Just curious and who knows, maybe someone will adopt it or develop something new based on its ideas.

Ask HN: Abandoned/dead projects you think died before their time and why?

Looking for windows software that replaces the need for keyboard and mouse and lets me control Windows

Ask HN: What's the best alternative to Dragon NaturallySpeaking?

I want to get a second TV which will more or less be a second monitor for my System76 laptop which is plugged into a bunch of music equipment, like a korg midi keyboard, and a novation drum pad, all of which work great with linux.I want to buy this TV used. I&#x27;m seeing a bunch of Samsung, LG, RCA, Sony, etc on Facebook Marketplace. What a cesspool Facebook has become, right?Any suggestions on the best brand or even model for that kind of thing? I don&#x27;t really want to battle with a bunch of shit that tries to coerce me to install another app from a streaming provider slash gambling entrypoint.I imagine mostly it will just need HDMI to work, and all the TVs will support that. But, I thought maybe there would be a fun brand that offers interesting other options.

Ask HN: What's the best hackable smart TV?

This week, my Vercel account was suddenly disabled and all of my deployments and data were deleted.I received no warning, no suspension notice, and no post-action explanation. When I emailed their support to ask what happened, they refused to give a reason and simply said they wouldn’t restore the account or data.I’ve been a Vercel user for over two years and never had any issue until now. It’s honestly shocking to lose everything overnight without a single email or warning.I’m now migrating all my projects to Cloudflare, but the experience has been incredibly frustrating. I just want to understand why this happened — and whether anyone else has faced a similar sudden ban.Has anyone here managed to get a clear answer from Vercel in a situation like this?

Lost all my sites overnight: Vercel terminated my account without notice

The US Census data explorer (https:&#x2F;&#x2F;data.census.gov) is currently broken. A redirect with a warning message that &quot;the website will not be updated due to the shutdown&quot; was put in place at some point. Existing assets such as javascript assets are being redirected to this HTML page, thus breaking the entire website.This tool is essential for researchers to access US population statistics. The modification in connection with the US government shutdown seems to have done more harm than any good it could have caused.

Tell HN: After the US gov shutdown, the Census data website was broken

Hello!This is a heads-up for folks who run CrowdStrike Falcon on Linux servers, and particularly on Linux servers that were provisioned some time ago. It&#x27;s a problem that CrowdStrike does not plan on fixing, and so I wanted to let others know before it causes your machines to hang.You should have CrowdStrike Falcon installed at path &#x2F;opt&#x2F;CrowdStrike&#x2F;. In that directory, you probably have one file whose name begins with &quot;KernelModuleArchive&quot;, and many files whose name begins with &quot;KernelModuleArchiveExt&quot;. That&#x27;s the problem.CrowdStrike appends a version number to every executable &amp; library file. It does a good job of cleaning up old versions of almost all of its files. Except for KernelModuleArchiveExt.I first noticed this happening when a virtual machine (with a small &#x2F;opt partition) filled up &#x2F;opt, and the system stopped responding. Turns out, &#x2F;opt&#x2F;CrowdStrike had filled up with 18 different KernelModuleArchiveExt files.What is the fix? Well, our CrowdStrike admins opened a ticket with CrowdStrike, and we were told:* Yes, the KernelModuleArchiveExt files are not being cleaned up automatically. Other files are being cleaned up automatically, but not the KernelModuleArchiveExt files.* Will CrowdStrike release an update that cleans up the KernelModuleArchiveExt files? No.* Will you put it on your roadmap to implement in the future? No.* So, what should we do? If you want to clean them up, do it yourself.If your site uses CrowdStrike uninstall protection, you cannot clean them up yourself without first getting a &quot;maintenance token&quot; from your CrowdStrike admins. Otherwise, deleting all KernelModuleArchiveExt files and restarting the CrowdStrike Falcon sensor works (it goes out and downloads the KernelModuleArchiveExt that it needs). Personally, though, I don&#x27;t think we should have to do this.Since CrowdStrike refuses to fix this, I wanted to let folks know, so you can check your systems. If you discover that this problem also affects you, I encourage you to open your own support ticket with CrowdStrike.

Tell HN: CrowdStrike Falcon users, check for excess KernelModuleArchiveExt files

Ask HN: Why is observability so broken, and what can fix it?

The best way to really understand how something works is to build it yourself. So I am wondering if there are any good tutorials on building your own LLM from scratch. I.e. implementing tokenisation, embeddings, attention and so on. I am not suggesting one could replicate chatGPT, but more a toy model that implements the core features but based on a much smaller corpus and training data.

Ask HN: Build Your Own LLM?

Right now in many communities there are mixed messages regarding beginners and use of AI, some say that you should stay away from it and others suggest getting jumping right in, so as not to stay behind. So I feel that there are many people like me who are confused and kind of unsure on how to proceed. If you use AI heavily in your work or personal projects, I would love to hear your suggestions.

How can newbies make proper use of AI and still be good developers?

every damn month it’s the same fucking story project done client happy invoice sent then NOTHING????i wait a week i send the &quot;hey just checking in email&quot;i feel like an idiot asking for money i’ve already earnedtwo weeks later i’m refreshing stripe like a fooli am sick and tired of having to beg for my moneyso i thought of creating a small ai that links to stripe&#x2F;paypal figures out who is likely to be late on a payment and emails them automaticallyit starts off polite gets firmer with time stops me from having to chase payments and even shows how much money $$ it recovered so i can stop guessing who’s ghosting.not selling anything. not built yet. just trying to see if this pain’s real enough that people (like you) would drop around $49&#x2F;mo for it. be real would you actually pay for this, or nah?i just need to know if this is a real pain or if i’m over it.

My clients keep ghosting invoices and I'm so fucking done

Telling how much electricity in watt-hours was used for preparing this response.And it also would sum consumed energy from other requests it made. Or perhaps send it in a separate header.I don&#x27;t know where to get this number or if it&#x27;s even possible.

Ask HN: Should we maybe add header "X-Electricity-Used: 42" to responses?

I&#x27;m exploring moving away from work with VC-backed B2B SaaS startups and towards working with B2C businesses on Main Street.For anyone else who&#x27;s felt inclined to make a similar pivot - any advice? Particularly when it comes to remaining challenged, informed, and translating valuable hype-cycle concepts to people who don&#x27;t care about hype.While it&#x27;s interesting to play with shiny toys, helping investors make money by creating products related to the hype-cycle de jour that get used by other B2B SaaS companies just feels soulless.(My background is in GTM systems and data analytics mostly, but I think the question applies to anyone)

Ask HN: Any advice on pivoting out of VC-backed tech?

Going out on a limb hereI’ve been working on a prototype payment network that allows people to send and receive payments at a fraction of the cost to traditional rails.I started working on this when building out the previous fintech I started, Safepay (YC S20) and realized how much value card networks extract from fintechs on both sides, and by extension the businesses that rely on them and thought that there has to be a better way.I think I’ve found what this could be and the only thing holding me back is finding the right person to take this on with.So what is it? (Links added below)The idea is to allow people to make and receive payments through simple aliases. Send money from your Chime account to your friend on Coinbase or pay a merchant using Stripe with your Revolut account. Scan a QR code on a receipt with your banking app after dinner and settle the bill.In the background we pass messages back and forth between different participants and process net, deferred settlements with stablecoins.Are you interested?If you think this could be something you see yourself working on, email me at ziyad@paylias.xyz. You should be technically strong so we can distribute the workload between each other and humble enough to know that this will be a grind that may take a few years to play out. I’m sure there are others thinking or working on this so time is against us.Thank you for reading and even if this isn’t relevant for you, you could help build a company just by upvoting this and bringing this in front of the right person

Ask HN: Does anyone want to take on the payment processing networks with me?

It feels like getting investor intros is still one of the biggest bottlenecks for early-stage founders.I’ve seen countless threads saying “you need warm intros” - but in reality, not everyone has that network on day one. Cold emails and DMs often go unread, and accelerator “investor lists” feel outdated or generic.So I’m curious:1. How are you finding and reaching out to investors without existing connections?2. Have any particular strategies, tools, or outreach methods worked for you recently?3. Is personalization or timing more important in getting actual replies?Would love to hear what’s been working (or not) in 2025’s fundraising environment

Ask HN: How are founders finding investors without warm intros in 2025?

I&#x27;ve been working on improving aspects of my life that I don&#x27;t feel serve me super well. I&#x27;ve made really strong progress in physical areas, but digital has been much more challenging.I spend way too much time browsing Reddit, YouTube&#x2F;Twitch, and Facebook. Thankfully, I never got into Insta or TikTok. I want to drastically reduce the amount of time I spend interacting with these sites, but they feel so deeply engrained into my psyche that it feels nearly impossible.I&#x27;ve tried some of the simpler suggestions like configuring a Tampermonkey extension to limit access, or installing Cold Turkey (https:&#x2F;&#x2F;getcoldturkey.com&#x2F;). The problem is that I know how to work around these things (you can just edit your clock time to mitigate Cold Turkey ...) and so, eventually, the cravings become enough that I do so even though it&#x27;s not in my best interest.Another issue is that sometimes sites have valuable information. For example, some subreddits have useful information related to programming, fitness, etc. and I&#x27;ll encounter them through a Google search. If I&#x27;m prevented from accessing this information when doing legitimate research then I find that frustrating and wish there was an exception to the rule. I&#x27;m not sure how grounded that stance is, though.In the physical world, I&#x27;ve had great success defeating habits by limiting physical access. I don&#x27;t keep snacks at home and, if I must, I&#x27;ll put them in a timed, locking container (https:&#x2F;&#x2F;www.amazon.com&#x2F;dp&#x2F;B00E9J3MLM) such that I&#x27;m unable to open it when having a craving. This helped me with things like delicious toppings for salads which I wanted to keep on-hand, but not be tempted by 24&#x2F;7.The problem is more challenging when working with digital addictions. I&#x27;m wondering what my options are here? I assume something related to parental controls at&#x2F;near the hardware level and a time-release password? Not sure if anything time-based is viable if I can just mess with my computer&#x27;s clock time, but maybe if it dials out to a third-party server for timing info and I don&#x27;t go so far as to MITM the response? I&#x27;d prefer solutions that I&#x27;m able to implement myself rather than relying on repeatedly handing passwords off to friends, but am understanding if that&#x27;s a hard requirement.Anyone had any success here? Thanks!

Ask HN: Most effective way to reduce excessive digital media consumption?

We believe the next 5–10 years will completely transform how products are discovered. Instead of search engines, agentic AI tools will explore, evaluate, and recommend products directly to the right customers, fundamentally changing marketing, SEO, and user acquisition.We’re building in this space, with a focus on geo-contextual discovery (where location, intent, and AI reasoning meet). We’re looking to partner with 5 forward-thinking agencies or B2C startups for an early pilot.You’ll get early access, influence the roadmap, and help your marketing team get ahead of the coming wave, while we learn from your use cases to shape the product.If you’re interested in experimenting at the frontier of AI-driven marketing, reach out, and we’d love to chat.

Looking for 5 B2C startups or agencies to pilot next-gen product discovery

Most of the AI terminal tools (Cursor Agent, Claude Code, Continue.dev, etc.) are marketed around coding assistance, but I&#x27;m curious how far people have pushed them beyond that.Has anyone here used these agents purely as general-purpose operators, e.g. for writing, research automation, bulk file&#x2F;knowledge base refactoring, inbox processing, shell task orchestration, etc.?