know if I learned something with my family or friends back home, our hikes are relatively low-stress. A medium-sized hike, sticking to the next to fight the fight. To do was to take the calculated risk. Worst case, I slowed the group down or turned around. I don’t know if I could’ve just stood there forever?
In that way they are rejected because we understand the faults of those that came before us, and use our knowledge to make people feel guilty or worthless. The free flow and debate of ideas are both essential to a democracy. If we cannot really attack fascism because we understand the faults of the FBI; swept aside and evacuated decades of science; destroyed individuals and families through inhumane immigration policies; and so deregulated the financial world in some way.
To me, the mark of a swooping, miraculous flight is at the center. Never mind a real dialogue, we could have advanced our learning by hearing each other out and engaging others as well. That is what I didn’t realize about paralysis. I was supposed to complete a much longer hike, a few times. I can honestly say I enjoyed my risky hiking experience, not because of the organizers of the white supremacy, hetero-normalcy, misogyny, anti-Semitism, Islamophobia and other forms of bigotry events. Consider this line of argument: One can be vital to ensuring that patients aren’t taking medications that conflict with one another.
In the distance like some sort of architectural Messiah, I wanted to weep. Actually, I think about Hillary Clinton. There are probably dozens of reasons why she lost the election. But she is a good predictor of what I thought myself pretty good at, having done it for all my life thus far: learning. I’m pretty familiar with learning. Every year since I waddled into preschool, I’ve been required by law (and my mom) to learn. From my ABC’s to Crime and Punishment, school has been blurred, and opinions have come to terms with that.
Progress is about, then, learning how to cook and how to wear long skirts. I am one of the people with visible overreach that could impact the self-esteem of people as objects and within set categories that render them inhuman and perpetually less worthy. They reject the basic values upon which academic freedom is in sync or entirely separate. The answer is a little scary.
When I got in my friend’s car, they decided to close with words of supposed hope. We spent the quarter looking at the top and a sense of accomplishment rather than cling to the specific speakers as they protest. Fewer still seem interested in doing so. The Fundamental Standard contains elements that are relevant to that discussion. But whether private or public, universities, like democratic states, are premised on an underlying lack of solutions, leaving me looking for answers.
***
This experiment turned out better than I expected. It also revealed a few interesting realities. First, it shows the current limit of computers. Although the sentences are mostly grammatically correct, it doesn’t quite make sense. This makes things interesting to read but useless in application. However, this is also only CS106B, and natural language processing capabilities are far beyond what I’m able to create at the moment. Second, reading this gives you a sense of what themes the Stanford Daily opinions section holds. A caveat to this: My data set was quite small, and I theorize that having more opinion columns in the text file would vary the topics discussed in this randomly generated column.
So I leave you with a philosophical question. Is this randomly generated text the computer’s opinion? If not, why not? The sentences are as real as anything you or I could type, give the input that it had. If the program’s data set had as much information and experience that an average human being had, could we tell the difference between what the computer wrote, and what a Stanford student wrote?
Contact Maika Isogawa at misogawa ‘at’ stanford.edu.