A key line seems to be buried in the "Summary" at the end of the post: "Python native enums are great for what they were designed to do, but..." Barring tricks I'm unaware of, enums in other popular languages, such as Go and TypeScript, don't support the author's more complex usage either—probably by design.
Out of curiosity, are there languages for which the enum implementation supports and encourages encoding complex data like this?
Often called sum types, it's really convenient when your language supports both building these kinds of enums as well as destructuring them with pattern matching.
If you don't have destructuring, they're less attractive. There's a proposal to add destructuring to python[1] so we'll see
The paper may not offer insightful prescriptions for experienced engineers, but can work like this still be useful for informing future studies in a meaningful way? The authors repeatedly note the widespread inadequacies of the current research landscape. (To anyone familiar with the literature, is their assessment accurate?) In my eyes, the message is that the paper represents an incremental step in the direction of a truly detailed understanding of the factors involved in developer productivity. Even if there's a broad intersection between the answers one would get from taxi drivers and from computer scientists, the distance between the two fields makes that an unexpected result, which should prompt us to change how we think about computer science* and/or how we think about studying it.
> The authors repeatedly note the widespread inadequacies of the current research landscape.
This is standard language in academic research papers. It is there to sell the importance of the research to the reader, in particular to journal editors or peers who review the article. It is mere puffery.
One of the linked articles [1] describes the arguments in favor of pro-ISP restrictions as concerns that municipal broadband would either be a waste of taxpayer money or an "unfair" threat to private-sector ISPs: "'The general rhetoric behind these laws, from the incumbents, is that cities are too incompetent to run their own networks, so it's a risk to taxpayers,' Craig Settles, a broadband consultant who works with cities to create municipal networks told me. 'But then, the other side of it is that cities are so competent that they represent unfair competition.'"
Is this a relatively accurate and complete characterization?
Not as contradictory as the speaker is making it out to be. The organization can be poorly managed but still price out competition via a state subsidized business model that would be unsustainable for anyone else.
I've been using Notion (https://www.notion.so/) for Kanban-style boards for a few months. After using Trello, Asana, and Jira, it's the only one that does basically what I want it to without spending hours on setup (Jira) or sheer annoyance (Asana). It won't be perfect for everyone, but for me Notion's task management is good enough and actually a pleasure to use so far.
Notion is great. It gives you so much flexibility to really mold the software into what you need, I would highly recommend it. As freelancer I use it as a Jira replacement but I also have personal to-do lists, etc. in Notion and it’s nice having them all in one place.
I'm curious about the "fine-tuning based detection" mentioned in the report ("Fine-tunes a language model to 'detect itself'... over a range of available settings"). Does anyone know good articles/papers (or have an off-the-top tl;dr) to get a high-level grasp of "self-detection" for generative models?
Hiya, I work at OpenAI. I think the Grover paper is a good place to read about some of this:https://arxiv.org/abs/1905.12616
We're likely publishing more on detecting fine-tuned outputs in the future, also.
Out of curiosity, are there languages for which the enum implementation supports and encourages encoding complex data like this?