HN user
Writes occasionally at technicalchops.com
I wrote this a year or so ago: https://www.technicalchops.com/articles/ai-goes-to-hollywood...
"The studios and creators who thrive in this new landscape will be those who can effectively harness AI’s capabilities while maintaining the human creativity and vision that ultimately drives the art of cinema."
It is in many ways thrilling to see this come to life, and I couldn't agree with you more.
The lede is buried at the end of this article.
Later iOS 26 beta releases show Apple reducing transparency and adding blur effects for better readability.
This is a beta release. It is a work in progress. When iOS 7’s betas came out the reaction was similarly negative. I would suggest we wait and see what the system evolves into; by the time we get to iOS 27 I am quite sure that Apple will have found the right balance.
A fascinating situation is facing polymarket - Zelenskyy was pictured wearing a suit (as described by the media) at the NATO summit on June 25th but this market is still unresolved. It appears there’s lots of argument here and that this market is set to be test of the platform.
When I tested, you can also screenshot disney+, netflix etc. so DRM is also broken by it!
WhatsApp was the one I tested it on
This video[1] seems to give some insight into what the process actually is, which I believe is also indicated by the output token cost.
Whereas GPT-4o spits out the first answer that comes to mind, o1 appears to follow a process closer to coming up with an answer, checking whether it meets the requirements and then revising it. The process of saying to an LLM "are you sure that's right? it looks wrong" and it coming back with "oh yes, of course, here's the right answer" is pretty familiar to most regular users, so seeing it baked into a model is great (and obviously more reflective of self-correcting human thought)
The comments on this thread are interesting. I use Laravel extensively. For big applications, serving lots of users. It works when the application is relatively complex, and the ecosystem is second to none. Need to just throw it up on a server? There's Forge[1]. A better CI/CD process? Envoyer[2]. Want serverless? Not for me, but Fathom[3] use it to deal with >100Ms of hits a day; there's Vapor[4]. All three of those are Laravel developed and maintained solutions.
If I'm throwing something small together then sure, I'll maybe use Flask or something lightweight[5]. But Laravel is very good for nearly every use-case where you intend to actually build something.
Then there's the bigger question: if you're building to meet a business use case, or well, to make money, then why wouldn't you use the most complete scaffold possible? I'd say Laravel is that. If it's too much of a pain to do something in PHP I can just stick in a call to a python file or really whatever language I want. But for the basics? A db? Auth? and lots of other stuff that I never want to personally build again? Yeah, give me Laravel everyday.
[1] https://forge.laravel.com/ [2] https://envoyer.io/ [3] https://usefathom.com/ [4] https://vapor.laravel.com/ [5] To prove I'm not a shill, this is from literally last night: https://github.com/simonminton/consensus-article
The article states that "They scan an image through a thin slit up to 2,000 times a second", whereas it has been widely reported that it is actually 40,000 times a second[1]
The article probably wasn't wrong, for when it was first written. This is a curious internet thing - this article is a decade old and has been updated incrementally to keep it somewhat relevant, however because it's about tech that keeps advancing it ends up being a misleading source.
If you look at all of the sources, they're from January 2014 but because the article is undated it leads you to think it's is correct. It's an interesting problem. An old textbook is clearly an old textbook, but a website can just have modern CSS applied, dates removed and there is no apparent guide to the freshness of the article. Internet problems.
[1] https://www.axios.com/2024/08/05/noah-lyles-wins-gold-track-...
I've been bullish[1] on this as a major aspect of generative AI for a while now, so it's great to see this paper published.
3D has an extremely steep learning curve once you try to do anything non-trivial, especially in terms of asset creation for VR etc. but my real interest is where this leads in terms of real-world items. One of the major hurdles is that in the real-world we aren't as forgiving as we are in VR/games. I'm not entirely surprised to see that most of the outputs are "artistic" ones, but I'm really interested to see where this ends up when we can give AI combined inputs from text/photos/LIDAR etc and have it make the model for a physical item that can be 3D printed.
[1] https://www.technicalchops.com/articles/ai-inputs-and-output...
I'd be interested in seeing a similar analysis but with a slight twist:
We use (in production!) a prompt that includes words to the effect of "If you don't get this right then I will be fired and lose my house". It consistently performs remarkably well - we used to use a similar tactic to force JSON output before that was an option, the failure rate was around 3/1000 (although it sometimes varied key names).
I'd like to see how the threats/tips to itself balance against exactly the same but for the "user"
I knew Matthew's brother years ago and met him a few times, he now does pretty cool (and reasonably priced) watches: https://www.mhdwatches.com/
It is about credibility. Ex-Harvard, ex-Goldman Sachs, ex-Apple - whatever it is, the institutions that you have previously been a part of can lend their credibility to you and therefore change the reader's perception of your own competence. The goal, of course, is to have built your own reputation which secures that credibility, but it is perfectly acceptable to say "these people who know what they are doing trusted me and therefore you should too".
I agree with the sentiment of anti-DRM but Anker (who are highly reputable) have been doing licensed cables for <$10 for years.
I understand what you are saying. I should clarify my comment.
- All users can delete their accounts (and all related data) at https://www.ringerhq.com/user/profile which is accessible from the dashboard
- We have deliberately not created a marketing list because I don't believe that is what the open source community wants, nor needs, from what we are doing here. I emailed OP and other users requesting feedback on the original version of the CV, not to market anything.
I appreciate that Ringer is not for everyone, and that's OK - there is no pressure to use the site.
However, I do have a duty to defend what we are (and are not) doing. This is not a data grab - it's an attempt to build something that returns value to the open source community. If people have fears or concerns about what we are building then I do genuinely want to hear them, both to understand and attempt to mitigate for the same, and where necessary allay them.
Hi Remram,
The emails have no unsubscribe link because they were sent directly from me, from my personal email account. There is no marketing mail list, which is why you have received 2 emails in total since you signed up almost a year ago.
The OAuth is there to enable us to continue building more features which will include using features only available using authenticated login. I am not sure why you have a particular dislike for this project but you have made it quite clear here, by email to me and in a previous HN post.
I am always open to understand what exactly your fears are about what we are doing - you have my email address.
Thanks - I'm trying to debug it but it seems very inconsistent. We had a clear hour and then it's restarted again!
Hi - I'm digging into this, I can see it in the logs and am trying to work out why - it looks like an issue in the OAuth process for some users only
Done! Apologies for the oversight - everything we release will be under the MIT License.
Great to see more in this space. We've been working on the same problem with https://www.ringerhq.com for the past ~year and have seen great success with specific projects. We're changing our approach slightly to give contributors and maintainers more tools after receiving feedback on specific features. If I can reiterate one piece of advice: listen to the community!
Hey - would you mind clarifying what you mean re whitelisting?
An expert can only be an expert on a repository that they have a history of contributing to, so I would hope that that would have an effect on the advice they give. If they still chose to give bad advice then I would be disappointed.
Now there is! ;-)
Yeah - this is where I see us fitting well. It's not really the standard install/import and you're done issues that docs deal with. It's the niche problem <> niche expertise area where I believe we can excel - these tend to be very competent teams or developers who would be willing to take that hour and pay for it because the value is high enough.
Thank you! I enjoy the big and not easy things, it's where the magic happens!
Yeah - it's all in the vetting. I will dogfood a solution shortly!
So we are logging requests and will reach out manually. Not today, given the Hacker News traffic load! But we are intending to put in a more scalable solution - what we don't want to do is spam people, which would be the result if it were fully automated.
We check contributions, but have a slightly more nuanced system planned and in testing. It's a difficult problem, but we do have thoughts on ways to ensure quality of experts remains high.
Hmm on this [1] page there is a link which should allow you to appear as an expert without completing the Stripe process. I think we might not be routing everyone through that page though!
You can stay that way until you receive a request, which you cannot accept until you have set up payments.
OK - this is something that I am going to work on. Traditionally marketplaces have been a race to the bottom, and I think (in fact know, through conversations) that many developers don't necessarily know the market rate for their skills. Some absolutely do (Shout out to KrauseFx https://www.ringerhq.com/i/fastlane/fastlane).
I do think there is a balance, especially when the request comes from an individual vs a business and we're actively looking into ways to solve this. I have partially realised that educating developers on value driven pricing is going to be part of my job now!
Thanks for the well wishes!
Hey! So this is something that we are really interested in, and want to support. I've had about a dozen conversations about this, especially where someone has created packages for a larger piece of software which necessitates a deep knowledge of the core project.
I want to allow it, but I need to work out a way to do so whilst making sure that we don't end up with a free for all - that means manual review perhaps, but I agree that it would be very helpful if we can make it work.
Re GitHub - we've built on GitHub initially but would love to extend to other platforms in time!
Hi pabs3, thanks for your comments! We already have an improved "clear English" version of the ToS, which you will be glad to hear includes updates on changes going through review now - getting the legals right has been a really big deal for me, especially to make sure they are fair to everyone!
Thanks you!