4,763 karma · joined October 1, 2009
I guess this just isn’t obvious to me. Where do you see the boundary between “interface to the os” and “programming language”?
> Verification becomes hard to reason about because there is no ground truth for points of interest, there are no red/green unit tests for taste. I’m sure these are familiar challenges to data scientists and that there are frameworks and evals for working on them. This will require more iteration and manual overrides. Hopefully with feedback and collaboration from the community. But for now I’ve shipped V1…
I suspect LLMs may be able to help us quantify our taste because they can keep track of so many data points all at once, where we have to lossily abstract these details away.
Exactly. There's little "win" here for the business, especially when the simpler "just cap at n requests" solution exists. Pursuing this solution:
- introduces more technical complexity (increased cost of developing, testing and maintentance)
- forces overage costs onto the vendor, who will then... increase prices to cover these costs, which upsets customers and makes the company less competitive.
- doesn't improve cost prediction (since n requests can be costed out a $/request quite easily and reliably)
(The cynical will say something about "the business has an incentive to _not_ fix this issue because they get paid for overages", but this assumes that Cloudflare has the, "monopolistic" market power to make customers unhappy and still charge whatever they want. In this case implementing billing caps wouldn't do anything because Cloudflare could just... charge whatever they want.)
In one case I was literally repeating back to the human what I’d just been told, and getting them to confirm that what I said was correct. First bill arrives and I find out the truth.
Second case I was told I’d have to cancel and create a new account to add a service. I decided to keep my existing account and learned that there is a web page where I can easily self-serve and add the additional service in one or two clicks. (I assume like the human actually made more money for “new account signups”.
My point is that the feeling of being a valued customer is really independent of whether you’re interacting with a human.
Repeatedly requesting time/budget to fix an ongoing issue is a requirement of any half-decent manager. If they’re reporting issues then just smiling blankly when asked “what can we do about it?” they’ve failed their basic job duty.
The argument here is that all code reviews are done with attention and care, but quality of a code review is highly dependent on the reviewer and the team’s review process, and in the real world the quality of reviews pretty much follow the same distribution curve as, say, agile project management: For the time invested in reviewing, a handful of teams get excellent utility from them, most teams get little benefit, and a sad few actually cause harm.
If most code reviews provide only a little benefit at base for most teams, recommending that most teams should also delay shipping quality work is going to sound a lot like bad advice.
> - Security: Each container has the isolation properties of a full VM, using a minimal set of core utilities and dynamic libraries to reduce resource utilization and attack surface.
> - Privacy: When sharing host data using container, you mount only necessary data into each VM. With a shared VM, you need to mount all data that you may ever want to use into the VM, so that it can be mounted selectively into containers.
> -Performance: Containers created using container require less memory than full VMs, with boot times that are comparable to containers running in a shared VM.
More details, including technical limitations (they’re looking for bug reports and contributions): “Container: Technical Overview” https://github.com/apple/container/blob/main/docs/technical-...
Bottom line is we didn’t have a measurement of safety before the port, and we don’t have one now.
What we do have is a known list of unsafe blocks, and we can use that as our safety measure. (I’m neither a Zig not Rust programmer, but I’m guessing that the unsafe parts of the Zig codebase were also mostly measurable so we could have had this measure.)
I do wonder if the next step is to move bun into WASM for an additional layer of security. Those unsafe blocks might be neutered by not granting WASM the ability to run them. That would give anthropic a “sandboxed by default” opportunity.
It’s a fun project to watch!
> It’s one that we’ve been facing for quite some time. The reason we wrote this report, however, is to highlight the fact that we’re sort of in this moment in time right now, with our relationship with the U.S. deteriorating and us trying to diversify our trading partners, to highlight the fact that we are still not really all that competitive. Our productivity growth is quite low and has been for a few years now. So, banging this drum about wanting to raise this issue around competitiveness, that was the goal of this.
With the U.S. moving from a cooperative trade partner to a trade competitor, Canada needs to up its game.
> Iran has started a Bitcoin-backed insurance service for Iranian shipping companies that want to transit the Strait of Hormuz, the semi-official Fars news agency reported, citing documents obtained from the country’s Ministry of Economy and Financial Affairs.
> According to a screen shot of the insurance company’s website, dubbed Hormuz Safe and shared by Fars news, it “provides Iranian shipping companies and cargo owners with fast, verifiable digital insurance.” Fars didn’t give a detailed break down of how the insurance works and whether it’s available to foreign shipping companies and vessels.
A better example would be if you’d changed the behavior of the library as you did this work, and the library changes introduced hard-to-detect bugs across the application.
The issues:
> The New York and California pension systems would become holders of SpaceX shares through their passive allocations if the company is admitted to major U.S. stock indexes.
> The officials… objected to the amount of power the board has given Musk over the company, including:
- voting control over the stock,
- veto power over his own removal as CEO, and
- protections from litigation, including mandatory arbitration for SpaceX shareholder claims.
> …
> In their letter, the pension leaders urged SpaceX to:
- adopt one-share, one-vote or sunset super-voting shares within seven years;
- install a majority-independent board and separate the CEO and chair roles;
- eliminate provisions protecting Musk from termination without his approval;
- scrap mandatory arbitration; and require independent approval of related-party transactions with Musk's other companies.
(Formatting mine; moved paragraph about becoming holders above the lists of concerns and recommendations.)
Gall notes that the universe naturally trends toward complexity and unintended consequences and therefore complex designs should be assumed to already be full of these unintended consequence “bugs”. He proposes that systems should be designed:
- with less scope to reduce unintended consequences,
- with less rigidity to allow for workarounds when unintended consequences arise, and
- to take advantage of “momentum” to reduce the required energy to use the system correctly. In other words make the right thing the easy thing, remembering that the easiest thing to do is nothing, thus systems will halt if operators get too busy with other tasks.)
The only claim here is that there’s a report that tracks GitHub star growth that is that is presumably read by VCs:
>> Union Labs is the most consequential case. It was ranked #1 on Runa Capital's ROSS Index for Q2 2025 - a widely cited VC industry report identifying the "hottest open-source startups" - with 54.2x star growth and 74,300 stars. Our analysis found 32.7% zero-repo accounts, 52% zero-follower accounts, and a fork-to-star ratio of 0.052. The StarScout analysis flagged it with 47.4% suspected fake stars. An influential investment-sourcing report that VCs rely on was topped by a project with nearly half its stars suspected as artificial.
Again, the claim here is that the report is “influential”. Maybe?
Union Labs is the most consequential case. It was ranked #1 on Runa Capital's ROSS Index for Q2 2025 - a widely cited VC industry report identifying the "hottest open-source startups" - with 54.2x star growth and 74,300 stars. Our analysis found 32.7% zero-repo accounts, 52% zero-follower accounts, and a fork-to-star ratio of 0.052. The StarScout analysis flagged it with 47.4% suspected fake stars. An influential investment-sourcing report that VCs rely on was topped by a project with nearly half its stars suspected as artificial.
i.e. Nursing jobs mostly go to women not because men can’t do them because “nurses aren’t men”, per our current cultural norms.
Cofounders and founding engineers just don’t talk enough about how they model the financial risk of investing in a high risk asset like a startup, and how they make sure that they’re not taking on more risk than they can personally manage if things don’t work out. Approaches will vary by person and personal circumstances, so it’s also helpful to know what constraints they’re working with (like needing to house and feed their family).