Posts by Roland 2 20 publicly visible posts • joined 23 Jul 2010
This cockup tale is unconvincing and appears to be a smokescreen designed to sow doubt on the plaintiff.
GPS combined with accelerometer dead reckoning offers extremely high accuracy (see Kalman filter on Wikipedia). The accumulated error would be well below 1% per trip, making it more accurate than wheel rotation counting over several months.
Here you go (deepseek-r1:32b):
>>> How much nutrient in an egg, when hard boiled. It has been laid by dog.
<think>
Okay, so I need to figure out how much nutrition is in a hard-boiled egg that's been laid by a
dog. Hmm, wait a minute—dogs don't lay eggs like chickens do. That doesn't make sense
biologically. Maybe the user made a mistake or was confused about something else.
[... 8 paragraphs of introspection about answering a bogus question leading to answering for chicken eggs ...]
</think>
The question seems to be based on a misunderstanding, as dogs do not lay eggs; they are
mammals that give birth to live young. However, if we assume the intended query is about a
chicken egg, here is the nutritional breakdown:
A large hard-boiled chicken egg contains approximately:
[... calories, fat, etc. ...]
Hard-boiling slightly reduces some water-soluble vitamins but preserves protein and fat
content. Cooking can enhance the digestibility of proteins compared to raw eggs. Thus,
hard-boiled chicken eggs are a nutrient-rich food, offering high-quality protein and essential
nutrients beneficial for overall health.
Not bad for a quantized 32B model that fits on my laptop
The way AIS works is that it broadcasts the ship's position, avoiding conflicts with time division multiplexing synchronized on the GPS signal.
Of course the intent is that each ship would broadcast its own GPS position, but nothing stops you broadcasting a false position. Might not be very credible if the position is thousands of miles from where the signal is picked up, but an error <100 miles is enough to provide plausible deniability.
I guess there will be an expanding market for maritime patrol drones to constantly watch areas of interest...
Just tried it (again): it did not have the right information, either pages not indexed, or query mi-interpreted.
You can't even tell why, since the haphazard parrot will always produce an assertive answer that is trained to look right.
I'd much rather prefer "sorry, no results" or "I interpret your query as ..."
Aside from that, totally agree about Google's enshitification.
I switched to Brave.
Having full backups is only the first step.
They also need to be tested regularly.
And you need to have at least yearly full restore to bare metal (to clean rootkits those guys will leave behind) drills.
Short of that, just running regular full backup will only increase you storage costs.
from the article:
" ... the lack of arrests remains a concern. "
" Governments will have crucial roles in the fight against ransomware."
What we have seen over the last ten years is a clear shift from independent criminals (Pirate) to state tolerated/sponsored actors (Privateers), in a handful of rogue states: Russia, China, Iran, N.Korea
Surely putting more pressure on organizations to improve security is going to (marginally) improve the situation, but no medium size business is going to be able to match the challenge of state approved actors. No action from law enforcement will be decisive unless offshore bases for cybercrime are disrupted.
Until the 19th century privateers from North Africa harassed Mediterranean trade, until punitive actions by western powers destroyed or dissuaded the sponsors. That's what's needed right now.
If lawmakers are going to be part of the cybercrime discussion, they need to understand this reality, otherwise we'll just get another layer of regulatory burden while perpetrators laugh all the way from Moscow to Dubai.
Agreed that config/argument parsing should be done by a well tested library, but nothing much to do with CLI arguments vs. json config files.
There are plenty of poor ways to hand parse a config file too, with associated security issues.
Using an adequate library (think argparse in python, or clap in Rust) is not only safer, it also brings good error messages in case of misuse, arguably better that a security hole.
> Who on earth chooses "personalized relevance model" for their data?
You do as soon as you use Google search.
Which is a much bigger concern than what kind of political gibberish you see on twitter, as we all now more or less the the world through the lens of web searches.
BTW, am I the only one to notice that Wikipedia entries have moved down the Google result list in the last months?
Could that be correlated to zero ad revenue from Wikipedia?
Just guessing.
My experience with young software engineers is not having them reimplement known algorithm.
It's failure to recognize they need a known algorithm in the first place!
Like "do a topological sort" instead of a set of heuristics trying to figure out the right order in something.
Or "use a regexp" instead of a 20 line nest of ifs with startsWith, charAt and substring.
good luck detecting that with AI, but worth trying.
I keep seeing endurance figures that seem ridiculous for a HDD.
For this model, the Toshiba spec sheet's exact wording is:
"550 Total TB Transferred per Year Workload Rating"
If this is an endurance limit as for an SSD, this would be less than 0.1 full drive read or write per day!
Or is this a prerequisite for the 2.5 Mh TTF ?
If it is, it means : "this drive may last, provided you don't use it"
Can some bright mind explain?
As as shop that did the move from tape to disk 6 years ago, I do confirm that the operational experience here is that disk is more reliable than tape. We no longer experience things like "not this backup, it's corrupted/incomplete/eaten by the dog, take the previous one", or wait hours to apply incremental dailies.
The BER issue (assuming there is one) it totally hidden by RAID 6 + replication.
As of cost, the whole driver for the project back then was cost reduction.
> Including thin provisioning is rediculous.
I think it's fantastic. Allows the customer to upgrade from 4:1 to 400:1 in a second by just thin provisioning more non-existant space.
Seriously, any vendor claim is legitimate if it documents exactly what ratio can be attributed to what reduction technology. Like Compression 2:1, dedupe 2:1 Thin provisioning:1.5:1 Overall 6:1
You may then get a quick figure for a given use case. Like database of unique png images, 20% thin provisioned: ratio = 1.2:1
I'm not sure Pure is much more transparent in that respect.
If your disk has 5ms latency and array/network has 1ms, you see 6ms of latency from the server. Let's say some amount of flash will give you a 80% hit rate, with 10 microsecond latency.
If you place the cache near to the disks, you get 2 ms average latency from server.
If you place the cache on PCI-e on server you now see 1.2 ms latency, and you alleviate 80% of the traffic on the array on a read workload, so you end up also saving on array iron.
While tackling OS integration, what would really be welcome would be a cacheFS integration, so that NFS mounts to Linux or VMware hosts could hide most of the network latency through a 620 Gb cache.
This would be a huge application performance and network traffic win, while being totally transparent to applications.
(right now our applications cache hot data manually to the Fusion-IO cards. not transparent!)
The Register 
Biting the hand that feeds IT
Copyright. All rights reserved © 1998–2025
