HNHacker News
TopNewBestAskShowJobs

mbauman

4,240 karma · joined February 3, 2012

mbauman.com
submissionscomments
mbauman··on Terrapin Attack for prefix truncation in SSH
> I have in the past suggested that the fact that the MDC has remained unbroken for 20+ years is evidence that it is secure.

This seems like... not a great stance to take on post about a vulnerability that existed for 20+ years.

I don't understand the defensiveness.

mbauman··on USB-C Cures Mosquito Bites
In my experience, it only really works if you do it within 5 minutes or so — but if you do it's absolutely magical.

It's even better for bee stings.

mbauman··on Figma and Adobe abandon proposed merger
Wow, thanks for the reference!
mbauman··on Figma and Adobe abandon proposed merger
I would expect that to only get paid if solely one side were responsible for blowing up the deal. This is expressly saying both have mutually agreed.
mbauman··on Barcelona Supercomputing Center presents Sargantana: new open-source RISC-V chip
There's both supply-chain and runtime security.
mbauman··on GitHub Renamed Me (2014)
It feels a little different if — as it seems to be the case — the GitHub account is used as a static site generator. Heck, the guy could've been depending upon CNAMEing to `rbishop.github.io`.
mbauman··on Email obfuscation: What still works in 2023?
So is the real story here that putting your email on the open web simply isn't a significant "vulnerability" anymore?

Honestly, why would you spider and scrape sites when you can just grab a leaked dataset?

It would be slightly more interesting to put the canaries in a more commonly scanned location, like a plain-text bio field on a popular site (like this one... or FB... or LinkedIn...).

mbauman··on You have a right to know why a health insurer denied your claim
In my experience, knowing why a health insurer denied a claim isn't useful; it just becomes a maze of medical billing codes and definitions that always ends at "insurer wins."
mbauman··on There's Math.random(), and then there's Math.random() (2015)
Oh I completely agree there's value in improving the accuracy of the cdf, especially for 32-bit float! There it can be quite a surprise that tests like `random() < x` can be off by as much as .6% for values of x on the order of 1f-5 and 6% for values on the order of 1f-6, even though they are an order of magnitude or two over `eps(1f0)`.

Following your post, I've found the following fast and straightforward and SIMD-friendly implementation that uses all 64 bits for a [0, 1) distribution:

    ```julia
    function random_float(rng)
        r = rand(rng, UInt64)
        last_bit = r & -r
        exponent = UInt64(2045)<<52 - reinterpret(UInt64, Float64(last_bit))
        exponent *= !iszero(r)
        fraction = ((r ⊻ last_bit)>>(8*sizeof(UInt64) - 52)) % UInt64
        return exponent | fraction
    end
    ```
mbauman··on There's Math.random(), and then there's Math.random() (2015)
That article makes the same unstated assumption, attempting to treat `random()`'s output (and transformations thereof) as true random variables. But it just changes the rounding mode and bin size in a manner that _happens_ to work better for some transforms like log.

It changes the fundamental property of the distribution — the cdf — without stating it.

With floating point realizations of [0, 1) intervals:

   cdf(p) = P(random() < p) := p
With (0, 1] intervals:

   cdf(p) = P(random() <= p) := p
mbauman··on There's Math.random(), and then there's Math.random() (2015)
That's not a nitpick at all — it has huge ramifications for how `Math.random()` needs to be interpreted. Many folks want it (and transformations of it!) to behave as though it were a true random variable sampled from a theoretically ideal [0, 1) uniform distribution.

But it's not, and will never be. It can only ever return values that have been rounded down to some floating point number. Even if you realize this, you still might expect it to be able to return _any_ floating point number x∈[0, 1) with `eps(x)` probability. But that's not typically the case, either. Typical implementations round down to the previous multiple of `eps(1.0)` or `eps(0.5)`.

mbauman··on Right turn on red? With pedestrian deaths rising, US cities are considering bans
That’s wildly incorrect.

https://axleaddict.com/safety/Obscure-Driving-Regulations-Ig...

mbauman··on Redfin Is Leaving the National Association of Realtors
> Otherwise whatever gains you had on paper also apply to the nearby houses, so you can't afford them.

Why are you so certain you couldn't you afford them? You were able to purchase a 200K house many years ago. Now you just need another 200K to buy that bigger house — and it's a bigger house! You did it once long ago, you can probably do it again even easier (in addition to everything else, you have a huge downpayment now). And no, you typically don't pay capital gains taxes against your primary residence.

In any case, the point is that you're in a way better place than if you hadn't made that 200K purchase. That's what I mean by "floating on the water" — you have a stake in the market so now as it moves, so does your asset.

mbauman··on Redfin Is Leaving the National Association of Realtors
Baloney. You have a house with relatively stable monthly payments. You will be able to purchase another hour if and when you need to move. You have a nest egg.

It doesn’t matter when you sell because now you’re floating on the water. As the tide rises, so do you.

mbauman··on How Google alters search queries to get at your wallet
We don’t know how many assumptions Megan made or even who was testifying and what they assumed or what evidence was presented in court. Yes, that’s problematic.
mbauman··on How Google alters search queries to get at your wallet
It is so extraordinarily valuable that folks like Megan Gray have been attending these trials. This story would be so much more compelling if we could also see that slide — but as a closed trial only those with the time and ability to actually go to the courthouse can see it.

On the same day this slide was presented, she also tweeted: https://nitter.net/megangrA/status/1704244948378194289

> I am so livid at what just happened re public access in the Google Search antitrust case right now that I don’t trust myself to tweet about it yet. Back later.

mbauman··on We're the Researchers who looked into the privacy of 25 of the top car brands
Unfortunately it can also matter if your boss knows. Or the company you’re interviewing with. It shouldn’t matter (and would be illegal if it could be proven), but it matters.
mbauman··on We're the Researchers who looked into the privacy of 25 of the top car brands
> from a practical perspective it is kinda true

Folks don't realize all the things they are hiding — and can be extraordinarily important to hide — even when they're actively hiding them! My personal favorite anecdote here is comes from a debate with a friend who was staunchly asserting that she had nothing to hide. I just blindly suggested — hypothetically — that one example would be if she were in early stages of pregnancy. Seven months later, guess what she was hiding during the argument.

mbauman··on Apple vs. Meta: The Illusion of Privacy
Does it matter? The "threads bit" isn't set until you click go. They could totally support zeroing that bit back out. Obviously you and I can imagine how that might be super difficult, but it needn't be. More to the point, not everyone would think that way. They just want an "undo".
mbauman··on eSignature Beta for Google Docs and Google Drive
It would also vary by use-case. Some regulated industries have specific standards required for an electronic signature — and hitting those standards is typically an add-on cost for systems like Docusign.
mbauman··on Maybe the problem is that Harvard exists
That was my favorite part of the article:

> We should have a Marxist version of Goodhart’s law, something like:

>> When a measure becomes a target, the privileged always find a way to win.

mbauman··on LK-99 is an online sensation but replication efforts fall short
I don't think it's too hard to imagine that other characteristics of the material could indeed get in the way of practical applications — perhaps brittle-ness or difficulty of manufacture or challenges in making it in the appropriate forms or ...
mbauman··on LK-99 is an online sensation but replication efforts fall short
> They didn't perform tests above that temperature.

Right, exactly, they're claiming that Tc is above 127°C. The claim is that it's a superconductor at (some) temperatures even higher than what was tested. It's also still a superconductor at temperatures below.

mbauman··on Rust vs. Julia in scientific computing
I think the consensus response from the Julia community is that the author is spot-on with a key point: that Rust is exceptional when you need/want static analysis. And Julia isn't great there.

This doesn't mean, however, that Julia doesn't solve the 2-language problem. There are large classes of problems and applications — especially in scientific computing — where a complete static analysis isn't required to move into production. It is definitely helpful for maintenance and security (and a requirement in some industries), and Julia's static analysis tooling is continuing to improve... especially with ongoing investments from key companies pushing us towards this goal.

Rust is a great language when you know exactly what you want to write. When you know exactly what the inputs are and what the outputs are and what the algorithm is. And it's really painful when you're not sure. Pick the right tool for the right job!

mbauman··on U.S. pedestrian deaths reach a 40-year high
I see that anecdote as a sobering reminder of the state of our roads and drivers. That sounds like inappropriate infrastructure (a 4-lane residential stroad?) combined with a person who prioritizes their ability to drive as they wish over others' lives, and who takes glee at their ability to dominate others, and a legal framework that empowers them to do so.
mbauman··on U.S. pedestrian deaths reach a 40-year high
Given that pedestrian (and especially child) visibility is so low in these gargantuan vehicles, I don't see how you can make such a definitive statement absolving their fault in causing collisions without data.
mbauman··on U.S. pedestrian deaths reach a 40-year high
E.g., The People's CDOT (Chicago Dept of Transportation): https://www.cbsnews.com/chicago/news/the-peoples-cdot-road-p...
mbauman··on U.S. pedestrian deaths reach a 40-year high
Perfect example here is that Indiana has just created a law that explicitly forbids a singular targeted city from adding _any_ new restrictions against turning right on red in _any_ intersection.

https://legiscan.com/IN/bill/HB1050/2023

mbauman··on Arrest made in SF killing of Bob Lee – alleged killer also worked in tech
Yes, let's jump straight to other conclusions now.
mbauman··on Current causes of death in children and adolescents in the United States (2022)
> I’ll update this comment when I find the source

So... you gonna update your beliefs in light of the changing landscape? We have the source; you probably heard it when it was true, but things have gotten worse. Or are you just going to seek alternative ways of supporting your pre-existing beliefs?

Everything is politics, indeed.

← PreviousPage 3 of 9Next →