Schema.org

Schema.org is useful markup to have on website as it makes it search engine friendly while helping them understand the content and internal structure which enables better search results. However, the schema.org website lacks clarity and is difficult to navigate like a clutter of information.

Benefits:

• Enables search engines to better understand content on sites that rank higher on search results
• Improve click-through rates and organically increase traffic on site 
• Provide more flexibility and context to how sites appear in search results
• Increase user search relevancy
• Improve strategy for content and context
• Improve user experience
• Flexibility on markup from microdata, rdfa, and jsonld
• Provides a meta vocabulary to define the context of the site
• Extract who, what, when, why from sites 

Drawbacks:

• Unfriendly schema.org community for suggestions, feedback, and improvements
• Submitting new changes or schemas is slow and often fraught with frustration 
• Terrible and difficult to navigate schema.org site as the information is cluttered
• Community is not very open and unwelcoming to new users
• No real reasoning and significant effort towards web of data queryability 
• Community is discriminatory towards user suggestions, submissions, and approval process 
• Very opinionated and closed community which makes it unconstructive 
• Huge Google bias with often rude and arrogant community members 
• Markup often is buggy, flawed, and inflexible to community changes 
• Process is fraught with trial and error
• Difficult to develop a strategy around the markup 
• Difficult to implement at scale with larger websites 
• Maintaining markup is a challenge 
• It is subjective and questionable whether the markup significantly improves discoverability 
• Limited tools that support and provide insights into the markup 
• Inflexible schema.org developer community makes the standard inaccessible, inextensible, and unmaintainable 
• Unclear documentation on the schema.org website 
• The markup is still very limited in context and scope especially for larger websites 
• Lacks sufficient domain coverage as a markup

Although, the project has a long history and many websites make use of such markup, it has significant drawbacks that justify alternative efforts. The project is also Google sponsored with a significant corporate bias which defeats the merit and accessibility for an open community engagement. The often slow process means the markup lacks speed in reaching its full potential. An active open community may speed up the process but this is likely to be a significant roadblock from the existing community of developers who are not very forthright with community engagement. Bugs in schemas take a very long time to resolve and usually recommendations are not appreciated in the community. There is significant concern for websites to use such markup where the community is often unapproachable and inflexible. After all, it is supposed to be a web standard which arises from a community effort and engagement. Lastly, a markup that lacks readability, reasoning, and trust as a web standard is likely to be insufficient for the spectrum of web crawling, semantic search, web of data, and AI in general.

Schema.org

Nostr

Nostr

Netflix Terrible Recommendations

Netflix is all hype. The quality of recommendations and personalization is terrible and lacks variety of content. The entire flow of rating system looks flawed. And, when you refresh the browser the same content that you thumbs down on reappears at the very top. The infinite scrolling is annoying. The algorithm does not personalize to watch habits of a user. Majority of the recommendations seem to be new and irregular. 

Netflix has an inaccurate and insufficient data collection gathering process which leaves an incomplete dataset for a recommendation model. The model is neither sufficiently able to gather how you use the platform nor how you don't use it while ignoring user interests and intents.

Netflix algorithms try to measure correlation but not sufficiently causation. It is not able to answer "why". This is in fact the whole point of a recommendation algorithm to utilize insights in order to make deeper contextual decisions on personalization to match items to users. 

Netflix algorithms lack sufficient reasoning skills to understand the habits of the user to provide better recommendations in respect of context, intents, and interests. The connections drawn between two points of data seem blurred. This may be as a result of tastes that are not fixed but tend to change. Unfortunately, there is also lack of filters for the user to provide additional data. This could include preferences as part of user profile. The fundamental filtering attributes of grouping items with the user in context seems to be missing for quality recommendations.  There is also little to no common sense in the recommendations. It often seems like the user will be trapped in a bubble of sorts. Also, one would assume that they would also recommend their own produced content to recuperate production costs through the platform and take benefit from user data. Furthermore, this may even provide insights on future production projects.

Finally, there is a significant lack of content variety on the platform. A huge bias towards Indian content over other regional content. The content library needs a complete reboot with more flexibility on user preference filters. And, an increase in frequency of new content to be able to sustainably compete with other streaming platform providers.