<rss version="2.0" xmlns:dc="http://purl.org/dc/elements/1.1/" xmlns:atom="http://www.w3.org/2005/Atom"><channel><title>Hacker News: bastawhiz</title><link>https://news.ycombinator.com/user?id=bastawhiz</link><description>Hacker News RSS</description><docs>https://hnrss.org/</docs><generator>hnrss v2.1.1</generator><lastBuildDate>Sat, 22 Aug 2026 01:58:41 +0000</lastBuildDate><atom:link href="https://hnrss.org/user?id=bastawhiz" rel="self" type="application/rss+xml"></atom:link><item><title><![CDATA[New comment by bastawhiz in "Omacom Foundation launches with $8M"]]></title><description><![CDATA[
<p>The real issue for me is moving all the mail over. Since the organization is non standard, imap doesn't map cleanly and it all ends up being a huge mess. I'm not looking forward to getting collections and labels and whatever else untangled.</p>
]]></description><pubDate>Fri, 21 Aug 2026 19:20:30 +0000</pubDate><link>https://news.ycombinator.com/item?id=49392708</link><dc:creator>bastawhiz</dc:creator><comments>https://news.ycombinator.com/item?id=49392708</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=49392708</guid></item><item><title><![CDATA[New comment by bastawhiz in "Omacom Foundation launches with $8M"]]></title><description><![CDATA[
<p>I have a really hard time getting excited by this. I'm a Hey customer, and it's one of my biggest regrets. DHH promises a lot, then gets sidetracked with killing DEI or spending years migrating off AWS (for all the supposed savings, I certainly haven't seen any) or bending over backwards to have a framework to avoid front end frameworks or whatever the hot new flavor of the month is. Meanwhile, my Android app email client performs terribly when I'm on LTE (!) and I'm strong armed into using a calendar app that I don't want or need, while obvious spam and phishing emails routinely doesn't get caught and the search feature is less than useless. If changing email addresses was less tedious, I'd have done it already. I'll never trust DHH with another of his company's products, let alone trust him with my computer's operating system.</p>
]]></description><pubDate>Fri, 21 Aug 2026 17:21:08 +0000</pubDate><link>https://news.ycombinator.com/item?id=49391202</link><dc:creator>bastawhiz</dc:creator><comments>https://news.ycombinator.com/item?id=49391202</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=49391202</guid></item><item><title><![CDATA[New comment by bastawhiz in "Felony Bench"]]></title><description><![CDATA[
<p>Well it's not a benchmark, and it's not really representative of...anything except volume of research and what gets publicized. This mostly just measures how much testing each company does on models with relaxed guardrails and then talks about it. I'm not sure what kind of conclusion you can draw from that. Meta might have the most evil models but if they're piddling around not testing it, they won't ever find themselves with a "high score."</p>
]]></description><pubDate>Fri, 21 Aug 2026 16:55:52 +0000</pubDate><link>https://news.ycombinator.com/item?id=49390845</link><dc:creator>bastawhiz</dc:creator><comments>https://news.ycombinator.com/item?id=49390845</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=49390845</guid></item><item><title><![CDATA[New comment by bastawhiz in "Show HN: Open-source Stripe Connect alternative"]]></title><description><![CDATA[
<p>It's worth noting that this offloads the KYC/AML component to someone else down the road, but doesn't protect you from possible money laundering risk (e.g., you send USDC to a sanctioned individual who launders it into monero or something else). I'm no lawyer but I did work on Stripe Connect's compliance pieces for five years and I would NOT take the risk of (as a business) sending money to USDC addresses without actually verifying identities.<p>It also doesn't offer you protection against fraud (which is a huge problem that marketplaces face).<p>Frankly the missing piece to this is a real, actual identity verification component. If you're not checking identities against sanctions lists at the absolute minimum, you're not doing your due diligence. I wouldn't take that risk. The last thing you need is to be on the receiving end of a US Treasury inquiry.</p>
]]></description><pubDate>Thu, 20 Aug 2026 16:22:52 +0000</pubDate><link>https://news.ycombinator.com/item?id=49376768</link><dc:creator>bastawhiz</dc:creator><comments>https://news.ycombinator.com/item?id=49376768</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=49376768</guid></item><item><title><![CDATA[New comment by bastawhiz in "GPT 5.6 Sol is the best "vision" model OpenAI ever released"]]></title><description><![CDATA[
<p><a href="https://ai.google.dev/gemini-api/docs/deprecations" rel="nofollow">https://ai.google.dev/gemini-api/docs/deprecations</a><p>That link shows 3.1 pro listed as deprecated with no replacement model.</p>
]]></description><pubDate>Mon, 17 Aug 2026 22:54:06 +0000</pubDate><link>https://news.ycombinator.com/item?id=49338811</link><dc:creator>bastawhiz</dc:creator><comments>https://news.ycombinator.com/item?id=49338811</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=49338811</guid></item><item><title><![CDATA[New comment by bastawhiz in "AI;DR (AI; Didn't Read)"]]></title><description><![CDATA[
<p>Can you read something and tell when an author you like wrote it without seeing the attribution? I can. I think most of us could, because people use certain phrases or have a particular style. That doesn't make it bad.<p>AI writing isn't actually all that bad. But it has a very specific style, and there are MANY tells. I use Claude every single day, and it's immediately obvious to me when I'm reading something Claude wrote. "It's not X, it's Y" "That's the reframe" "The gap is" I could go on.<p>I too like using em dashes, and have been since I learned the keyboard shortcut over a decade ago. I've never been accused of posting ai-written text because of it.<p>AI is indeed trained on good writing, but it's also trained on bad writing. And it's post trained on specific things that are writing that aren't good or bad for reasons unrelated to style. What you're left with is a one-size-fits-all writing style. Doing math? Doing medicine? Doing anime waifu role play? It all starts from the same generic milquetoast style until you tell it not to. And to me (and apparently many other people) that's super easy to pick out.</p>
]]></description><pubDate>Mon, 17 Aug 2026 22:49:29 +0000</pubDate><link>https://news.ycombinator.com/item?id=49338751</link><dc:creator>bastawhiz</dc:creator><comments>https://news.ycombinator.com/item?id=49338751</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=49338751</guid></item><item><title><![CDATA[New comment by bastawhiz in "GPT 5.6 Sol is the best "vision" model OpenAI ever released"]]></title><description><![CDATA[
<p>3 and 3.1 Pro are both marked as deprecated by Google. Even if they're the best Google offers, it would be foolish to choose a model that's explicitly deprecated.<p>It's not a technical problem, it's a commercial one. If Google can't ship a model to replace the one they deprecated, that tells you everything you need to know about choosing a Gemini model for whatever you're trying to do.</p>
]]></description><pubDate>Mon, 17 Aug 2026 20:18:26 +0000</pubDate><link>https://news.ycombinator.com/item?id=49337002</link><dc:creator>bastawhiz</dc:creator><comments>https://news.ycombinator.com/item?id=49337002</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=49337002</guid></item><item><title><![CDATA[New comment by bastawhiz in "GPT 5.6 Sol is the best "vision" model OpenAI ever released"]]></title><description><![CDATA[
<p>3 Flash never left "preview" status and is listed as deprecated.<p><a href="https://ai.google.dev/gemini-api/docs/deprecations" rel="nofollow">https://ai.google.dev/gemini-api/docs/deprecations</a></p>
]]></description><pubDate>Mon, 17 Aug 2026 18:05:25 +0000</pubDate><link>https://news.ycombinator.com/item?id=49335152</link><dc:creator>bastawhiz</dc:creator><comments>https://news.ycombinator.com/item?id=49335152</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=49335152</guid></item><item><title><![CDATA[New comment by bastawhiz in "GPT 5.6 Sol is the best "vision" model OpenAI ever released"]]></title><description><![CDATA[
<p>3 Pro is quickly approaching one year old. There's almost no reason to benchmark it, especially since a new version of Gemini Pro was supposed to be released mid 2026 and hasn't seen the light of day.</p>
]]></description><pubDate>Mon, 17 Aug 2026 18:00:55 +0000</pubDate><link>https://news.ycombinator.com/item?id=49335096</link><dc:creator>bastawhiz</dc:creator><comments>https://news.ycombinator.com/item?id=49335096</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=49335096</guid></item><item><title><![CDATA[New comment by bastawhiz in "Self hosted email continues to steeply decline"]]></title><description><![CDATA[
<p>And soon, I'll be lazy and use an LLM to do the reading for me. Eventually the LLMs will train on the state of the world and decide to be lazy also, and the "writer's" LLM prompt will just be shared with the "reader": nobody actually needed all those words anyway! Maybe someday we'll get rid of the links on HN and we can just share and consume headlines.</p>
]]></description><pubDate>Mon, 17 Aug 2026 12:25:24 +0000</pubDate><link>https://news.ycombinator.com/item?id=49329741</link><dc:creator>bastawhiz</dc:creator><comments>https://news.ycombinator.com/item?id=49329741</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=49329741</guid></item><item><title><![CDATA[New comment by bastawhiz in "Signal is working on a paid option to create an account without a phone number"]]></title><description><![CDATA[
<p>I would bet that the overwhelming majority of folks trying to use Signal without a phone number are on devices that aren't phones and where the user doesn't want to use their phone. For instance:<p>- Folks running businesses without a business phone number<p>- People who want to give their kids the ability to text them, but their kids only have an iPad<p>- People who want to give texting capabilities to bots without having their own phone number involved</p>
]]></description><pubDate>Mon, 10 Aug 2026 23:22:40 +0000</pubDate><link>https://news.ycombinator.com/item?id=49251232</link><dc:creator>bastawhiz</dc:creator><comments>https://news.ycombinator.com/item?id=49251232</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=49251232</guid></item><item><title><![CDATA[New comment by bastawhiz in "Mark Zuckerberg attacks 'closed' AI rivals as Meta returns to open models"]]></title><description><![CDATA[
<p>Because it took off. All of a sudden they captured more customers than they could have imagined they would have, and throwing away the lead they unintentionally gave themselves (in terms of usage and mindshare) would have undone all of that and more.</p>
]]></description><pubDate>Mon, 10 Aug 2026 21:26:33 +0000</pubDate><link>https://news.ycombinator.com/item?id=49249978</link><dc:creator>bastawhiz</dc:creator><comments>https://news.ycombinator.com/item?id=49249978</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=49249978</guid></item><item><title><![CDATA[New comment by bastawhiz in "How We Pushed CDC into Postgres"]]></title><description><![CDATA[
<p>Aurora zero etl was a nightmare for us. Almost any schema changes require a VACUUM FULL for it to continue functioning. On a few occasions, it just stopped running without an obvious explanation, requiring slow and lengthy back and forth threads with AWS support. If it worked as advertised, it would be great, but I can't recommend it for any serious production system.</p>
]]></description><pubDate>Mon, 10 Aug 2026 03:11:20 +0000</pubDate><link>https://news.ycombinator.com/item?id=49238811</link><dc:creator>bastawhiz</dc:creator><comments>https://news.ycombinator.com/item?id=49238811</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=49238811</guid></item><item><title><![CDATA[New comment by bastawhiz in "How We Pushed CDC into Postgres"]]></title><description><![CDATA[
<p>Clickhouse really nailed this with the acquisition of peerdb. I used it with many terabyte databases and I essentially never thought about it. The only thing we really had to watch for was trying to replicate too much at once (because of the physical compute/io capacity of the postgres or clickhouse clusters).</p>
]]></description><pubDate>Mon, 10 Aug 2026 03:09:26 +0000</pubDate><link>https://news.ycombinator.com/item?id=49238797</link><dc:creator>bastawhiz</dc:creator><comments>https://news.ycombinator.com/item?id=49238797</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=49238797</guid></item><item><title><![CDATA[New comment by bastawhiz in "Ask HN: What are you working on? (August 2026)"]]></title><description><![CDATA[
<p>A search engine powered by ai? A search engine for ai? A search engine whose corpus covers data about ai?</p>
]]></description><pubDate>Mon, 10 Aug 2026 00:33:33 +0000</pubDate><link>https://news.ycombinator.com/item?id=49237854</link><dc:creator>bastawhiz</dc:creator><comments>https://news.ycombinator.com/item?id=49237854</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=49237854</guid></item><item><title><![CDATA[New comment by bastawhiz in "Taxi drivers rarely die of Alzheimer's"]]></title><description><![CDATA[
<p>I'm gonna go out on a limb and say that not even a double digit percentage of taxi drivers' careers are ended due to violent attacks. Taxi driver doesn't even appear in the U.S. Bureau of Labor Statistics top 20 most dangerous jobs.</p>
]]></description><pubDate>Sun, 09 Aug 2026 22:33:32 +0000</pubDate><link>https://news.ycombinator.com/item?id=49236882</link><dc:creator>bastawhiz</dc:creator><comments>https://news.ycombinator.com/item?id=49236882</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=49236882</guid></item><item><title><![CDATA[New comment by bastawhiz in "Fixing my tooltip accessibility mistake"]]></title><description><![CDATA[
<p>That's the thing, though. Tooltips aren't just text. Qt, GTK, and Windows all support some flavor of rich text tooltips, so it's not even an operating system constraint. The simplest solution for this in an accessible way looks exactly like Aria labels.<p>But here's the thing: a text tooltip assumes the label you show to screen readers is the tooltip. That's not always true. In this case, the tooltip serves the purpose of labeling a button icon. A tooltip on a text button doesn't have the same purpose from an accessibility standpoint. There's simply no universal way to define what a tooltip <i>is</i> in a way that semantically correct.</p>
]]></description><pubDate>Sun, 09 Aug 2026 16:04:00 +0000</pubDate><link>https://news.ycombinator.com/item?id=49232647</link><dc:creator>bastawhiz</dc:creator><comments>https://news.ycombinator.com/item?id=49232647</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=49232647</guid></item><item><title><![CDATA[New comment by bastawhiz in "Fixing my tooltip accessibility mistake"]]></title><description><![CDATA[
<p>The title attribute isn't supportive of accessibility. It also doesn't work on touch screens.<p>From <a href="https://www.a11y-collective.com/blog/aria-label-vs-title/" rel="nofollow">https://www.a11y-collective.com/blog/aria-label-vs-title/</a><p>> The content of the title attribute is often ignored or read inconsistently, leaving users who rely on assistive technology at a disadvantage.<p>Regardless of whether screen readers <i>should</i> better support title attributes, that's not the world we live in and blind users don't benefit from websites that fail to work because of principled takes.</p>
]]></description><pubDate>Sun, 09 Aug 2026 15:23:18 +0000</pubDate><link>https://news.ycombinator.com/item?id=49232272</link><dc:creator>bastawhiz</dc:creator><comments>https://news.ycombinator.com/item?id=49232272</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=49232272</guid></item><item><title><![CDATA[New comment by bastawhiz in "AMD acquires Taalas to boost inference performance by etching models in silicon"]]></title><description><![CDATA[
<p>The weights might fit in cache, if you're using a small model. If you wanted to have a 20B+ parameter model, that's just going in RAM. You could put more RAM in the device and pay the perf cost or have a dedicated chip. Most devices already have a dedicated chip, this just changes which silicon you're spending the money on.</p>
]]></description><pubDate>Thu, 06 Aug 2026 22:39:24 +0000</pubDate><link>https://news.ycombinator.com/item?id=49203604</link><dc:creator>bastawhiz</dc:creator><comments>https://news.ycombinator.com/item?id=49203604</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=49203604</guid></item><item><title><![CDATA[New comment by bastawhiz in "Dress made of living mycelium can renew and repair itself"]]></title><description><![CDATA[
<p>Exactly. You can place a patch with some glue over a hole in fabric to achieve the same thing in a traditional garment. Iron-on repair patches are also similar.<p>It's one thing to say "get it wet and leave it in the sun". It's another thing to require identical replacement material and nutrient slurry (I don't know about you, but I think I can source fabric glue more easily than potato water).</p>
]]></description><pubDate>Thu, 06 Aug 2026 19:22:17 +0000</pubDate><link>https://news.ycombinator.com/item?id=49201115</link><dc:creator>bastawhiz</dc:creator><comments>https://news.ycombinator.com/item?id=49201115</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=49201115</guid></item></channel></rss>