HNHacker News
TopNewBestAskShowJobs

lhnz

5,754 karma · joined June 16, 2010

www: https://sebinsua.com

github: https://github.com/sebinsua

gist: https://gist.github.com/sebinsua

twitter: https://twitter.com/sebinsua

[ my public key: https://keybase.io/sebinsua; my proof: https://keybase.io/sebinsua/sigs/h5SINjJKRvoKp0ggBhJuVn61d2Zfb3sbIp7ChK3OlTY ]

submissionscomments
lhnz··on Google User Data Is Police's Top Shortcut for Solving Crimes
Nobody other than yourself is arguing that reasonable doubt is being generally ignored by juries. You either misunderstood the Kalven-Zeisel study that you brought to the argument or you failed to read it.

The study showed that juries had a higher threshold for conviction than judges, which is a direct repudiation of your point that "reasonable doubt is generally ignored by juries. Interviews of post conviction juries show they often convict people who they have serious doubts actually committed the crime." It also explained that similar results to this had been reproduced many times over.

It's not even conclusive that it is purely due to juries having a different interpretation of reasonable doubt as it is explained that "Much more research is needed to map experimentally the differences and similarities between the judgments of judges and juries before concluding that judges are better than juries at specific tasks (e.g., assessing risk) or that deliberations enable juries to outperform judges on other tasks (e.g., assessing conflicting testimony)."

To be clear, your claim that juries are more likely to convict was disproven by your own data and your claim that they ignore "reasonable doubt" is unlikely given that they err on the side of reasonable doubt. The later article you posted doesn't really have anything to say about either of these things, but instead presents better ways to articulate the idea of reasonable doubt to a jury.

I don't find your latest argument about dice rolling persuasive either as (1) it's not an empirical argument, and (2) this oversimplifies the complex dynamics of legal decision-making and doesn't effectively challenge the implications drawn from the Kalven-Zeisel study -- unlike the randomness of dice, legal decisions are derived from a deliberative process grounded in law, evidence, and rational argumentation.

lhnz··on Google User Data Is Police's Top Shortcut for Solving Crimes
The 78% agreement rate between judges and juries suggests a shared interpretation of evidence, which indirectly indicates both are following the same 'reasonable doubt' standard.

The 3% disagreement where judges would have convicted doesn't prove juries 'ignore' reasonable doubt; it reflects normal variance in human judgment.

Legal decisions aren't coin tosses—such a high agreement is unlikely if both parties weren't generally applying the same legal standard.

lhnz··on Google User Data Is Police's Top Shortcut for Solving Crimes
The Kalven-Zeisel study indicates that juries and judges agree on verdicts 78% of the time, suggesting they have similar interpretations of the evidence. When they do disagree, juries are more lenient 19% of the time, indicating they don't 'ignore' the standard of reasonable doubt but may actually err on the side of it. Your supplementary article doesn't prove that juries generally misunderstand 'reasonable doubt' either. So, neither of your sources substantiate your claim that juries 'generally ignore' the standard of reasonable doubt and instead choose to convict when they have "serious doubts".
lhnz··on Google User Data Is Police's Top Shortcut for Solving Crimes

  > Yet you made no reference to it in your poorly executed attack.
I am not attacking you and I did make direct reference to this.

You wrote:

  > In practice reasonable doubt is generally ignored by juries.
  > Interviews of post conviction juries show they often convict
  > people who they have serious doubts actually committed the crime.
  > They say stuff like, “If he’s innocent it will be reversed on appeal.”
I wrote:

  > While the system is not perfect and errors do occur, the claim
  > that juries "generally" ignore the standard of reasonable doubt
  > is a sweeping statement.
You then wrote:

  > It’s a sweeping statement backed up by actual evidence.
I then wrote:

  > Actually your post containing the sweeping statement doesn't 
  > contain any evidence.
It's correct that you didn't support the claim.

Your own evidence didn't substantiate your claim, it directly refuted it.

You posted a Kalven-Zeisel study comparing the decision-making processes and outcomes of judges and juries in trials which concluded that judges and juries generally agree on verdicts in a high percentage of cases (78%). In fact, it is stated that "when they disagreed, the judge would have convicted when the jury acquitted in 19% of the cases, and the jury convicted when the judge would have acquitted in 3% of the cases—a net leniency rate of 16%". The study that you've given me shows that when the jury and judge disagree, the jury is more lenient than the judge. This directly refutes your subsequent claim that "Interviews of post conviction juries show they often convict people who they have serious doubts actually committed the crime. They say stuff like, “If he’s innocent it will be reversed on appeal”" and indicates that the jury err on the side of reasonable doubt rather than ignore it.

The article you have subsequently posted is about the interpretations and articulation of what "reasonable doubt" means to juries. It does not substantiate any claims about "reasonable doubt" being generally ignored by juries or make any claims about them being more likely to convict. It's conclusions are about how "reasonable doubt" might be better communicated or understood.

lhnz··on Google User Data Is Police's Top Shortcut for Solving Crimes
The claim you made which I referred to as a sweeping statement was that juries "generally" ignore the standard of reasonable doubt.

This study you cite is not relevant to that claim.

lhnz··on Google User Data Is Police's Top Shortcut for Solving Crimes
Actually your post containing the sweeping statement doesn't contain any evidence.

You haven't substantiated your claim.

lhnz··on Google User Data Is Police's Top Shortcut for Solving Crimes
While the system is not perfect and errors do occur, the claim that juries "generally" ignore the standard of reasonable doubt is a sweeping statement. The principle of "innocent until proven guilty beyond a reasonable doubt" is a cornerstone of many legal systems and taught and emphasized to jurors in instructions before they deliberate.
lhnz··on Google User Data Is Police's Top Shortcut for Solving Crimes
Even if the phone was completely stationary and you made no recorded interactions/transactions at all with it, it wouldn't get "thrown out as an alibi". This evidence would get considered with regards to other corroborating evidence that you provide.

Obviously, if you picked up the phone and chatted to your wife, or used it to buy something at the local store, it validates your location data even further.

lhnz··on Google User Data Is Police's Top Shortcut for Solving Crimes
The evidence would have to be looked at in regards to the totality of the evidence. They couldn't just say you handed the phone to somebody else without knowing this and showing evidence that this was the case. Assertions without evidence are generally not persuasive, and in criminal cases they must prove that you committed the crime beyond "reasonable doubt". The burden of proof is on the prosecution.

And, if there was corroboration with the digital trail of the mobile device, and you appearing on a CCTV at a location matching the GPS at some point or chatting to somebody on the phone, it'd possibly further break the insinuation that you didn't have the device during this time period.

lhnz··on Google User Data Is Police's Top Shortcut for Solving Crimes
This isn't true. Contemporaneous evidence may often be admissable in court, subject to other legal tests and challenges regarding its relevance, reliability and integrity.

Any evidence that you have that can corroborate your story will be useful to you, hence digital trails are defensive tools to innocent people, even though appearing on CCTV or near a crime could be sufficient for further investigation including questioning, detainment, or other forms of inquiry.

lhnz··on Ask HN: Who wants to be hired? (October 2023)
SEEKING WORK | LONDON, UK / REMOTE WORLDWIDE

Take a read of my blog or GitHub if you wish to quickly gauge my skillset: https://sebinsua.com

I'm a full-stack product engineer. I usually program in TypeScript, Python and Rust but have hundreds of hours of experience using a wide range of languages and technologies. I am an expert within the Node.js and React ecosystems, where I have 12 and 9 years of experience respectively.

  Location: London, UK

  Remote: Yes

  Willing to relocate: No

  Technologies: TypeScript, JavaScript, Python, Node.js, Next.js, Rust,
                SQL, PostgreSQL, React, HTML, CSS, Docker, Git,
                Kubernetes, Linux, Jupyter, FastAPI, Pandas, NumPy,
                Machine Learning, Large Language Models (LLMs), etc.

  Résumé/CV: https://sebinsua.com/resume.pdf

  Email: me@sebinsua.com

  GitHub: https://github.com/sebinsua

  Gists: https://gist.github.com/sebinsua

  LinkedIn: https://linkedin.com/in/sebinsua/
lhnz··on Dictionary of Algorithms and Data Structures
I wrote something sort of adjacent to what you're talking about. It wasn't about applications persay but instead it was a guide / decision tree that took everything I learnt on how to make choices on what datastructure or algorithmic approaches applied to different problems (from doing the Blind 75 problem set).

It's not authoritative as I am not an expert yet but might still be of interest: https://sebinsua.com/algorithmic-bathwater#what-kind-of-prob...

lhnz··on Ways to capture changes in Postgres
I once consulted at a company with a very large monolithic SQL Server. It actually wasn't Postgres but let's pretend it was.

It had been around for decades and over time it had ended up being used for all sorts of things within the company. In fact, it was more or less true that every application and business process within the whole company stored its data within this database.

A key issue we had was that because this database had many different applications that queried it, and there were a huge number of processes and procedures that inserted or updated data within it, sometimes queries would break due to upstream insert/update processes being amended or new ones added that broke application-level invariants -- or when a normal process operated differently when there was bad data.

It was very difficult to work out what had happened because often everything that you looked at was written a decade before you and the employees had long since left the company.

Would it be possible to capture changes from a Postgres database in some kind of DAG in order that you could find out things like:

- What processes are inserting, updating or deleting data and historically how are they behaving? For example, do they operate differently ever?

- How are different applications' querying this data? Are there any statistics about their queries which are generally true? Historically how are these statistics changing?

I don't know if there is prior art here, or what kind of approach might allow a tool like this to be made?

(I've thought of making something like this before but I think this is an area in which you'd want to be a core Postgres engineer to make good choices.)

lhnz··on v0: Generate UI with simple text prompts
A lot of intellectual/software work is a bit like this though.

It'll be interesting to see whether these AI tools scale into more complex work, as right now what you tend to see is a "good start point" but eventually the context of what you're building gets too large and the AI starts to produce junk.

lhnz··on Death by a Thousand Microservices

  > First, a whole army of developers writing Javascript for the browser
  > started self-identifying as “full-stack”, diving into server development
  > and asynchronous code.
This is gatekeeping--we're all ultimately self-identifying as software engineers. But more importantly it's not correct. Many of the original engineers working with Node had non-UI backgrounds. Node.js made asynchronous programming easy and promised that you could use the same language on the back-end and front-end, which seemed like good bang-for-buck for people that wanted to focus on products.

Also, not to point the blame somewhere else, but wasn't it Golang engineers that went really hard on the microservices koolaid?

lhnz··on Bun v1.0.0
Maybe the trade-offs are better?

https://twitter.com/threepointone/status/1698271991927648643

lhnz··on Bun v1.0.0

  > The transition from CommonJS to ES modules has been slow and full of terrors. 
  > Bun supports both module systems, all the time. No need to worry about file
  > extensions, .js vs .cjs vs .mjs, or including "type": "module" in your 
  > package.json.
  >
  > You can even use import and require(), in the same file. It just works.
This is the highlight for me. The Node.js ecosystem is more-or-less completely broken otherwise. This might save it.

(I think that the most impressive thing about Bun isn't its performance but the pragmatic, developer-friendly choices that have been consistently made by Jarred.)

lhnz··on Ask HN: Who is hiring? (September 2023)
Upvoting because of the WebGL header text thing you did. Very nice!
lhnz··on Ask HN: Who wants to be hired? (September 2023)
SEEKING WORK | LONDON, UK / REMOTE WORLDWIDE

Take a read of my blog or GitHub if you wish to quickly gauge my skillset: https://sebinsua.com

I'm a full-stack product engineer. I usually program in TypeScript, Python and Rust but have hundreds of hours of experience using a wide range of languages and technologies. I am an expert within the Node.js and React ecosystems, where I have 12 and 9 years of experience respectively.

  Technologies: TypeScript, JavaScript, Python, Node.js, Next.js, Rust,
                SQL, PostgreSQL, React, HTML, CSS, Docker, Git,
                Kubernetes, Linux, Jupyter, FastAPI, Pandas, NumPy,
                Machine Learning, Large Language Models (LLMs), etc.
Résumé/CV: https://sebinsua.com/resume.pdf

GitHub: https://github.com/sebinsua

Gists: https://gist.github.com/sebinsua

LinkedIn: https://linkedin.com/in/sebinsua/

Email: me@sebinsua.com

lhnz··on Scrum is a cancer
Yes, sorry, that was sarcasm. It's incredibly funny that Scrum's most redeeming feature is for it to be so configurable. Scrum has the configurability of SAP.
lhnz··on OpenAI predicted to generate over $1B
Why wouldn't corporations and users just stick to the original AI chat model that gave them the best results the first time? I think that people are lazy about switching because it's very difficult to personally validate whether something is better or worse quickly/robustly, and when you do get bad responses it's really annoying and makes you churn back to the original model that you used.

I do sometimes wonder whether they could be replaced with a free model running on a user's local machine, however, on the other hand we might always wish to pay for best-in-class (according to some evaluation criteria, that we eventually give up on trying to self-evaluate).

lhnz··on Scrum is a cancer
The good thing about Scrum is that there are hundreds of little dials that can be modified if you're not getting any value from it.

Due to this level of flexibility, it's always the right approach.

lhnz··on Transactions in a Microservice World
I agree with you, but, I once responded along these lines in an interview and it's not what people want to hear.
lhnz··on Ask HN: Who is hiring? (August 2023)
Unfortunately I do live in London and have a family so do require a London salary, but I would like to say that "Companies House" desperately needs to be improved so this is a valuable job.

Specifically it needs to be analysed from the perspective of fraudulent activities by UK businesses or foreign businesses that operate in the UK. People are able to register businesses using fake names, split their identity by changing their registered address, close their company before they need to post accounts, set the registered address of their company to a residential address they do not have access to, etc.

Obviously some of these things need to be altered at the policy-level, but if you were just to allow the weirdness to be analysed and bring it the right people's attention (e.g. https://twitter.com/DanNeidle/status/1679472841106882567), you would be doing the UK a big favour.

lhnz··on Ask HN: Who is hiring? (August 2023)
Absolutely amazing startup idea. Please open up your remote opportunities to other countries! :)
lhnz··on Ask HN: Who is hiring? (August 2023)
I know it's awkward in some ways, but would you consider remote workers? I'm from London (in the UK) but really like your product.
lhnz··on Ask HN: Who wants to be hired? (August 2023)
SEEKING WORK | LONDON, UK / REMOTE WORLDWIDE

Take a read of my blog or GitHub if you wish to quickly gauge my skillset: https://sebinsua.com

I'm a full-stack product engineer. I usually program in TypeScript, Python and Rust but have hundreds of hours of experience using a wide range of languages and technologies. I am an expert within the Node.js and React ecosystems, where I have 12 and 9 years of experience respectively.

  Technologies: TypeScript, JavaScript, Python, Node.js, Next.js, Rust,
                SQL, PostgreSQL, React, HTML, CSS, Docker, Git,
                Kubernetes, Linux, Jupyter, FastAPI, Pandas, NumPy,
                Machine Learning, Large Language Models (LLMs), etc.
Résumé/CV: https://sebinsua.com/resume.pdf

GitHub: https://github.com/sebinsua

Gists: https://gist.github.com/sebinsua

LinkedIn: https://linkedin.com/in/sebinsua/

Email: me@sebinsua.com

lhnz··on In the LLM space, "open source" is being used to mean "downloadable weights"
It tells you that they think their moat is data quality/quantity.
lhnz··on Privatisation has been a costly failure in Britain

  > In truth, what needs to happen is the government needs to realise that
  > what limits our wealth is our resource constraints and enact policies 
  > on that basis, instead of fussing about artificial financial constraints.
Indeed, the government must understand that wealth hinges on available resources, not financial limits. Policies should reflect this.

Yet, the UK seems hesitant to utilise natural resources, develop housing/infrastructure, or encourage investment in capital or innovation.

We've been rich and industrialised for so long we have forgotten that wealth isn’t created out of nothing. Our attitude towards fossil fuels is particularly indicative of this. Coal made this country an industrial superpower. We forgot that the key to our economic power started and will always start with energy -- and that this is often dirty.

Today we a country of financial services and coffee shops, because we believe we are above the truly productive sectors of the economy.

lhnz··on Privatisation has been a costly failure in Britain
It's true but also older people are addicted to destroying the economy here, constantly demanding decisions that impair the ability of the country to build houses, build businesses, build new infrastructure, etc, all while expecting high living standards. The only way it could be financed was by selling off the country's wealth.
← PreviousPage 3 of 34Next →