> The same way you test any system - you find a sampling of test subjects, have them interact with the system and then evaluate those interactions.
That’s not strictly how I test my systems. I can release with confidence because of a litany of SWE best practices learned and borrowed from decades of my own and other people’s experiences.
> No system is guaranteed to never fail, it's all about degree of effectiveness and resilience.
It seems like the product space for services built on generative AI is diminishing by the day with respect to “effectiveness and resilience”. I was just laughing with some friends about how terrible most of the results are when using Apple’s new Genmoji feature. Apple, the company with one of the largest market caps in the world.
I can definitely use LLMs and other generative AI directly, and understand the caveats, and even get great results from them. But so far every service I’ve interacted with that was a “white label” repackaging of generative AI has been absolute dogwater.
I don’t think that this is a dupe, the ramifications of citizens being removed from Global Entry and PreCheck lists wasn’t part of the NYT article AFAIU.
Curious that at its current score and comment count it’s no longer on the front page, despite being neither flagged nor marked as dupe.
Edit: guess it’s flipping in between page 1 and 2 per refresh.
Personally, I don’t think that anyone should be speaking authoritatively on this subject because it seems to me to be untested constitutional law, unless a constitutional scholar would like to chime in.
In which case, it’s up to the Supreme Court to either explicitly (through judgment) or implicitly (through denying a hearing of the case) decide.
Ok, I am a Constitutional Scholar and there is no way that not getting to jump the line at the airport is a “material adverse action” because you can simply get in line. In general, there need to be damages when you talk about liability.
I mean usually an Actual Expert pretends to inform and educate the people he’s interacting with. I didn’t really think he came across as informative or educational.
The stated purpose for biometrics and photos with PreCheck and Global Entry is to identify you, so it’s not likely against its stated purpose to use it for identification, per-se.
Consider the information can be used for more than just identifying you... if you have sufficient quality biometrics they can be used to _impersonate_ you, including "fingering" you for things you didn't do. Police forces have "planted" evidence for decades now, biometrics can be just another thing that can be planted. The problem is, you can't fight it, because it's absolutely unique to you (with some extreme exceptions).
This is one of _many_ reasons why biometrics need to be a personal civil liberty. The individual must have the right to say "no" to _any_ "requirement" for giving up biometric data, unless they are convicted as a criminal (IMO). Because once you deliver that information, you _cannot_ trust any other party _to actually do what they say will do and destroy said data_, and that's not even considering just poor storage of said data.
Once your biometrics are in a database, you're fucked *for life* because it's completely unrealistic to have it destroyed with absolute certainty. This needs to be a *global human right*, as hard as those are to come by still.
But it's still awful. It doesn't matter at this moment that other governments may be doing this. We don't want that for us (and I don't want it for others either).
Guess who has essentially unlimited jurisdictional limits? ICE.
So they can pretend they are ‘checking for immigration status’ using the existing photos and biometrics, while simultaneously gathering information on who is at what protest.
Then the info gets shared once gathered - with or without plausible deniability - and blam. Bobs your uncle.
> Guess what the stated jurisdictional limits are for CBP? 100 miles from any possible border
To quote a prominent US historian:
In a constitutional regime, such as ours, the law applies everywhere and at all times. In a republic, such as ours, it applies to everyone. For that logic of law to be undone, the aspiring tyrant looks for openings, for cracks to pry open.
One of these is the border. The country stops at the border. And so the law stops at the border. And so for the tyrant an obvious move is to extend the border so that is everywhere, to turn the whole country as a border area, where no rules apply.
Stalin did this with border zones and deportations in the 1930s that preceded the Great Terror. Hitler did it with immigration raids in 1938 that targeted undocumented Jews and forced them across the border.
That last part isn't true. Citizens who impede ICE officers in the performance of their duties can be arrested by ICE. That is specifically written into the law, and it's a statute that can be interpreted pretty broadly.
It’s not legal to deport U.S. citizens but they have anyway. A judge in Minnesota has said that ICE has violated around 100 court orders. We are living in a personalist dictatorship. The courts are ignored when their rulings are inconvenient.
> The question you asked, as pointed out, is a non sequitor
Not what non sequitur means nor how it’s spelled. And repeating a point in the same comment doesn’t count as pointing it out previously.
To the extent there is non sequitur in this thread, it’s in jumping into a legal discussion halfway to argue the law doesn’t actually matter because you feel like it.
Ah. My bad spelling. That is a great, pertinent thing to point out. I did abuse the meaning of non sequitor. I was trying to convey a sense that is lost on you without writing a treatise. The law doesn’t matter because we are living in a personaist dictatorship. Asking for the policy or legal basis of ICE’s actions is pointless and ignores the reality that ICE doesn’t care about this and no authority in the country is willing and/or able to stop their abuses.
Fair. I’ve seen this site posted on HN before, mostly in the context of business travel and tech in the airline industry, so I didn’t consider it too low quality.
The rest of the Republican Party is completely devoid of charisma, especially the kind that drew so many voters to Trump. There is no drop-in replacement.
Lots of money will be spent trying to manufacture a replacement, though. That will be fun to watch. If you thought the last-minute rally around Kamala was tough to watch…
Isn’t the VP generally the shoe-in nominee? Vance lacks charisma and gravitas, but he only has to be better than the Democratic candidate. For every Bill and Barack, the Democrats have also given us a Kamala, Hillary, and Al. Never underestimate their ability to pick a loser.
But what the republican party has, is a lot of isolationist voters who cannot be moved by appeals to markets or international trade. They don’t care about that stuff.
Sure, the republicans will look hilarious trying to replace Trump for a while … but those Americans aren’t going anywhere and will gladly vote for the next Trump whenever they show up, same as they voted for Reagan and Bush II.
The American attitude driving this current period is much deeper and wider than one man, and people thinking it will all go away when one old man steps down are going to be “surprised” when we’re dealing with this again in ten years or twenty years or three years.
Don’t get me wrong, I’ll be the first to jump up and say there’s a deep cultural rot in America that, if it weren’t for the fortune of incredible financial success, would have us be seen as a hellhole of antisocial maniacs.
That being said, I just don’t buy into the notion that the strategy of the party from 2016-2024 (maybe 100 Trump rallies per year?) can carry over into the late 2020s / early 2030s.
If anything, this is me saying everyone is aware that the current window for reactionary politics in America is closing as Trump loses his vigor and gets closer to being too old to do what he did between 2014 and 2024. The reactionaries in the government and behind the scenes may make one last desperate grab at maintaining power.
That's not the point: the point is America did this twice. The world is not going to deal with America radically flip flopping every policy position every 4 years, and escalating that every time.
The US has just finished (maybe?) threatening to invade a NATO allied country. The occurrence rate of that has gone from "never" to "at least once". The delta change on that is infinity: there will never be a world in several generations where that is not a strategic risk the world has to deal with every 4 years.
> That's not the point: the point is America did this twice. The world is not going to deal with America radically flip flopping every policy position every 4 years, and escalating that every time.
I’ll admit, I’m becoming confused about the point of our back-and-forth.
All I’m trying to express is that probably by the end of 2026, and definitely by 2028, the people who are trying to enact reactionary change (Stephen Miller, PayPal Mafia, Heritage Foundation, etc.) will have to adjust their strategy. They are losing their charismatic leader, if not because of constitutional limits on presidential terms, then by his very obvious reduced vigor (he will not be able to do 100 rallies in a calendar year again).
On the world stage, yes, America has stumbled. Maybe even worse, some international folks are realizing that the America that they thought existed was just a Hollywood mirage, and that we were always one recession and a few thousand votes in Florida from becoming a global pariah.
There's no law of nature that proves democracy can't be overthrown, but ICE is currently struggling against popular resistance to merely enforce immigration law in a single medium-sized city. Right now there's not much indication they have either the desire or operational capacity to pull off nationwide voter suppression. (And a number of special elections over the past year have defeated regime-backed candidates without ICE involvement.)
If they tried to merely immigration law, they would had no issue.
They were successful at creating fear and making conflict avoidant people stay home. Despite large amount of protesters, that is real effect.
And it would create real vote suppression. You would had people not voting out of fear. And the opposing side having less votes (even if they win, they win less)
You, I, and regime officials know that they're trying to do more than merely enforce immigration law. But that's not the story they're telling the staff. Your average immigration enforcement officer still believes they're doing nothing more than executing warrants on people who are unlawfully present in the country - indeed, even with all the chaos and murders, there may not even be a majority of officers who've personally performed any misconduct. I'm not saying they're "good apples", but neither will they necessarily go along with an election subversion scheme that has no plausible fig leaf.
Equally importantly, this is not an operation that could be planned in secret or at the drop of the hat. If the Trump regime tries to do something like this, we'll know weeks or months in advance, and there will be plenty of time to make it clear that anyone who participates will face swift and severe justice.
Definitely an odd duck out in the context of Trump, Obama, Bush II, Clinton, Bush, Reagan.
Also remember it’s not just being charismatic, but charismatic enough to keep people distracted from increasingly unpopular reactionary politics that defy even conservative beliefs (e.g. gun control, speech policing, deficit spending, plenary executive).
Gun control (for minorities), speech policing (for liberals), deficit spending (when they are in charge), and a plenary executive (when Obama isn't president) are core conservative beliefs.
They say that they don't like those things, but you can't listen to what politicians and talking heads on TV say. Politicians and talking heads lie all the fucking time. You have to look at what people do.
They aren't stupid in not understanding the hypocrisy.
We are... for thinking that they don't know that they are hypocrites.
This article is different because it actually talks about code review, which I don’t see very often. Especially in the ultra-hype 1000x “we built an operating system in a day using agentic coding”, it seems like code review is implied to be a thing of the past.
As long as code continues to need to be reviewed to (mostly) maintain a chain of liability, I don’t see SWE going anywhere like both hypebros and doomers seem to be intent on posting about.
Code review will simply become the bottle neck for SWE. In other words, reading and understanding code so that when SHTF the shareholders know who to hold accountable.
1. You know exactly what the code should look like ahead of time. Agents are handy here, but this is rare and takes no time at all as is
2. You don’t know exactly what it should be. Building the mental model by hand is faster than working backwards from finished work. Agents help with speeding up your knowledge here, but having them build the model is not good
Code review is and always was the bottle neck. You are doing code review when you write code.
This is one of the key elements that will shift the SWE discipline. We will move to more explicitly editing systems by their footprint and guarantees, and verifying that what exists matches those.
“Editing systems” will remain being a rigid and formal process, though. The discipline won’t become softer and more easily understood (SWEs are going to be replaced!), if anything it will become harder and less easily understood.
Shareholders don’t want a “release” button that when they press it there’s a XX% chance that the entire software stack collapses because an increasingly complex piece of software is understood by no one but a black box called “AI” that can’t be fired or otherwise held accountable.
It’s definitely a result of the money at play, which is unprecedented in scale and (imo) speculation.
But this is, in theory, why we have laws: to fight power imbalances, and money is of course power.
Tough for me to be optimistic about law and order right now though, especially when it comes to the president’s biggest donors and the vice president’s handlers.
Ah my bad. But also, if we’re comparing buildout of infrastructure to the construction of the American Railroad system, especially in the context of lawbreaking and general immoral and unethical behavior…
Point kind of proven, yeah? One more argument for the “return to the gilded age” debates.
Edit: you’re speaking kind of authoritatively on the subject though. Care to share some figures? The AI bubble is definitely measured in trillions in 2026 USD. Was the railroad buildout trillions of dollars?
As a percentage of GDP investments in the railroad buildout in the US was comparable or slightly higher than AI-related investments. But they are on the same order of magnitude, which says a lot about the scale of AI.
> AI infrastructure has risen by $400 billion since 2022. A notable chunk of this spending has been focused on information processing equipment, which spiked at a 39% annualized rate in the first half of 2025. Harvard economist Jason Furman commented that investment in information processing equipment & software is equivalent to only 4% of US GDP, but was responsible for 92% of GDP growth in the first half of 2025. If you exclude these categories, the US economy grew at only a 0.1% annual rate in the first half.
Land value underneath railroad tracks is an interesting subject. Most land value is reasonably calculated by width * length, and maybe some airspace rights. And that makes sense to our human brains, because we can look at a parcel of land and acknowledge it might be worth $10^x for some x given inflation.
But railroads kind of fail with this because you might have a landowner who prices the edge of their parcel at $1,000,000,000,000 because they know you need that exact piece of land for your railroad, and if the railroad is super long you might run into 10 of these maniacs.
Meanwhile the vast majority of your line might be worth less than any adjacent farmland, square foot by square foot, especially if it’s rocky or unstable etc.
Having a continuous line of land for many miles also has its own intrinsic value, much more than owning any particular segment (especially as it allows you to build a railroad hah).
Anyway, suffice to say, I don’t think “land value underneath railroads from the 18th century” is something that’s easily estimated.
That’s not strictly how I test my systems. I can release with confidence because of a litany of SWE best practices learned and borrowed from decades of my own and other people’s experiences.
> No system is guaranteed to never fail, it's all about degree of effectiveness and resilience.
It seems like the product space for services built on generative AI is diminishing by the day with respect to “effectiveness and resilience”. I was just laughing with some friends about how terrible most of the results are when using Apple’s new Genmoji feature. Apple, the company with one of the largest market caps in the world.
I can definitely use LLMs and other generative AI directly, and understand the caveats, and even get great results from them. But so far every service I’ve interacted with that was a “white label” repackaging of generative AI has been absolute dogwater.
reply