<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: gizmo686</title><link>https://news.ycombinator.com/user?id=gizmo686</link><description>Hacker News RSS</description><docs>https://hnrss.org/</docs><generator>hnrss v2.1.1</generator><lastBuildDate>Fri, 20 Nov 2026 07:09:28 +0000</lastBuildDate><atom:link href="https://hnrss.org/user?id=gizmo686" rel="self" type="application/rss+xml"></atom:link><item><title><![CDATA[New comment by gizmo686 in "Nvidia allegedly engineered scarcity by funding AI startups to buy its GPUs"]]></title><description><![CDATA[
<p>Almost none. Dealerships will arrange financing for you, but the actual financing comes from a third party financial institution. The dealership walks away from the deal with the full cash value of the sale; and maybe a kickback from the bank.
The concerning thing with vendor financing is that the vendor is supposed to be in the widget business, not the financing business. If they are getting into the financing bussiness, that is generally a sign that the professional financers don't want to get involved; which is not a good sign.</p>
]]></description><pubDate>Sat, 14 Nov 2026 05:08:55 +0000</pubDate><link>https://news.ycombinator.com/item?id=50494171</link><dc:creator>gizmo686</dc:creator><comments>https://news.ycombinator.com/item?id=50494171</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=50494171</guid></item><item><title><![CDATA[New comment by gizmo686 in "Nintendo Takes Down Emulators and Firmware in Massive GitHub DMCA Sweep"]]></title><description><![CDATA[
<p>Notably, neither lockpicks nor lockpicks are illegal.<p>By making circumvention itself illegal the DMCA gives the producer to essentially invent new IP law.<p>Want to take a DVD you bought while vacationing in China and play it back home in the US. That is perfectly legal. However, if the DVD has even the flimsiest form of region locking, you would need to circumvent them, which is a crime.</p>
]]></description><pubDate>Fri, 13 Nov 2026 03:38:00 +0000</pubDate><link>https://news.ycombinator.com/item?id=50477920</link><dc:creator>gizmo686</dc:creator><comments>https://news.ycombinator.com/item?id=50477920</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=50477920</guid></item><item><title><![CDATA[New comment by gizmo686 in "Only 16% of U.S. fast-charging stations meet federal NEVI standards (open data)"]]></title><description><![CDATA[
<p>I already see EV charging symbols on exist signs.<p>As to every gas station having a fast charger; the answer is probably never. I doubt that the EV economy would ever be able to support that number of fast charging stations.<p>Consider that with ICE cars, approximatly every mile driven is powered by gas from a gas station. With EVs, a significant amount of miles driven is powered by residential electricity. Many more miles driven can be powered by slow destination chargers (e.g. at workplaces). This means that the addressable market of fast charges are: A) long road trips, and B) people without convenient access to slow chargers.<p>In addition to that, most fast chargers are just plopped onto random shopping center parking lots. The needed infastructure is just space, and a large enough grid connection.</p>
]]></description><pubDate>Fri, 06 Nov 2026 19:28:27 +0000</pubDate><link>https://news.ycombinator.com/item?id=50397148</link><dc:creator>gizmo686</dc:creator><comments>https://news.ycombinator.com/item?id=50397148</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=50397148</guid></item><item><title><![CDATA[New comment by gizmo686 in "Only 16% of U.S. fast-charging stations meet federal NEVI standards (open data)"]]></title><description><![CDATA[
<p>This number is pretty much meaningless. The standards being measured are:<p>1) four CCS ports
2) 150 kW each
3) within 1 mile of a designated corridor<p>1 is worth something, but not much as the industry has been migrating to NACS. When I bought my used CCS car, it came with a fast-charge NACS->CCS converter, so dealing with a NACS charger is not that big of a deal.
Requiring 4 ports in general is a more useful standardad. The data does keep a seperate count of dc fast ports, so you could check for 4 ports if you want to, but the pass/fail does not consider non-ccs ports.<p>2 is reasonable.<p>3 is just a sign that you are completly mis-applying the standard. Probably because you are a computer.</p>
]]></description><pubDate>Fri, 06 Nov 2026 19:05:49 +0000</pubDate><link>https://news.ycombinator.com/item?id=50396828</link><dc:creator>gizmo686</dc:creator><comments>https://news.ycombinator.com/item?id=50396828</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=50396828</guid></item><item><title><![CDATA[New comment by gizmo686 in "Chomsky and the two cultures of statistical learning (2011)"]]></title><description><![CDATA[
<p>What reason to we have to believe that LLMs are the closest model we have for how humans learn language.<p>We considered the hypothesis that humans are a general intelligence that figures language out by exposure. The evidence has consistently come back that that is not how humans actually work. A model where humans are hardwired with a lot of knowledge about language, then fill in the gaps based on exposure still appears to be far more accurate.<p>The only thing LLMs show is that it is possible to learn language de-novo.</p>
]]></description><pubDate>Fri, 06 Nov 2026 14:14:11 +0000</pubDate><link>https://news.ycombinator.com/item?id=50392511</link><dc:creator>gizmo686</dc:creator><comments>https://news.ycombinator.com/item?id=50392511</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=50392511</guid></item><item><title><![CDATA[New comment by gizmo686 in "LLMs are making formal software verification feasible at a much larger scale"]]></title><description><![CDATA[
<p>If you have a model of the programming language, then you can make proofs about the source code itself. At this point, your assumption is that you are accuratly modelling the compiler.<p>You can go a step farther and model the instruction set and architecture, and which point you can make proofs about the final binary directly (thereby sidestepping the problem of compiler bugs). At this point, your assumption is just that your physical hardware actually matches your model.</p>
]]></description><pubDate>Fri, 23 Oct 2026 00:24:10 +0000</pubDate><link>https://news.ycombinator.com/item?id=50202968</link><dc:creator>gizmo686</dc:creator><comments>https://news.ycombinator.com/item?id=50202968</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=50202968</guid></item><item><title><![CDATA[New comment by gizmo686 in "Sixteen years after Deepwater Horizon, deep sea life shows few signs of recovery"]]></title><description><![CDATA[
<p>The environment around Chernobyl has already recovered, and is now healthier than it would have been without a nuclear disaster (due entirely to it's function as a human repellent).<p>The area is still bad for humans; but not terrible for the ecosystem.</p>
]]></description><pubDate>Wed, 21 Oct 2026 04:25:49 +0000</pubDate><link>https://news.ycombinator.com/item?id=50176071</link><dc:creator>gizmo686</dc:creator><comments>https://news.ycombinator.com/item?id=50176071</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=50176071</guid></item><item><title><![CDATA[New comment by gizmo686 in "RPKI now protects 70% of internet routes"]]></title><description><![CDATA[
<p>Without RPKI, it would be easy for many governments to cut off Russia's internet access by launching a BGP attack.<p>Assuming the Netherlands was acting unnilaterally, Russian ISPs would be able to coordinate with other ISPs to get back online via other routes. Assuming Russia's enemy wanted to press the issue, it would quickly devolve into a political question of who wants to allow Russian peering, who does not, and who is willing to not peer with peers of Russia (and peers of peers, etc).<p>This has always been the case. The internet is a network of networks; all of which are subject to control by their host governments should they choose to exercise it. With or without RPKI, a partitioning of the Internet for a significant period of time was always going to be a political decision.</p>
]]></description><pubDate>Fri, 16 Oct 2026 02:22:59 +0000</pubDate><link>https://news.ycombinator.com/item?id=50113134</link><dc:creator>gizmo686</dc:creator><comments>https://news.ycombinator.com/item?id=50113134</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=50113134</guid></item><item><title><![CDATA[New comment by gizmo686 in "Iran’s precision strikes picked apart Gulf infrastructure, blunting US war plans"]]></title><description><![CDATA[
<p>> You should look into how many air craft carriers the US has<p>Air craft carriers such as the USS Abraham Lincoln? Which had to remain at sea for 286 days, including 208 consequtive days without making port. During which time, the military suffered a massive logistical failure to adaquetly supply it, presumably due to them moving their logistical base from Bahrain to Diego Garcia. Then, was relieved only by diverting the USS George Washington and USS Theodore Roosevelt to the region, leaving no carriers in the West Pacific<p>> Also there are somewhere between 20 and 28 US military bases in the middle east alone<p>Bases which have been comming under repeated air attack; the extent and damage of which is kept quiete, but appears to be much more substantial than the US military wants to admit. And which are increasingly vulnerable as they loose air defense abilities due to both loss of supporting infastructure, and a depletion of interceptors.</p>
]]></description><pubDate>Fri, 16 Oct 2026 02:12:35 +0000</pubDate><link>https://news.ycombinator.com/item?id=50113028</link><dc:creator>gizmo686</dc:creator><comments>https://news.ycombinator.com/item?id=50113028</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=50113028</guid></item><item><title><![CDATA[New comment by gizmo686 in "Market selloff, rising rates test AI data centre deal boom"]]></title><description><![CDATA[
<p>You can get rich by stealing pennies, if you steal them from enough people. It is still wrong, even if no individual lost enough to care about going after you.<p>If enough people try profiting by stealling pennies, the cumulative losses are significant; and can threaten confidence in and the strenght of the market.</p>
]]></description><pubDate>Thu, 15 Oct 2026 21:24:34 +0000</pubDate><link>https://news.ycombinator.com/item?id=50110498</link><dc:creator>gizmo686</dc:creator><comments>https://news.ycombinator.com/item?id=50110498</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=50110498</guid></item><item><title><![CDATA[New comment by gizmo686 in "Market selloff, rising rates test AI data centre deal boom"]]></title><description><![CDATA[
<p>You would have made a lot of money betting against technology back in 2000 ... assuming you got the timing right.<p>There are a bunch of ways that AI the technology could be massivly successful, while AI the industry crashes.</p>
]]></description><pubDate>Thu, 15 Oct 2026 19:37:39 +0000</pubDate><link>https://news.ycombinator.com/item?id=50109124</link><dc:creator>gizmo686</dc:creator><comments>https://news.ycombinator.com/item?id=50109124</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=50109124</guid></item><item><title><![CDATA[New comment by gizmo686 in "Market selloff, rising rates test AI data centre deal boom"]]></title><description><![CDATA[
<p>When a stock hits the public market, it gets incorporated into broad market index funds (subject to the rules of those funds). If an overvalues stock gets dumped onto the public market, then crashes after being bought by index funds, then holders of those index funds are harmed.</p>
]]></description><pubDate>Thu, 15 Oct 2026 19:32:27 +0000</pubDate><link>https://news.ycombinator.com/item?id=50109039</link><dc:creator>gizmo686</dc:creator><comments>https://news.ycombinator.com/item?id=50109039</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=50109039</guid></item><item><title><![CDATA[New comment by gizmo686 in "ASML's strong orders ease fears AI spending boom is running out of steam"]]></title><description><![CDATA[
<p>If you invested in XLK (State Street Technology Select Sector SPDR ETF) In March of 2000 (just before the .com bust), you would have underperfomred SPY (S&P 500) until June 2026. As of today, the spread between the two is just over 1%, with good reason to believe that we are in another tech bubble.<p><a href="https://www.etftimemachine.com/en/compare/XLK-vs-SPY?start=2000-03-14&end=today&drip=1#etf-cards" rel="nofollow">https://www.etftimemachine.com/en/compare/XLK-vs-SPY?start=2...</a><p>Had you invested in XLK when it opened in 1998, then following the .com crash, it would take until early 2020 to return to parity to the S&P 500.</p>
]]></description><pubDate>Wed, 14 Oct 2026 17:19:02 +0000</pubDate><link>https://news.ycombinator.com/item?id=50091253</link><dc:creator>gizmo686</dc:creator><comments>https://news.ycombinator.com/item?id=50091253</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=50091253</guid></item><item><title><![CDATA[New comment by gizmo686 in "BB(10) > Graham's Number"]]></title><description><![CDATA[
<p>There is some notational subtletly here. Strictly speaking, BB(n) represents the output of the BB function for the input n. In that sense BB(n) is computable.<p>However, the BB function itself is not computable. It makes sense to ask "for what inputs is the BB function computable?". When people talk about BB(n) not being computable, they are really talking about the computability of the BB function at n, not the computability of the integer that happens to be BB(n).</p>
]]></description><pubDate>Wed, 14 Oct 2026 16:16:25 +0000</pubDate><link>https://news.ycombinator.com/item?id=50090085</link><dc:creator>gizmo686</dc:creator><comments>https://news.ycombinator.com/item?id=50090085</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=50090085</guid></item><item><title><![CDATA[New comment by gizmo686 in "BB(10) > Graham's Number"]]></title><description><![CDATA[
<p>The integer C is computable. However, we do not know what integer C is without first computing BB(10); and doing that might be not be possible.</p>
]]></description><pubDate>Tue, 13 Oct 2026 19:41:22 +0000</pubDate><link>https://news.ycombinator.com/item?id=50077484</link><dc:creator>gizmo686</dc:creator><comments>https://news.ycombinator.com/item?id=50077484</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=50077484</guid></item><item><title><![CDATA[New comment by gizmo686 in "BB(10) > Graham's Number"]]></title><description><![CDATA[
<p>Given a constanct C, you can construct a constant sequence f(n) = C.<p>Since f(n) is trivially computable, we know that BB(n) grows faster than f(n). It follows that there must be some n such that BB(n) > C.</p>
]]></description><pubDate>Tue, 13 Oct 2026 16:55:21 +0000</pubDate><link>https://news.ycombinator.com/item?id=50074865</link><dc:creator>gizmo686</dc:creator><comments>https://news.ycombinator.com/item?id=50074865</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=50074865</guid></item><item><title><![CDATA[New comment by gizmo686 in "BB(10) > Graham's Number"]]></title><description><![CDATA[
<p>The BB sequence still has infinitely many entries. Any turing machine will either halt or not halt. For any N, there is a finite set of N-state turing machines which halt, and there is therefore a finite number of steps after which all halting N-state turing machines would halt.<p>There is a point where these values become uncomputable, but they still exist (for some definition of exist).</p>
]]></description><pubDate>Tue, 13 Oct 2026 16:49:30 +0000</pubDate><link>https://news.ycombinator.com/item?id=50074758</link><dc:creator>gizmo686</dc:creator><comments>https://news.ycombinator.com/item?id=50074758</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=50074758</guid></item><item><title><![CDATA[New comment by gizmo686 in "AI models can provably hide arbitrary capabilities"]]></title><description><![CDATA[
<p>Where is the association being stored if not the model? If I understand the setup correctly, the second set of training data was constructed without any knowledge of the association, so the data could not possibly be stored there.<p>This looks like a pretty classic case of stenography. You are able to hide the data in the weights such that the data is not apparent on casual (or even careful) inspection. However, the information is still there, and someone who knows to look for it can still find it, even if they do not know the specific message.</p>
]]></description><pubDate>Tue, 13 Oct 2026 16:27:16 +0000</pubDate><link>https://news.ycombinator.com/item?id=50074354</link><dc:creator>gizmo686</dc:creator><comments>https://news.ycombinator.com/item?id=50074354</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=50074354</guid></item><item><title><![CDATA[New comment by gizmo686 in "Solar panels that plug into your wall"]]></title><description><![CDATA[
<p>It backfeeds onto the grid and becomes the utility companies problem to deal with. This is similar to them dealing with a house that is an unbalanced consumer between the legs.</p>
]]></description><pubDate>Mon, 12 Oct 2026 15:30:17 +0000</pubDate><link>https://news.ycombinator.com/item?id=50057869</link><dc:creator>gizmo686</dc:creator><comments>https://news.ycombinator.com/item?id=50057869</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=50057869</guid></item><item><title><![CDATA[New comment by gizmo686 in "Solar panels that plug into your wall"]]></title><description><![CDATA[
<p>We know how to make inverters that can produce an AC voltage on their own. It is not particularly difficult or expensive. The reason why we do not put those inverters in plug-in solar units is entirely because we don't want them to have that ability.</p>
]]></description><pubDate>Mon, 12 Oct 2026 15:25:03 +0000</pubDate><link>https://news.ycombinator.com/item?id=50057790</link><dc:creator>gizmo686</dc:creator><comments>https://news.ycombinator.com/item?id=50057790</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=50057790</guid></item></channel></rss>