WEEK 13

 

The semester flew by. It is already the end. Here is the data that I gathered from this experiment.

Inputs	Incognito / Normal mode	Before / After search	Number of relevant videos	Percentage of relevant videos
Cooking	Incognito	Before	6	5.0%
	Incognito	After	118	98.3%
Flying	Incognito	Before	3	2.5%
	Incognito	After	117	97.5%
	Normal	Before	2	3.3%
	Normal	After	58	96.6%
Reading	Incognito	Before	0	0.0%
	Incognito	After	114	95.0%
	Normal	Before	0	0.0%
	Normal	After	115	95.8%
Sleeping	Incognito	Before	1	0.8%
	Incognito	After	120	100.0%
	Normal	Before	2	1.7%
	Normal	After	120	100.0%

 

Before putting the inputs onto a search bar, the outputs with Incognito mode or without Incognito mode have very low similarities to the inputs. After putting the inputs into the search bar, the outputs with Incognito mode or without Incognito mode have a range between 95% and 100%. I expected I would see more related outputs to inputs before putting the inputs as the experiment progressed. My hypothesis is that the outputs are more influenced by arbitrary inputs in the YouTube algorithm as data sets are collected. My hypothesis is not supported by the data. However, I think this hypothesis may be supported with a different approach such as changing inputs, watching certain videos, clicking certain things and so on. It was fun to do this research and consider what would make the results different.