Hacker News Reader

mgon Sam Altman on X: "we are pausing new ChatGPT Plus sign-ups for a bit"

I've been watching the interest in ChatGPT and its rivals via Google Trends for a while:

https://twitter.com/marekgibney/status/1724712305894600929

It's fascinating that ChatGPT seems to stand a chance to even surpass Reddit in interest.

I wonder what the recent spike for "Bing" means? Is that from their image creator? Or did the chat on Bing take off?

mgon 37, the median value for the second prime factor of an integer

Then let's see who manages to write the least interesting comment in this thread.

mgon 37, the median value for the second prime factor of an integer

Well, that is about the question which is the most uninteresting number. And the paradox that - whichever number it is - this attribute makes it interesting.

But since we are not looking for the most uninteresting number, but the most interesting one, we do not have to fight with issues of that calibre.

mgon 37, the median value for the second prime factor of an integer

I agree that this makes 37 somewhat interesting.

Certainly more interesting than - say - 31.

31 is a prime number too, and therefore somewhat interesting. But for sure not as interesting as 37, which as we just learned, is the median value for the second prime factor of an integer.

Any suggestions of integers which are even more interesting?

And while we are at it, is there an integer which qualifies to be the most interesting?

mgon Websites That Changed the World (2006)

A friend and I had a discussion today, whether Amazon was a small internet company among many, back when it IPOed in 1997. Or if it already stood out as one of the biggest IPOs of that time.

To our surprise, we did not find a list of dotcom IPOs from the 90s together with their IPO market cap.

If someone here knows about such a list, I would love to see it.

mgon GPTs: Custom versions of ChatGPT

You don't need to use an open source LLM for the approach I described. You can still send the prompts to OpenAI's GPT-4 or any other LLM which is available as a service.

mgon GPTs: Custom versions of ChatGPT

So these "GPTs" are the combination of predefined prompts and custom API access? Not customly trained LLMs?

If so, I guess you can make such a "GPT" on your own server and independent from a specific LLM service by using a prompt like

    ...you have available an API "WEATHER_API". If you need the
    weather for a given city for a given day, please say
    WEATHER_API('berlin', '2022-11-24')
    and I will give you a new prompt including the data you
    asked for...

Or is there some magic done by OpenAI which goes beyond this?

mgon Mastering DOM manipulation with vanilla JavaScript

Well, since html elements already have a querySelectorAll function, you could just assign it to a new method "qsA":

    HTMLElement.prototype.qsA = HTMLElement.prototype.querySelectorAll;

Now myTable.qsA('tr') will work. You can try it right here on the HN page in the browser console:

    HTMLElement.prototype.qsA = HTMLElement.prototype.querySelectorAll; 
    document.querySelector('#hnmain').qsA('tr');

Why do you prefer chaining?

mgon Mastering DOM manipulation with vanilla JavaScript

    import {qsA} from './dqs.js';

    rows = qsA(myTable, 'tr');

mgon Mastering DOM manipulation with vanilla JavaScript

handlebars.js runs in the browser:

https://plnkr.co/edit/LXb5TOMoyAxdbTxB

mgon Mastering DOM manipulation with vanilla JavaScript

That's what I did before I put those lines (and some more) into dqs.js

mgon Mastering DOM manipulation with vanilla JavaScript

What's wrong with solo devs?

You are talking to me on a site done by one.

This is the frontend code:

https://news.ycombinator.com/hn.js

mgon Mastering DOM manipulation with vanilla JavaScript

These days, I get by with just two dependencies. One is dqs.js:

https://github.com/no-gravity/dqs.js

Which turns document.querySelector('.some .thing') into dqs('.some .thing').

The other one is Handlebars:

https://github.com/handlebars-lang/handlebars.js/

Which I use for all my frontend templating needs.

mgon AI companies have all kinds of arguments against paying for copyrighted content

For humans, it works.

mgon AI companies have all kinds of arguments against paying for copyrighted content

In wonder if copyrighted content will be needed at all in the future of AI training.

AlphaZero learned to play chess via self-play, not by reading books about chess.

Why couldn't the same happen for art for example?

For coding, won't a sufficiently advanced neural net be able to figure out how to use a programming language when given just the documentation?

And when most of our interactions are with AI, it will learn from our conversations with it. Asking some AI system why feature X was removed from programming language Y in version Z teaches it something. The next person who asks it which feature was removed from Y in version Z might be told "X". Without the AI system ever having to read about it. The interaction with AI could become a self-learning loop in and on itself.

mgon Email obfuscation rendered almost ineffective against ChatGPT

Is there no type of work a human can do in a short time which has value?

Say in 20 seconds.

If there is such a task, then it can't be automated. Because then it would not have value. So we could base a captcha on performing that task.

If there is no such task - what is the minimum amount of time in which a person can create value?

If there is no task for any length of time in which a human can create value - well, then we are all jobless :)

mgon Apple Spent $1B on the M3 Tape-Out, Says Analyst

    Apple claims the transition from TSMC's N5 to
    N3 has resulted in a 30% boost in performance
    for the M3's performance cores compared to M1's.

Is that 30% more performance while weight and power consumption are the same?

mgon Yann LeCun: AI one-percenters seizing power forever is real doomsday scenario

There are more nuclear warheads in Russia than in the USA.

mgon Yann LeCun: AI one-percenters seizing power forever is real doomsday scenario

The moat he describes the one-percenters might aquire is government regulation. Creating an environment where only an exclusive club of companies is able to work on AI.

Is that possible? Has that happened in any other industry?

I can imagine it happening in a single country. But wouldn't research route around that country then? How is the situation in Europe, China, India, Canada, Japan ... ?

mgon Htmx, Rust and Shuttle: A New Rapid Prototyping Stack

This is approaching the code size of the vanilla js version. And it does not have the confirmation modal.

If there would be additional computational logic involved (say calculating how many rides this car has done this year so far to say "Really delete this car? It delivered 124 rides this year.") - would you also cram all of that into the button element?

What I am trying to say: I don't see which real world use cases are tackled by htmx. I usually don't have a use case as simple as "When element X is clicked, load url Y and replace element Z with it".

mgon Htmx, Rust and Shuttle: A New Rapid Prototyping Stack

Hmm.. I don't think you can simplify their example to just that one line. For example, where is the url which is supposed to be requested to delete the item on the server.

mgon Htmx, Rust and Shuttle: A New Rapid Prototyping Stack

I wonder what to think of htmx.

Their primary example is this:

    <button hx-post="/clicked" hx-swap="outerHTML">
        Click Me
    </button>

But that is rarely what I want to do when a button is clicked. My typical example would be I have list of - say - cars, and each car provides a button to delete it:

    <button onclick="carList.deleteItem(this)">
        Delete this car
    </button>

And then in deleteItem(), I typically:

    let item=getItem(e);
    if (!confirm(`Really delete ${item['name']}?`)) return;
    let item_id=getItemId(e);
    items.splice(item_id,1);
    document.getElementById('items').innerHTML = template(items);

Where "template" is a handlebars template.

Would htmx have anything to say about this?

mgon Phind Model beats GPT-4 at coding, with GPT-3.5 speed and 16k context

In Vim, I tried to assign a shortcut to send the selected text to Phind (or any other LLM) and came up with this:

:'<,'>y|call system('firefox <url>?q='.shellescape(@*).' &')

The only problem left is that the text is not urlencoded.

There probably is some elegant way to urlencode it. But I did not come up with one yet.

mgon Camptown – HTML-based audio player

Since there is no demo and the repo looks super barebones (a loop which templates a list of mp3s in a directory?), do the upvotes mean that there is demand for an HTML based audio player?

What features would everybody want from such a player?

mgon The Windows installer of ImageMagick will no longer be signed

It is interesting, that the lack of a feature that would cost $629 to add is significant enough to make it the the HN front page.

Makes me feel like I would like to learn more about open source. What drives it's development and what the business models are.

There are 152 contributors to this project who wrote 21,686 commits. If each commit took an hour of work, and we value each hour at $50, that is $1,084,300 worth of time.

How can a project gather over a million dollars worth of work time, but not $629 for a certification service?

mgon One Year into Musk, X Monthly Users Are Down 15% and Ad Revenue Has Slumped 54%

I dislike what Musk has done to Twitter so far. It feels closed, greedy and messy.

But in fairness, one has to say that after building a $650B company (Tesla) and a $150B company (Space X), it is not necessarily a sign of business failure when one buys a $44B company they want to restructure and cut revenues in half in the process.

Like when you have built a 65 floor skyscraper and buy a 4 floor house as the base for your next project. You wouldn't probably be that scared to take off 2 of the floors if your plan is to reshape it into the next skyscraper.

mgon The downside of AI language translation

Douglas Hofstadter is one of my childhood heroes. I have devoured all of his books and always felt that his curiosity about artificial intelligence was similar to mine.

Now we made this huge jump in AI capabilities, and I am really surprised that he mostly seems to have negative feelings about it.

To me, it is quite the opposite. With these new LLMs, a childhood dream has come true. And I am mesmerized by the progress and possibilities.

mgon My ranking of the best numbers from 0-100

Only positive integers from 0 to 100?

If we would allow all real numbers, I would surely include pi and -1 in the top 100.

Since you already have 10 in there at position 4, I would say there is nothing so special about 100 that it deserves to be higher up than - say - pi.

So I would assign 3.14159 to position 7.

I think I would include -1 at position 8.

mgon 10-year Treasury yield rises to 5%, highest level for the key rate in 16 years

The CPI is not a good way to evaluate the value of the dollar. Especially not when discussing if rate hikes impact the stock market.

And again: Look at decades, not at years. Assets are way more expensive these days than they were in 2008. The S&P 500 more than tripled. The Nasdaq 100 increased more than 7-fold.

The economy is like a large swimming pool. If an elephant jumps into one side of the pool (increase of M0), it won't lift the water level on the other side (whatever prices you look at) in a linear way immediately. It will cause complex waves. But when the dust settles, you see higher prices.