While I'm willing to believe that Anthropic common stock might somehow be a reasonable investment because it might go "to the moon", it seems obvious that this is a high-risk bet. This is shaping up to be a winner-take-all industry. Does anyone really think there is a 90% chance that Anthropic will be that winner?

If you thought there really was only a 10% chance that they'd fail to be in business in a decade, why wouldn't you just buy their common stock and hope for a 100x return? How do the rating agencies justify this degree of confidence in Anthropic's ability to repay these debts? Or do I somehow misunderstand what these ratings mean?

That is, I claim that while it is no faster to execute a comparison against zero, it still can be faster to finish a loop at zero. This is because internal to the processor, the comparison and the branch are two separate ops. The comparison essentially does an arithmetic operation that sets a flag but throws away the result, and then the branch acts on it.

But if you finish at zero, the flag you are using is already set by the decrement, so no additional comparison op is necessary. Now normally this doesn't matter, since the processor is multiscalar with sufficient execution ports to fit in an extra comparison with no increase in total cycles spent. But in a tiny and highly optimized loop, occasionally you can get about 20% gain from having one fewer op to execute.

Getting this to actually happen can be complicated because modern processors do op-fusion that can "fuse" the CMP and branch back into something that can be executed on a single port. But DEC-JNZ can be fused too, which still gives it an advantage. More details on this here: https://chipswiki.org/Macro-operation_fusion

Now does anyone want to finish the "ground speed readout" routine (https://theaviationgeekclub.com/sr-71-blackbird-pilot-tells-...) by pointing out actually how with modern processors I'm actually wrong too?

If you've come up with something better, I'd encourage you to contact Daniel and see if he's interested in helping you write it up into a paper. He's friendly, and may be better positioned to get the optimizations into mainstream compilers.

Having grown up somewhere where about half the school teachers took the week off for deer hunting, I'm pleasantly surprised that there exist places where this might still be the case even for tech workers. Would I be guessing right that you are describing somewhere in Wisconsin?

If such tests exist, it would seem like they could go a long way to increasing the safety of home distilling. If no such test is possible, maybe it makes sense to regulate very tightly?

In his defense, this is written on his personal blog where (quite possibly) many readers already have some reason to trust his advice. And while I'm not sure what counts as appropriate credentials, on his linked HN bio (https://news.ycombinator.com/user?id=bookofjoe) he claims to be a "Retired (38 years in practice) neurosurgical anesthesiologist". He also links to a lot of Google Scholar papers on medical topics for which he is a co-author.

(Oops, beaten by Dan, but I'll leave it up anyway. While I agree with Dan that you might have phrased it a bit better, I did think it was a fair question.)

First, the numbers. My searching says that for some particular lake in Nebraska, "Average annual gross lake evaporation ranges from about 48 inches to 54 inches". Let's call it 4 feet per year, which means 4 acre-feet of water evaporates per acre per year. Trusting some numbers someone gave in a comment earlier, the data center in question used about 40 acre-feet per year. So 10 acres of lake would evaporate about the same rate as the data center. At a very rough guess, there are something like 200,000 acres of non-seasonal lakes in Nebraska, so this is indeed a tiny percentage.

Next, the relevance. The lakes are more or less in a steady state. But if you were actually to double the evaporation rate for a 10 acre lake, you might quickly end up with a muddy pit instead of a lake. The fish would be gone, and the users of that lake would have good reason to complain. If lake was 10 times larger, so that you were only increasing the evaporation by 10%, over time you'd still probably have a major effect on the shoreline and any infrastructure there. What used to be a useable pier might look over a mud flat rather than the water, and launching a boat would be impossible until you built a new one.

But the real issue is that water use isn't actually fungible. Water continuing to evaporate from a lake at the rate it always has definitionally doesn't make people's life worse than it was. But suddenly pumping of the same amount of groundwater might be really bad for the neighbors who previously were using the same water to irrigate their fields. And trying to double the output of a small municipal water system might (at least in the short term) make things much harder for all the existing users on the system. This doesn't necessarily mean it shouldn't be done, but it's a harm that should be considered.

That's my good faith attempt to answer you. In return, I have a quick research question that I think is essential to understanding whether or not data center water usage is a big deal or not, and that I haven't seen mentioned yet in the comments: Where is this water for these new data centers coming from? Is it from new wells they drill? Existing irrigation wells they buy and repurpose? From the local municipal water supply? Rainwater collection? Diverting a nearby river?

I feel like there's no way to answer the question of whether data center water usage matters without considering the actual local impact, and no way of determining the impact without considering the specifics of the source. But somehow I haven't seen any mention of this on either side of the debate.

To the flaggers: flag comments that you think violate the rules, but don't flag something just because you strongly disagree with the claim it makes.

Comments URL: https://news.ycombinator.com/item?id=49725404

Points: 71

# Comments: 19

For reference, did you mess this up yourself or did it require AI?

  The Milk-and-Water School 
  Alas! she would not hear my prayer!
  Yet it were rash to tear my hair;
  Disfigured, I should be less fair.

  She was unwise, I may say blind;
  Once she was lovingly inclined;
  Some circumstance has changed her mind.


  The Strong-Minded or Matter-of-Fact School 
  Well! so my offer was no go!
  She might do worse, I told her so;
  She was a fool to answer "No".

  However, things are as they stood;
  Nor would I have her if I could,
  For there are plenty more as good.


  The Spasmodic or German School 
  Firebrands and Daggers! hope hath fled!
  To atoms dash the doubly dead!
  My brain is fire--my heart is lead!

  Her soul is flint, and what am I?
  Scorch'd by her fierce, relentless eye,
  Nothingness is my destiny!

I liked your piece, and agree with almost all of it, but I'm surprised by your faith in the accuracy of Pangram at detecting AI writing. Is your faith based on testing it with lots of writing of known origins, or are you just saying that it reaches the same conclusion that you do as a talented human?

In particular, I wondered if you have tried running all of your own writings through it to verify that it thinks you are human. I was struck by Freddie deBoer's recent piece where he did this and said it often failed: https://freddiedeboer.substack.com/p/i-wouldnt-say-pangram-i...

What percentage of false positive rejections would you find acceptable? Would you accept this even if it forced you to change the way you write?

My problem with the article was that "patch" doesn't seem to add any useful information. Unless he was using a restrictive definition intending to exclude things like crossover cables (which still work), he could have just as well said "transfer files over a blue Ethernet cable" with equal precision.

If he meant to say that at this point in time "you can use _any_ ethernet cable to transfer files directly between two computers", I think it would have been better to say that rather than appearing to add a restriction on the type of cable required.

Traditionally (pre-2000?), one had to use a special "crossover cable" to do direct connections like this, but apparently modern Gigabit ethernet adapters are able to detect this situation automatically?

Hopefully you realize that it's not actually introspecting into its previous thought process, and instead it is just making up a plausible after-the-fact explanation that matches the format of similar explanations found in its training set?

I too often find these explanations enlightening, but please don't forget that you need to read all such responses as those of a pathologically lying psychopath desparate to trick you, not as a trusted expert who actually understands its internals.