Sure and that's a new technique for indexing and querying.
Where's the new design tension? Indexes always had to be monitored for freshness and queries have always needed cleaning or parsing.
1,019 karma · joined July 26, 2024
Where's the new design tension? Indexes always had to be monitored for freshness and queries have always needed cleaning or parsing.
How is RAG any different from the search systems we've been building before LLMs? Is it the sudden need for everyone to design a search API and engine that's driven this trend?
If so, I'd like to see more design patterns around existing search problems:
- Correcting or backtracking based on feedback.
- Measuring relevance.
- Comparison with task-based pre-written queries. Does every LLM task need a full blown search engine? Why not a tightly scoped domain API for data retrieval?
I don't want to think of memory management in most projects, I want to focus on business logic, so I use Node.js instead of Rust, even if the latter gives me more control.
Similarly, I don't want to think of DOM management in most projects, I want to focus on data dependencies, so I use React instead of Web Components, even if the latter gives me more control.
No, spreadsheets popularized reactivity. And the general point is incredibly weak.
Don't use frameworks and make your own? Sure, have your fun. But then try teaching your framework to your company of 1000 and see how quickly you realize your view of the "problems" are only a slice of the pie.
The whole post felt like a personal criticism of Jarred.
How do you justify this when you factor in cost and time?
I don't know, I understand the principle, but I don't see how you can determine the value of a principle outside of a specific context.
Even for accessibility, we can't target every context in the name of being accessible. We still have to pick which contexts of inaccessibility we'll need to support with more attention.
Engineering attention is finite. Why would you spend time thinking about 8 clicks when most people will only need ~3?
Not all user-action possibilities are equally important, and if they are, then you better have infinite resources to spend on engineering.
So, now, I can hate with cause: it reads like someone who cares about what their MFA friends think.
Meaning, it puts most of its emphasis on description, and so little on situational engagement. Which makes sense, I suppose, for an LLM.
I wrote about this from the perspective of someone with no connections [0] but I think even if you're well connected, reaching out to people from other networks is a useful way to gain access to great thinking, information, and opportunity.
[0] https://www.nair.sh/guides-and-opinions/marketing-under-pres...
2. Can I ask how? When I was building something like this, I realized there's an element of burning tokens for correctness. Meaning, splitting things into small units and small processes, each using a separate LLM output to be later combined. For a 1k page document, what kind of token usage do you see?
Over the long term, democracies have mostly avoided self-implosion, so this political see-saw does seem to work.
The question is: is there anything we can do to make it feel better to live in?
You're saying it "works" for some people but maybe not in a way or scale that was originally intended for it to "work".
But isn't that how all learning in a complex system takes place? Doesn't mean you need to rush to cancel something?
What's wrong in saying it doesn't do what we thought it would but it does do something useful? Maybe adjust further investment based on results obtained. Ignore sunk cost.
The political experiments should be chosen and driven by small units in order to be safely contained.
Planning shouldn't be a top-down activity, strategy should be. Meaning , the bounds for decisions and the rules of play must be centrally determined, but not the experiments themselves.
I don't completely follow the last bit of your comment. It sounds like you agree that a centralized planner doesn't have enough context to make system wide decisions. Am I right?
It's not enough to try things, we also need a social system in place to allow trying to be safe and useful, even on failure.
We're missing guardrails to allow safe experimentation and we're missing institutions to provide affordances.
I think the difficult bit is figuring out how to seperate the goodness of centralized decision shaping and the badness of centralized power accumulation.
In the past I'd built an internal tool that transforms insurance PDFs to structured data. I wanted to extract explicit data dependencies between fields to perform validation.
Insurance forms can sometimes have 30-40 pages and they can have fields on page 40 that depend on fields on page 4 with a few nested if conditions. Would Parsewise be able to extract those relationships?
If yes, how do you do it for large documents?
Can you simplify how form dynamism works? I skimmed the docs and saw 'states', but it didn't immediately click how it works.
Do we build a tree of rules outside of the components? Are states attached to each component, bottoms-up, and then the form tree is managed by the library?
And it works, to some degree.
And how do teachers teach? They don't start by trying to argue or by trying to prove students wrong. They teach by showing what's fascinating.
Taking the time to show people what's fascinating, what's perplexing, where the tension lies, and how it's resolved, is teaching.
Argument construction in social contexts is ironically ego-driven. Demonstrating something interesting, on the other hand, means asking yourself what what they would find interesting about what you want to tell them.
The scientists do,
Step 1: Build Model
Step 2: Think of implications
Step 3: Check if observations make sense based on implications
Step 4: If wrong, refine model or go back to Step 1. If right, submit to other people and ask them to verify.
The consultants do,
Step 1: Build model.
Step 2: Tell people this is the right model.
I understand design can have many goals, but surely it should atleast feel like ... something good? I've never once used a Google interface and felt anything good.
How ... how is that legal? Why would that ever be made legal?
Apparently businesses can use contracts to opt out of regular public courts and agree on using a neutral decision-maker; an arbitrator.
But then the post says:
> Meta got its arbitrator – a lawyer who is paid by Meta to adjudicate contractual disputes instead of an actual judge
Huh? How's that legal?
Turns out, the law requires arbitrators to be neutral, but not the people choosing the arbitrators.
Arbitration services are businesses. So even though Meta doesn't directly pay the arbitrator, they pay the business picking the arbitrator.
Meaning, Meta has a long-term relationship with the arbitration service provider. They can choose to take their business elsewhere, if unhappy.
Imagine being Wynn-Williams, having a company of this size put a target on your head. I wonder how many live in silence because the paycheck is too good or the punishment too bad.
But an even larger point: most of HN is probably employed by a company that aspires to be Meta; HN is run by a VC fund that wants to make many Metas; and worse, unfortunately, I sometimes dream of being a Zuckerberg.
I am thoroughly seduced by a power I've never felt, even if I see it as poison.
The demo is not egregious by itself, it's the choice of the demo for this circumstance. Why boy-hits-boy-with-rock-for-girl for an animation DSL?