Very slick and potentially very powerful. After a few minutes playing with it, I have a few recommendations:
- Variables should have more types, like an array of objects
- Prompting should incorporate Jinja2/Nunjucks
- For every prompt, I should be able to create many different test examples, along with an answer key, and measure how well it does across many tests
- It should auto-save. I did a lot of prompting work and then clicked another icon. When I came back, all my work was gone. (In fact, I don't see where to save at all! Maybe I'm just missing it.)
Casetext | Software Engineer, Research Scientist, Machine Learning Engineers, and More | San Francisco, CA | ONSITE
This is an insanely fun time to join us. If you are interested in search, NLP, recommender engines, cutting-edge front-end technologies, or making a difference through your work, you should check us out.
Our startup, Casetext (YCS13) was mentioned a few times here, so I thought I'd stop in.
The crux of the article is that most legal research solutions have ignored the immense power contained in the links between laws:
> Laws frequently reference other laws in order to reuse definitions, introduce exceptions, or make it clear that two concepts are meant to work together. Consequential laws tend to get referenced in other laws as their influence spreads throughout the legal system. Experienced lawyers build up detailed mental maps of these links, allowing them to jump immediately to core issues of complex legal problems.
> However, most laws can only be searched using the dark-age, Lycos strategy—guess at keywords and hope—and it’s often necessary to pay for even that limited functionality.
We at Casetext are taking a very different approach than the "dark-age, Lycos strategy" that you have to pay for:
1. On Casetext, the law is free, as is basic search. Honestly, it's insane that Westlaw and LexisNexis charge as much as they do for basic keyword search over a database that should have been free to begin with.
2. We make money by charging for advanced, data-driven ways that lawyers can research more efficiently. CARA, our premium product, enables a lawyer to drag-and-drop upload a document they're working on, and will recommend the research that the lawyer missed but is very relevant to what they're working on (https://casetext.com/cara). A key ingredient behind this awesome tech is the network of citations that the article mentions.
Whether it's us or other startups, I agree with the article that in the next few years you'll see a trend towards more "Google for Law" -- companies will make legal research free, and their comparative advantage will be on their technology, often driven by ML/AI. As a lawyer/coder, it's a pretty exciting time to be in the space.
Sure. Something to monitor reddit for highly upvoted and interesting content. Once it hits a certain threshold, add the post to WP. Generate the cache entry and you're pretty well set.
Well, I was commenting in particular on the fact that the site is currently a private WordPress server. The plan appears to be a bit more involved than aggregation, comprising interviews with Reddit OPs and curated articles written by Reddit staff themselves (although they’re probably hiring mostly new people to work on that).
A really good book to read on the subject is AnnaLee Saxenian's Regional Advantage: Culture and Competition in Silicon Valley and Route 128 (http://www.amazon.com/Regional-Advantage-Culture-Competition...). It discusses why Route 128 failed while Silicon Valley flourished. A major part is that under California law, non-competes are not enforceable in California. As people moved freely between competing companies, their ideas, information, and best practices traveled with them. The diffusion of good ideas gave the region as a whole a competitive edge.
Yes - this is an instance where a law can help an ecosystem at the expense of individual players. In aggregate though, it's net neutral for most companies, since they benefit from being able to hire from competitors. The companies doing interesting things probably benefit, while the ones doing boring things get a little hurt.
The (largely untrained) economist in me says that mobility of labor is very important for growth and economic efficiency, as systems work best when the best people flock to the most important well funded ideas.
Thanks. You touch upon one of my pet peeves. A huge amount of legislative energy is spent to ensure that capital can move around unfettered. On the contrary, movement of skill and intellectual property seems purposefully hindered. Implicitly it says that capital is much more important than the others, I simply disagree.
There's an economics term for this, though I'm spacing on the name. It's the same reason the taxi lobby is so strong - it's a subset of 1 issue voters with concentrated financial support of key issues. It's also behind why tax loopholes are so hard to close - a small subset cares a lot, but the masses less so.
It helps for capital to have cheaper labor, and vice versa. :-) The system grows best when both are free. (Incidentally, this is also why I'm ok with stock buybacks. If companies can't put their money to good use, retire the stock and let people invest it elsewhere)
It's not a net neutral for most companies. It's a huge benefit for all the companies in that area. Skilled professionals have a choice of region and company to work in. IF they choose one with better terms for them, then that region will have a more skilled pool of workers. This benefits companies in that region. One could argue that the stronger ecosystem and pool of employees creates significant benefits that outweigh any corporate advantage due to restricting employee mobility. That's what the above reference is arguing.
Completely agree with your awesome comment except for one niggle...
> it's net neutral for most companies
My take is that if there were only two companies this would be true. Considering that no company employs more than a tiny fraction of the skilled labour market the vast majority of labour innovation occurs elsewhere.
A company would have to be doing an immense amount of innovation in order to not have a net benefit from free movement of labour.
We could talk about whether the Googles, NSAs, or Amazons of the world hit this threshold, but I'd argue that nobody else does.
And in response, California firms repeatedly engaged in (illegal) noncompete and anti-poaching collusion to both drive down wages and limit labour mobility.
I'd definitely like to see some personal penalty. Though Jobs is already dead.
Worth noting: much as I despise Facebook, Zuckerberg did the Right Thing here and didn't play along (though it's not clear whether or not FB reported the collusion).
For what it's worth, the author, Ira Rothken, has been litigating these cases for well over a decade. Notably, he is Kim Dotcom's lawyer for Megaupload-related litigation. It's really cool that he's sharing some creative thoughts about the different safe harbors ISPs can use to fight copyright claims.
Keep in mind that there are a lot of lawsuits that are done for strategic reasons, even where the factual grounding isn't super strong. I'm not saying that's necessarily happening here, but it could be the case.
- Variables should have more types, like an array of objects
- Prompting should incorporate Jinja2/Nunjucks
- For every prompt, I should be able to create many different test examples, along with an answer key, and measure how well it does across many tests
- It should auto-save. I did a lot of prompting work and then clicked another icon. When I came back, all my work was gone. (In fact, I don't see where to save at all! Maybe I'm just missing it.)