From my experience new models are slower and use more tokens even on questions which gpt 4 answered correctly. It is mostly because newer models tend to be more verbose (even with prompt requesting short answers).
I cannot stand writing style of this page. I have read almost half and the most important thing they say is, that I will learn later. Ok, I will never know.
Exactly this is my problem with all AI tools. I want someone else to create working tools for me so I can focus on my product. It is the same with other tools. I do not want to spent huge amounts of energy and time to setup my IDE, operating system or desk layout. I guess it is too early to have that now.
It depends if duplication is accidental or real. I.e. if two taxes are using the same formula, it is accidental. If you use the same physic formula on multipla places, it is real duplication.
Exactly, I stoped using all social media and I feel way better and my screen time went significantly down. It is harder to communicate with some people (have to use email or sms) but it reduced mostly meaningless small talk we can do in person.
Why do you think we lost free internet right now? In my opinion we lost it when all corporations switched to mass surveilance. Now we just stop pretending that social media corporations are not responsible for anything.
LLMs are getting so expensive that for many people are already unavailable (except for subsidized subscriptions). This might just accelerate general unavailability of frontiers models to general public which would happen in near future anyway. Frontier LLMs are just switching to B2B only earlier.
Because writing huge amounts of code is easy for humans too. Agents already proved that they can do it. But are agents able to maintain it? I do not know and unless I know for sure, I am not fully committing to AI generated code.
i.e. I am able to write about 1k lines of code of "acceptable" quality per week. Which means in 1 year, there will be about 5Ok LoC. I am pretty sure, that I would have to spent like 60-80% of time to maintain 1st year code and the rest to make new features in the second year so I would have to hire more people and spent time to onboard them to maintain velocity. All of that are rough estimates, probably overoptimistic and way worse in 3rd year. Good luck doing such estimates with code agents. Even worse if you already have huge amounts of legacy code.
Exactly my experience. I always refactor first myself then delegate boring tasks to AI. It saves me energy, time and also tokens. If code is not prepared for easy implementation agents always fail.
I am self hosting forgejo on my synology NAS. It is easier than it looks. Synology provides me access from the internet so I do not need static IP address. It took me at most 20 minutes to write (copy paste) docker compose file to make it run and another hour to import repositories from github and gitlab. Only maintenance I do is update to new version once a while which takes about 5 minutes. You can set it up to sync repositories back to code forges.
If you do not have a lot of users you can easily set it up too.
I wanted to give money to charity and they have whole form protected by recaptcha. So I would have to allow all my personal information and amount donated sent to google (and agree with google terms for data processing). I have contacted them but they did not understand why this is problem they just wanted to protect themself against bots. IMHO unless those things are not disallowed by antitrust laws we have lost.
I do not understand why this is so unpopular today? I feel like everyone now thinks that basically all of SW engineering is outdated. We are supposed to forget all lessons learned and let agents to go through this? My opinion is to not care who did the job. But we should apply the same standard to human and AI output. I do not buy "we should not look at code". If we should not look at it what we should check instead to have the same control over final product? Because not having control over final product is so stupid right now.
Better than average phone sold today. The only problem might be lack of android upgrades otherwise it is straight upgrade for most people. This is reason why replaceable battery is important. If you leave IT bubble people happily use ancient phones and do not need upgrades if battery is ok and there is space to save new photos.
What I do not like on Zed or electron GUIs is lack of customization. Older IDEs using sdks like Swing, JWT, QT, GTK etc. allowed user to design its user interface using drag and drop. ie compare older IDEs like eclipse or idea and try to create layout which fills screen with information important for you. And then try to do the same with vscode or zed. I like functionality and speed of zed but UI customization is too limited for me. It might be design choice or sdk limit I am not sure.
I have just combined those together. I use astronvim as main editor but when I need IDE I switch to Idea. I use ideavim with configuration as close as possible to astronvim. So text editing is the same for me in both programs. Modal editors are still great and they can do a lot of work that looks like magic for AI era trained developers.