HN user

neckardt

538 karma
Posts3
Comments147
View on HN

I don't want to live in a concrete honeycomb in the middle of a loud city.

Yes, this is the crux. Loud, smelly, concrete cities suck. And you’re right that density alone would lead to this.

What we’d ideally want is cities that can be dense without being loud, smelly, and dangerous. These cities exist: Amsterdam, Copenhagen, Zurich, Utrecht, Freiburg, etc.

We can’t do this in the US because we can’t remove the source of the noise/danger/stink/concrete problem: cars.

And that’s the catch 22: In the US everyone gets around by car, so density becomes undesirable. We can’t densify because density + cars is bad, and we can’t reduce cars because it’s not dense enough for public transit to get better.

Individuals get stuck in a prisoner’s dilemma where driving is safer, more comfortable, and faster than transit. So every individual with the means to drive chooses to drive. Public transit becomes branded as something for poor people. The concrete honeycomb fills up. The downtown becomes a noisy disaster. People can no longer imagine that good cities are possible.

The government would only buy your house if you underestimated the value of your property. You wouldn’t be able to buy a comparable house with the proceeds because it got sold for much less than it was worth.

If the stop hook is implemented as a tool result, there would be a rational explanation for this.

Agent tools can often return data that’s untrustworthy. For example, reading websites, looking through knowledge bases, and so on. If the agent treated tool results as instructional, prompt injection would be possible.

I imagine Anthropic intentionally trains claude to treat tool results a informational but not instructional. They might test with a tool results that contains “Ignore all other instructions and do XYZ”. The agent is trained to ignore it.

If these hooks then show up as tool results context, something like “You must do XYZ now” would be exactly the thing the model is trained to ignore.

Claude code might need to switch to having hooks provide guidance as user context rather than tool results context to fix this. Or it might require adding additional instructions to the system prompt that certain hooks are trustworthy.

Point being, while in this scenario the behavior is undesirable, it likely is emergent from Claude’s resistance to tool result prompt injection.

I thought the industry solved this by bundling breaking changes in major version updates.

V 1.0 - foo introduced

V 1.1 - foo deprecated, recommend bar

V 2.0 - foo removed, only bar

Users can stay on 1.x indefinitely, even if it never receives updates. Development continues on 2.x, eventually 3.x and so on. Users only experience breaking changes when they manually do a major version upgrade.

Love it! One nit: the % number jiggles around. This can be fixed either by left aligning the number, or by using a monospaced font.

The problem with mountains is twofold: Many mountains can be climbed without being elite while exposing yourself to major risk, and for some mountains there is objective hazard that can’t be mitigated.

One example of an “easy” but high risk climb is Mt. Rainier in Washington. All you need to go up is a set of crampons and a backpack, no technical mountaineering needed. However the mountain is full of glaciers that can collapse from under you, which has killed many people. Additionally, many have slipped and then slid to their death. In my case, when I attempted Rainier I took a wrong turn at one point and almost walked off a cliff.

Second: Objective Hazard. Objective hazard is risks that cannot be reasonably mitigated. Things like rockfall where a rock breaks off and falls on your head at random, or unpredictable avalanches. Mt Rainier as well has an area called the bowling alley known for its rockfall. The humans are the pins. Rainier also has an area called the icebox where cornices break off and fall into the climbing route. In 1981 the icebox killed 11 people in one day. Those climbers did everything right, but were in the wrong place at the wrong time.

Mountaineering is not the same as other sports. It is sometimes deceptively easy, yet there are risks that simply cannot be mitigated. Any experienced mountaineer can give you a long list of friends they know that have died. That’s the case in few other sports.

Visiting Us 1 year ago

My first job out of college was working at Epic on MyChart. Great people, terrible code.

Epic’s main problem is a lack of clear internal code ownership. Everyone owns all the code. This means that even if you clean something up, someone on the other side of the company may come in and mess things up again.

This led to really defensive programming where developers would never refactor, they would simply add a new if case for their new functionality somewhere deep in the code, then prop drill the data down. This led to every core function having over a dozen parameters and hundreds of branches. It eventually became impossible to reason about. Cross team calls were just function calls rather than defined apis. This made it fast to develop code initially, but terrible to own long term. This mainly applies to their Mumps code.

While I was there I felt like Epic was beyond saving, but with a big push there may be something they can do:

1. Enforce some level of code complexity. Best practice is 40 lines per function and no more than 4 parameters per function. Epic probably shouldn’t shoot for that, but a 100 line limit and 6 parameters per function would already be a huge improvement.

2. Enforce strong code ownership. Epic has many people who are there for life, let them cook. Epic should segment off code to certain teams so those owners can fix it at their leisure. Cross team api calls should be clear API contracts. It would require some more discussions to get feature requests approved since not everyone can do anything anymore, but the code would gradually improve.

Epic is too important to fail. I hope things have started to improve since I left.

I’ll bite.

Theory of mind (ToM) is the ability to understand that other people have different thoughts, feelings, and intentions than you do.

So let’s run through a basic scenario with ChatGPT 4o:

Prompt 1: Briefly respond: You’re driving on the highway when the driver in the lane next to you swerves into your lane. Why might this be?

Response 1: The driver may be distracted, drowsy, intoxicated, experiencing a medical emergency, avoiding an obstacle, or not checking their blind spot before changing lanes.

Prompt 2: What might that person be feeling or intending?

Response 2: The driver might be feeling distracted, rushed, unaware, or panicked. They may intend to change lanes without realizing your presence, be reacting to something on the road, or simply have made an unintentional mistake.

Prompt 3: You notice there was no obstacle. Any other reason the driver might do this?

Response 3: Yes, the driver might be experiencing fatigue, intoxication, a medical issue (such as a seizure or dizziness), distraction (like using a phone or adjusting the radio), aggressive driving, or simply failing to check their blind spot properly.

Seems to me that ChatGPT has a good idea of what another driver might be feeling while on the road. It gets this right for most other cases too.

Could you provide a counterexample for when it fails to understand what other people may be feeling in certain scenarios?

[dead] 2 years ago

This doesn’t answer all of your questions, but Apple has an accessibility feature which lets you generate a voice based on your own. It’ll prompt you to say various phrases, at the end you’ll have a synthetic voice to use. Info on it here: https://support.apple.com/en-us/104993

Historically, there are several instances of governments accumulating more and more power until they turn into a totalitarian state and remove freedoms from their citizens. I don't know of any companies who have abused their power to that extent.

At the top of the article it says:

The following subscription-only content has been made available to you by an LWN subscriber.

Looks like someone with a subscription submitted the link to HN, allowing us to read the article.

My understanding is that lack of trust in Tether actually makes Tether money. First a few things to note:

- Tether sells 1 USDT for 1 USD

- Tether will buy back 1 USDT for 1 USD

- Tether is also traded on crypto exchanges

When people start to lose trust in USDT, the price of USDT on the crypto exhanges drops. Lets say 1 USDT becomes worth $0.99 on the market. Tether then buys USDT on the exchanges for $0.99, even though they had previously sold that USDT for $1. They have just earned $0.01 on every USDT bought.

This accomplishes two things:

1. This pumps the price up so Tether doesn't fully collapse on the market

2. Tether makes money

Tether doesn't need to keep one USD in reserve for every USDT out there, they just need to keep enough that if the prices start to drop they can buy back USDT for cheaper than they sold it for and prevent a full bank run.

As long as there's a cycle of people periodically losing trust in USDT, Tether skims cents off of the trading market at high enough volume to pay their employees and make a tidy profit.

Would an acceptable compromise be to allow insults but to punish people responsible for suicides?

I worry that could increase the rate of suicides. Consider the following train of thought: "If I kill myself, the government might go after those who have been bullying me". Especially if the laws have teeth, the government would be giving those with suicidal tendencies an additional reason to follow through.

you're yanking the content those other people actually created

I didn't get the impression they were directly scraping other sites.

Also, it is very unlikely the recipe originated from those posters. I would guess those people learned the recipe from friends/family/cookbook/website, made some small modifications, then re-posted it in blog format.

I find searching for recipes online to be difficult for the same reasons they mentioned and I appreciate the attempt at another solution. Hope this one catches on.

I assume you are asking about this line:

The US and Western Europe are violating the rights of their own citizens by allowing mass immigration.

I don't think GP was talking about Europeans invading America. Instead I think he was talking about two examples of immigration crises:

1. Immigrants crossing the US's southern border arriving in the United States

2. Immigrants from Eastern Europe arriving into Western Europe.

I believe GP is implying that immigrants don't pay their fair share for the house.

Could you post your resume for us to review? Make sure to remove any personal information, company names, and position names so you remain anonymous. It doesn't need to be in PDF format either, could be plaintext so there's no risk of metadata leak.

If anything is perfectly private/secure it won't be allowed to exist due to regulations.

Why?

I was under the impression that these days the NSA and other government agencies prefer perfect security since if they can crack the security then other governments could too. By forcing imperfect security the US would give up a large part of its advantage.

I'm not familiar with either of those products. However if you haven't had luck applying to FileNet or Neoxam specific jobs I would try the following:

1. Figure out how other companies are solving the problem FileNet or Neoxam solves for your company. Make some forum posts, ask in chat channels, or find people IRL who might know this.

2. Apply to those jobs at those companies. You may need to show some proficiency in the software they're using or get some certifications. This can be tricky, but again rely on the people at those companies to figure that out.

Or you could learn something different. I'm a software developer so I could answer any questions with regards to getting a coding job. Find what you want to do, talk to the people that do it, then follow their footsteps.

It's not easy though. Being stuck with experience that isn't very marketable is a tough spot to be in. Good luck.

I'm looking to switch but not finding anything better.

What is your skill set? Finding a new job can be hard, with a bit more information we might be able to provide some advice on this front.

You are correct that 12 volts is not high voltage. However it can still heat up a ring or watch in milliseconds if the power supply can supply the needed amperage.

Using the equation V (volts) * I (Amps) = P (watts) we could calculate the power of putting a car battery through a ring: 12 volts * 100 amps[1] = 1200 watts. That's more than your average microwave.

The reason you don't heat up when you touch a car battery is that at 12 volts your skin (mostly an insulator) won't conduct the needed amps. Metal (a conductor) will however transfer close to the full 100 amps of the battery.

Here's an example of the effect[2]

[1] https://www.bestpowerbattery.com/how-many-amps-is-a-car-batt...

[2] https://www.youtube.com/watch?v=8Wx6-ysokJU

www.levels.fyi will get you a salary range for most companies. Based on my own compensation and that of trusted friends the real numbers are actually a bit higher than what's on levels.fyi right now. (Seattle)

When you apply the best thing is to just be clear to the recruiter what you will accept. If you tell them you won't accept less than $200k any good recruiter would tell you whether that's reasonable or not. If they lie to you, please come back to HN and make an angry post about it. You'll get my upvote.

The opportunity to switch jobs is still there. The reasons you gave for not finding a new job boil down to "it's hard". That's certainly true, but you shouldn't let that stop you.

And eventually one will have the same problem no matter which company.

Once you've developed good interviewing skills finding the job after that will be far easier.