Waypoint 1 - 9/16 Pre-Class Reflection
During these first two sessions, I found communication to be most effective when in small groups. More specifically, communication was most effective when we had broken into smaller groups to work on a given assignment. The format made it easy to share one's understanding as well as for people to ask clarifying questions as needed. One person's understanding became the group's understanding very quickly. In contrast, in larger group formats, or in inter-group formats, where back and forth communication was slower or not possible, this was far less effective. The ability to ask clarifying questions and to have a back and forth between members was integral for successful communication. Even when creating supposedly comprehensive directions, there were little things that the authors considered implicit that were lost, resulting in a various different mistakes. The ability to quickly get clarification would have solved this very effectively.
As far as features that support communication, the biggest thing that comes to mind, both from the activities and from this reflection, is that communication needs to be open and reciprocal. It is incredibly important to be able to ask questions, and have a brief discussion about them. I would argue, it is far more important to be able (feel comfortable/have space) to ask so called "trivial" questions. Quick clarifying questions are often the most efficient at resolving confusion in previous communication. Everyone makes mistakes, and no explanation or set of instructions is going to be perfect. Being able to ask, "what did you mean by this?" or "are you sure about ____" is incredibly important for ones understanding. For communication to be effective, it must go both ways. Simply sitting and listening to a lecture (or the written equivalent) is far, far less effective.
Waypoint 2 - 9/23 Networking
To form a networking connection, I reached out to Andrea C. Villanti at the Rutgers University - New Brunswick. I selected specifically because I could apply an "algorithm" with which I was already familiar.
To break down that process slightly more:
Research a topic of interest, ideally related to a current area of research/study.
Identify a person who has published in that area, the more recently the publication the better.
Read the work in question carefully.
Reach out to the author(s) in question, ideally with genuine questions/interest regarding their work.
Most academics will eventually respond, as they tend to really enjoy talking about their work.
In this case, I was introduced to their recent paper "Public Comments and Federal Rulemaking for Menthol Cigarettes and Flavored Cigars" through my current work. I am currently working remotely as a research assistant in lab looking into the same FDA Docket, if with a different process and research question. Naturally, when this paper was published, the group noticed. Since it is in an area related to something I am currently working on, I believe I was in a position to ask more potentially interesting questions. Additionally, even if this did not result in a long-term professional relationship, I would probably be able to glean some insights useful for my current work.
Andrea Villanti is listed as the first author on the paper, so I reached out to her. I have not yet heard back. However, I do not find this particularly surprising, especially given the short time frame. Academics tend to be a little bit hit or miss on whether or not cold emails actually get answered, particularly when they come from outside their institution. And when they do get answered, it can often take a little bit. So, I remain hopeful that I will hear back at some point.
Waypoint 3 - 9/30
Forrest - Genetic Algorithms
- Modeled after the process of evolution
- Process
o Create random population
o Judge fitness
o Recreate population based on best members, allowing for mutation
o Repeat
- Argument components
o Initial independent sampling via randomization
o High fitness individuals are preserved, improving overall fitness
o Crossover combines elements from multiple strings, increasing probability of success
- Implicit parallelism
- Schema Theorem – genetic algorithms guarantee exponentially increased samples of the best schema in the next step
- Wildly applicable
o Multiparameter function optimization most common
- Not universally successful
Moseley – On Dice Games:
- People throughout history have continually tried, and failed, to meaningfully replicate variants of art via mechanical constructions
- Wurfelspiele: German dice games
o Referencing a specific one or the category?
o Operates as a form of ludomusical machine
- As musical machines have advanced, their failing to surpass human creativity becomes even more stark
- Games as a way to reframe the unknowable
- Such constructs often emphasize an air of mystery in their function
- The rules/restrictions of such games, while patently arbitrary, can encourage creative problem solving
- Does music have a functional baseline/algorithm that can be meaningful taught/programmed?
- Successful versions of artificially generated music heavily rely on human creativity/intervention
While both pieces are deeply invested in the algorithimic process, they take a very different approach. The Forrest piece focuses on a specific space in which algorithims, while sometimes amorphous, have concrete start and end states, as well as no need for human intervention beyond its conception. In contrast, the Moseley piece is most concerned with an area in which pure algorithims fail. In the field of music, while you can create certain baselines or rules to guide creation, every interation that is not ultimately reliant on human creativity is doomed to be reduced to all but gibberish. This piece is about an experimental field, rather than an established one.
I feel like these pieces are fundamentally quite different. The Forrest piece provides a firm description of the process for solving a certain set of problems while the Moseley peice, in contrast, serves as more of a theortical framework through which algorithimic attempts at creative generation may be viewed. They are both interesting, if for very different reasons. The former is overtly applicable to a variety of computational problems, both solved and not. The latter, in turn, opens space for a more playful discussion over the fundamental necessities of creative works. I find this piece more gnerally interesting to consider in the present moment, with the way LLMs are changing aspects of many artistic industries, including music. The most successful of these attempts, that I have seen, as the essay describes, reuslt from a commbination of mechanical generation and human intervention and creativity.