It looks nice and friendly, but for developers I can recommend exploring caddy[1] or nginx[2]. It's a useful technology to have worked with, even if they're ultimately only used for proxying analytics.
242 karma · joined January 21, 2022
website: flessner.dev
Other than that, time and crypto are two things I also wouldn't code myself, both are just too easy to mess up.
The "copilot experiences", that finishes the next few lines can be useful and intuitive - an "agent" writing anything more than boilerplate is bound to create more work than it lifted in my experience.
Where I am having a blast with LLMs is learning new programming languages more deeply. I am trying to understand Rust better - and LLMs can produce nice reasoning to whether one should use "Vec<impl XYZ>" or "Vec<Box<dyn XYZ>>". I am sure this trivial for any experienced Rust developer though.
It's just everything else that's dreadful.
As an example, let's talk about "vibe coding" - It's a new term describing heavy LLM usage in programming, usually associated with Generation Z.
If I am asking an LLM to generate a German translation for "vibe coder" it comes up with the neutral "Vibe-Programmierer". When asking it to be more creative it came up with "Schwingungsschmied" ("vibration smith"?) - What?
I personally came up with the following words:
* Gefühlsprogrammierer ("A programmer, that focuses on intuition and feeling.")
* Freischnauzeprogrammierer ("Free-mouthed programmer - highlighting straightforwardness and the creative expression of vibe coding." - colloquial)
Interesstingly, LLMs can describe both these terms, they just can't create them naturally. I tested this on all major LLMs and the results were similar. Generating a picture of a "vibe coder" also highlights more of a moody atmosphere instead of the Generation Z aspects that are associated with it on social media nowadays.
Current generation of AI models can't think of anything truly new. Everything is simply a blend of prior work. I am not saying that this doesn't have economic value, but it means these AI models are closer to lossy compression algorithms than they are to AGI.
The following quote by Sam Altman from about 5 years ago is interesting.
"We have made a soft promise to investors that once we build this sort-of generally intelligent system, basically we will ask it to figure out a way to generate an investment return."
That's a statement I wouldn't even dream about making today.
It's only sad that the SDKs are often on the heavy side, I remember that the npm package used to be multiple megabytes as it bundled large parts of the core AWS SDK. Nowadays, I believe it's still around half a megabyte.
In all seriousness if such a system were to exist, why wouldn't the creator simply use it to make great returns for themself instead of scraps for everyone else?
I am not signing up for something I don't know. Before "committing" I want to see capabilities, documentation, pricing. Where's all that?
But yeah, I'd personally recommend Docker for self-hosting. Kubernetes or Proxmox always end up being too much to handle for personal use - or even small to medium sized companies.
There's also OVH, they have plenty of global regions, but I haven't used them yet.
However, the SolidJS ecosystem falls behind when building a full web application: The router is okay? Metadata is very separate... and SolidStart isn't pulling it all together in the way it should. Working with async is also a mess.
This makes it feel like you're having to learn 3-4 entirely separate systems.
In this particular case the creators were also harmed the most - the users didn't strictly get the "best" deals with Honey, but something is still better than nothing.
For hobbyists, I don't think that most of them strictly want to go through all these fillings. You could likely deliver "dumber" results, providing just bullet points on what's particularly interesting about a specific ticker.
Just as a heads-up, I am not attacking your product, it's just that the financial industry is very broad and you have to be mindful about your target audience. I have simply seen quite a few products that are "LLM with dataset X" fail because they don't integrate nicely with preexisting workflows.
Many hobbyists and "algo traders" only care for actionable results. Going through millions of records manually, even with AI, isn't really helpful.
Maybe I am in contact with too many quants, but this isn't even 5% of what's relevant for any given financial decision. Find out what the other 95% are about, integrate them into the product - now you have Bloomberg Terminal, great.
From a developer perspective many of these inconsistencies are rooting from inconsistent access patterns - operating system (ABI?), applications (ICP?), remote (TCP, HTTP, LDAP, FTP, ...?). All of these are "execution" or "information" but have to be programmed against differently.
Yet I just want StackOverflow for "life" questions.
The new generation of mods (e.g. Create) also have their charm, but they lost the simplicity - as did the game in general. And I am glad that's the case, wouldn't it be way worse if everything stayed the same?
Fits right in.
Great products are in my opinion a function of skill and care. The only benefit a "popular" tool or language gets you is a greater developer pool for hiring.
But yeah, it's quite baffling how in a couple years we seemingly went from stealing email addresses to credit cards to straight up billions of dollars.
If we can expect that everything shifts online eventually, where will this end? Clicked on the wrong link? Guess your house is gone... tough luck.
What helped me was looking at the "atomic concepts"
* What do I need to get done? (Tasks)
* When will I work on what? (Calendar)
* How do I keep information around? (Notes)
My "evergreen" information (like lecture notes, book notes) was happily living in Obsidian the past years, so criteria three was already "satisfied". I never found a true "system", so most of my notes are in a Zettelkasten-esque style.
I was stunned to discover that I didn't have a proper solution for "Tasks" or "Calendar". As an immediate fix I simply bought a DIN A6 notebook and a pen. Eventually, I started using the Apple Calendar with a Shortcut that could tally up the time for me - it was insightful. I went from >20 hours of social media a week to nothing (except HN) within a month.
I am still experimenting, currently I am trying to move the "Tasks" into a "daily note" in Obsidian. I have also tried to do some "Journaling", but I found it to not be effective. What I have found to be absolutely necessary though is having a dedicated time in the morning and evening to review everything, plan the day, defer tasks etc.
If it simply generates code from natural language then I am still fundamentally working with code. Aider as an example is useful for this, but anything that isn't a common function/component/class it falls apart even with flagship models.
If I actually put my "natural language code" under git then it'll lack specificity at compile time likely leading to large inconsistencies between versions. This is horrible user experience - like the random changes Excel makes every few years, but every week instead.
And everyone that has migrated a somewhat large database knows it isn't doable within minutes.
I am almost in disbelief that LLMs are the thing that reached the "tipping point" for most companies to magically care for ML. The amount of products, that could have been built properly 5 years ago, that exist now in a slower form because of "reasoning" LLMs, is likely astonishing.
Not in the sense that everyone should always aim for it, but simply as a "reality check". You won't be able to directly compete with large established players - so don't.
Unironically, that might be my lesson I take away from this - that I should maybe just try more, bumble about a bit, take things less serious...
Honestly, thank you for this website and great reaction to this feedback. More power to you!
Drive a story and tell me how you as the author feel, I care for your story and motivation to build this website - weave it into this work. Not to be overly rude, but it feels like you wanted another shiny thing in your portfolio instead of helping lonely people on Valentine's day.
Tell me 10 great things I could be doing with my time instead of being lonely and miserable. This website entertained me for 2 minutes, but what should I do with the other 23 hours and 58 minutes to escape the crushing loneliness?
I am sorry for writing this in a slightly excessive way, but as a user those might be my thoughts. What can I take with me from this experience essentially?
Also, I wouldn't ask for location without a good reason - as others have already mentioned it's better to put it behind Cloudflare as an example and use the country headers for basic regionalization.