Tokenization doesn't explain this kind of a mistake. If you ask about "djsjc" you would get the proper answer of 5. The claim that this is performance trade-off does not hold.
I cannot edit the question, but would like to say that I'm extremely impressed by ChatGTP and entire question was an honest curiosity about the limitations of it. It is strange that many responses are about blaming my question and example as just wrong and not about the limitations of the ChatGPT model (admirable anyway).
Our company tracks all the laptops and mobile phones. On onboarding we send them to the programmer and during offboarding we collect them back. Shipping label are always generated by us. At the end of life we sometimes give away devices to their users after resetting them. In case of laptops we mostly have a second life for them as long as they are fully functional - they are used by administration, as a second device for projects as needed, etc. But we do not plan to wind down anytime soon, so this might be a different story with a startup going bankrupt. In particular once the person responsible for administration is no longer with the company ;)
This article doesn't touch much about why plumbing is hard. I'm from Poland so I'm not only IT but also a plumber ;)
Plumbing is hard because it is not forgiving. It's as binary as IT except you can learn the outcome with some delay, once you learnt about a damage caused by a leak. Either you do a pressure tests right or repair can be expensive. And bugfixing is always tricky.
Water also goes down whether you like it or not. Think about all possible leaks inside the shower cabin. Or what is even more impressive that under a pressure the water goes everywhere possible.
Plumbing is similar to electrical engineering, except it usually doesn't kill immidiately (though working with gas is tricky anyway) but requires similar strict mental model to do right.
And when you see a plumber it seems like this person is just a physical worker. So work status misconception must be leveled with money...
I had a funny experience a week ago. One night working on a hobby web project it took me three or four hours to debug something, I finally got to bed around midnight thinking "boy, programming is hard".
The next day at work we had to find a broken heat wire in a tiled bathroom floor, running 1000 volts through the wires to try to fuse the broken wire, then heating the floor up and searching with heat-sensitive paper overlays for the likely broken spot, then breaking the tile with a hammer and digging the wire out of the mortar bed. After we found it I thought, "I'd rather hunt software bugs".
I once had to jackhammer a 4 foot wide, 26 foot long, 4 foot deep trench in my basement to replace the sanitary sewer in my house. It was old terracotta pipe and had tree roots growing into it and eventually blocked the flow. It was doing that that helped me be so thankful to have an office job. I also had to lug all the rubble upstairs in 5 gallon buckets.
I didn't. It's not good practice to fill in the trench with rubble (rubble is the big chunks of concrete with re-bar still in it, not the dirt). I left the dirt down there, along with any smaller chunks of concrete, anything smaller than a golf ball or so. Good practice states that you should fill the trench directly around the pipe with gravel which i brought down fresh, compact that and then fill dirt on top of that, which I had left down there, and then finally with newly poured concrete and re-bar. Filling in the trench with the rubble would potentially damage the new pipe I just put in. It actually wasn't bad bringing in the new gravel and sacks of concrete because gravity does the work for you. We just opened one of the windows and put two 2x4 boards to make a ramp and slid the bags down that. Just had to carry it from the truck to the window.
Good question, I don't know. The technician on site didn't have one. I guessing the camera would be faster, but the result image about the same. The nice thing about the paper is you easily can check multiple spots at once, so it's faster to divide down to the problem area.
Very cool! The tech did have some device that could measure the distance to the fault when tied into wire upstream, but I'm not sure if it was a TDR, I think he was making more rule of thumb calculations based on the characteristics of the wire. I would say though that between his voltage box, amp box, and general test tools he had the coolest tech on the jobsite in general.
My complaint is that plumbers frequently do poor work for the money.
We had some come and install an instant water heater and they cut an ugly hole in the side of the house without much thought.
At one office I worked in they called Roto-Rooter (a non-union franchise that is likely to wreck your pipes and require a call to the union plumbers afterwards) who claimed that we'd flushed a condom down the drain (very hard to believe) and wrecked the pipes so we had to call the union plumber.
Another time the sink wasn't running so we called the union plumbers, they unscrewed the aerator from the faucet, saw some crud come out and the water run and left in triumph, sure of their ability to outthink a group of mere computer nerds.
Us computer nerds were sitting at the faucet immediately after that, running it and talking about it. The now aerator free faucet clogged up again within 2 minutes of the plumbers leaving.
> My complaint is that plumbers frequently do poor work for the money.
Considering the quality of many expensive website and software implementations I've been required to use throughout the years at various jobs, this problem is not unique to plumbers.
I live in Poland. We are recently being recognized as a developed economy most often. Still, we earn 1/3 in dollar terms per Capita compared to the US or Germany.
I think 4 days week would result in people having two jobs regularly. I wonder if the situation is simalar with basic job workers in more advanced economies.
One interesing point is that people in my IT company were willing to put 5-10% of their income in exchange for work from home. But this is IT.
I think that the right way to go is not to reduce the number of hours but just allow people to work 4 days per week with hourly rate intact. This might be a very welcomed option by many people.
Also - signing up for this in the high inflation time might work well instead of raising compensations. So this is a good time to carry such experiments.
I am also from Poland and I would be willing to sacrifice 20% of my salary for 1 extra day off per week. I would certainly not look for another job. Also in my employment agreement there is a rule stating that I need my current employer approval before I get a side gig.
I even expect that this would have a rather minor impact on my work (like 10% decrease). I think for jobs like writing CRUD screens for entire week the productivity drop may be more significant.
Actually this inspires me to ask my employer this January for such offer.
Disclaimer: Single person, no children or other commitments.
> Also in my employment agreement there is a rule stating that I need my current employer approval before I get a side gig.
I do not think this is lawful. In our agreements we only prohibit doing a work for our competitors or directly competitive work. If you like your employer then do not come up with this with them, but stay informed of your rights.
> were willing to put 5-10% of their income in exchange for work from home.
Why? The company is going to use your home as an office and they should be paying that 5-10% extra.
Why business paying corporate landlord for office space is okay, but when Joe Public offers his own lowly place then it's a no no?
When people don't recognise their value, they are prime for being exploited.
We did the experiment to check if people are serious about WFH. There is a lot of work to make it happen inside a company. In particular to treat remote people as first class citizens. WFH was also a choice after covid and not necessity.
With a pay cut we are sure people are serious about WFH and not only demanding.
>Still, we earn 1/3 in dollar terms per Capita compared to the US or Germany.
It's not a meaningful comparison without cost of living adjustment. You're not paying 3000 dollars for a one bedroom apartment either. In fact last time I visited Poland I think a place in central Warsaw was like 600 bucks. Poland is exceptionally affordable.
For those on tech wages, which have reached near parity with the west, or work remotely for western companies, yes. But not everyone in Poland earns tech wages.
Congratulations for your contribution to PM, creating and sharing your toolbox. I wish I will be able to accomplish releasing such work for my process as well at some point in time.
Building a team is always a turn-key project. Tools are a part of it.
Time will tell if this is going to be a widely adopted framework or you are able to inspire several people with your ideas and some of them get adopted in other tools. Even if it only allowed you to better understand your ideas it is worth it. Once again - congratulations and continue inspiring others!
Reed-Solomon is the foundation of today's computing. It is used in data storage (hdd, ssd) and in data transfer protocols. It allows for building of a reliable system on top of an unreliable real life fenomens with desired level of certainty. This is so incredible tech that once implemented we can just forget about it in the higher level abstractions.
My story about ipv6 is from a travel. I connected to a friend's wifi. It worked for some time and then a strange thing happened. I was able to google but I was not able to connect to my company's services. I debugged the cause and it looked like the router received ipv6 address but not an ipv4. There was some tornado going around that might something to it.
This was the first time I felt like being on the ipv6 network. I also wonder why I was not able to access to ipv4 servers, I assume that in India people have no problem accessing ipv4 only networks.
Anyway - this was the first time I had to deal with ipv6 :)
On the business side. One ipv4 costs us $3/mo. We never had any problem with this, didn't feel any business reason to upgrade and in particular we (as most of the companies) do not have any permanent test of ipv6 connectivity going on. So even if this had been setup at some point, we would never notice a regression here.
So - unless our customers started asking for ipv6, government pushed some regulation or the address' prise would go up - I do not think we will do anything about ipv6, like most other companies.
The $3 are an early warning signal. You can start upgrading slowly on the cheap when you have all the time in the world, or you can do it for much more money and higher risks come crunch time, when you "see a business reason."
I don't have an axe to grind with IPv6 in particular, but I wish managers who consistently make short-sighted management decisions when the writing is so obviously on the wall, would face unemployment.
Generally, IPv6-first networks use shared IPv4 addresses for connectivity to legacy servers. Think CG-NAT but slightly more optimised.
I don't know why a tornado would only knock out IPv4, seems to me that the prefix administration for IPv6 should run on the same location as the DHCP administration for IPv4.
I would've switched most of my systems over to IPv6 by now if it wasn't for my workplace using Ubiquity hardware that still lacks IPv6 hardware acceleration, forcing the admins to disable it or face unnecessary network slowdowns. I like to be able to use my password manager and such on their WiFi just in case.
Indian regulation forcing IPv6 availability may lead to limited IPv4 availability in one of the biggest developing economies in the world, which may lead to business reasons for switching over. I can only hope, we've been stuck with IPv4 for way too long.
Sweden is going to join NATO soon. So do not expect VPNs operating from their teritory and under their law to be any different then the rest of Europe, as it was before.
Meh, I generally agree with the laws of my country and of the EU, so I don’t really care if they can obtain a warrant to wiretap me. I prefer to live in a society where law enforcement can be effective (with checks!) than one where bad things can happen with zero negative consequences.
If you plan to spend 1M somewhere to expand then buying a report for $1000 seems OK if you only believe the author did some research. In particular a research better than your assistant spending 40h on it which is more than $1000.
I cannot edit the question, but would like to say that I'm extremely impressed by ChatGTP and entire question was an honest curiosity about the limitations of it. It is strange that many responses are about blaming my question and example as just wrong and not about the limitations of the ChatGPT model (admirable anyway).