His example of a web crawler stinks. I can get behind the philosophy, but his specific web crawler example at a minimum needs to retry, respect robots.txt, and rate limit per domain. So no, xargs + curl is a dumb example.
Replace curl with wget, get robots.txt support. Write timed-out URLs into a file for the next identical run.
The idea stays: you can jerry-rig a web crawler from pedestrian, readily available parts, and quickly test the idea for which you wanted to crawl the Web.
The point is that bash, xargs, wget, make, etc are well-understood, well-maintained tools. They are not even necessarily simple, but they likely contain very few unknown unknowns.
Based on bot traffic to my personal blog over the past year or so, the minimum viable web crawler apparently does not need any of those things (much to my own personal annoyance, to be very clear).
Which is why I've never quite understood the appeal of portable SIMD libraries for performance-critical code. If I'm explicitly writing SIMD rather than relying on the auto-vectorizer, it's usually because I want access to the particular capabilities of the target ISA.
For many problems, choosing the right instruction or instruction sequence makes a large difference. Portable SIMD abstractions necessarily expose some common semantic layer, but SIMD ISAs don't actually have equivalent capabilities. Instructions like pshufb, for example, enable algorithmic tricks that don't necessarily have an equally efficient analogue on another architecture.
If maximum performance matters, I generally want intrinsics and architecture-specific implementations; if portability matters more, I'd rather move further up the abstraction stack and use something designed to target multiple architectures, such as ISPC. There are certainly cases where portable SIMD gets close enough to optimal, but I don't think there's a compiler or abstraction that can express every useful SIMD idiom and lower it equally efficiently across fundamentally different ISAs.
Because usually they achieve a very good middle ground, they are useful for when autovectorization isn't good enough, and it is possible to give a little help to the compiler.
There are many ways that performance matters without trying to win a F1 race.
Go isn't alone, .NET, Java have similar portable libraries, and C++ is in the process of getting one.
One I spent a few months working on was pathfinding for trucks. The goal is to find dominant solutions over {shortest time, lowest cost (tolls + fuel), avg road speed variance - traffic sensitivity} and then return 3-4 routes that are equal distance from each other in this dimensional space for users to pick from.
As you say, the most useful things happen in low-dimensional spaces.
social media is a completely different market though - since there are massive returns to scale, it's incredibly hard for a new entrant to break in.
model training and inference is the opposite. people switch LLMs like people change clothes in the morning. there are popular services (openrouter) that make moving as easy as changing a model string.
Which poses the question: Why does Google even need to catch up? At least currently, the name of the game is integration. The actual model is a commodity.
Is Gemini really doomed? I'm still bullish on Google:
1) they have more free cash flow and capital than God due to the ads business
2) they have data - intent from web searches, youtube videos, google books and music
3) they have dedicated inference hardware
for all these reasons, is being 6 months behind the frontier actually a structural, long term disadvantage? some day the pace of improvement will slow, and google will vacuum up the market. they'll be able to compete with open-weight models just on pure cost advantage from their vertical integration
That also own one of the 2 major mobile operating systems, with Gemini tightly integrated and all of the data they can gather from that. Why do you think OpenAI wants to do hardware? Owning delivery is going to be important, and right now, Google and Apple own the delivery mechanisms (to consumers).
They may not capture enterprise use, but I don't think they have to. That's only one piece of the market. AI that's useful to consumers will still get delivered via a smartphone, and Google is in a great place to capture that.
I also don't think LLMs have to be a "winner takes all" situation. Value isn't going to come from having direct access to a chatbot or selling API inference, value is going to be in the form of a specific product (for most, devs aside here). Something a consumer, or a non-tech business can buy off the shelf and plug and play. A "ready made" customer service agent system, a "ready made" BI platform using AI, etc.
For consumers, that's probably going to look like whatever is bundled and tightly integrated into their mobile OS of choice.
My prediction is that the EU will eventually bring in legislation that will force platform providers to provide pluggable APIs so that you can use whatever LLM you want.
> is being 6 months behind the frontier actually a structural, long term disadvantage?
Yeah, this is one of the things I find so weird on the discourse. If you get there negligeably later, but without astonishing spend and waste, you might even be better off in the long term.
I have thought that for a while and assumed that was apples approach to AI. Wait until everyone burns through investor cash, invents the better tech, and can monetize. Then copy that business model and polish it, or buy out the competition and polish. That’s usually apples move and it makes sense for google to do something similar.
"Doomed" no, but it's pretty clear that they just had a bad cycle and are struggling to keep up with the frontier.
Whether this happened because they bet on "world models -> better reasoning" and that bet didn't pay off, or failed a frontier run for technical reasons like OpenAI did with 4.5, or something else went down? We don't know.
Will they bleed talent, fall further behind until they give up, or clean the organizational and infrastructural cobwebs and get back in the saddle? We don't know.
So long as search revenue isn't correlated to AI revenue or threatened by AI revenue, Google at any point can bail on AI spending and suddenly the free cash flow machine is back on. Yeah they have a far larger debt load they now have to service but cash flow from search revenue is so strong it wouldn't be much of a blip.
I think that's backwards. Anything that changes the performance of a model deserves a minor version bump. A new model has to be qualified before being pushed to production; but we don't get the choice here, just cross your fingers there are no regressions at all on all possible tasks the model might be asked to do.
I should have worded it differently. Having a model selector as `string` seems like the wrong choice. At the very least, I should be able to select 4.X, like how we select dependencies, so people who rely on specific behavior can pin to 4.0 and I can say version >= 4 if I don't care.
That's on you for using a preview/beta. It was properly qualified when it was released. No one complains when ios goes from beta to release; even though it will be ios 27 beta -> ios 27 next month
its an open source model if it is important enough that you have to worry about a new version breaking something then why the fuck was that not running on your own servers this is not an Anthropic or OpenAI closed model that you only have access through an api
ironically, any frontier LLM will easily generate a tutorial at any detail you like explaining what these are.
if don't have time for that, just know that these are technical parameters that affect how likely it is an llm will produce the same result after being asked the same question.
An LLM is stateless so every time you send a message a command is run with the parameters, the prompts, and a copy of the chat history when using one of those chat interfaces, and then it returns text. If the output includes explaining it to an LLM that would then be included in the copy of the chat history placed in the new command.
It would be a bit like opening a text editor and typing in that you want to increase the font size. Someone external to the text would have to come along and click the font options.
Yes but it implies it's doing its own "brain surgery" so if it got params wrong it could break itself. In that case you would have another AI behind the scenes doing the tweaking so it can understand you at a "subconscious" level.
using sql everywhere; never having to remember dataframe syntax: timeless