HN user
Founder and President of BV Software. marcus@ bv software dot com
Even if they could "do it better" it is still locked into a single device/vehicle and can't match the eco-system that Apple and Google have developed across a family of devices and services. In order to really compete, they would need to enter markets already dominated by big tech companies.
10MB for the Google Homepage! 44MB for Facebook Homepage! I have not been paying attention to website bloat. Wow, and people were annoyed when a site had to download a whole JQuery library for a single function.
Anthropic clearly doesn't understand that customers see their brand as "Claude", Google's brand as "Gemini" and OpenAI's brand as "ChatGPT." They have so many plans and exclusions that they risk customer confusion. I was surprised when I was pay $200/month for Claude Code, finding it super helpful, and then I had to pay separately to get API access for an experiment. Why are so many parts of "Claude" separate from each other, especially on a $200/month subscription.
Anthropic better get this sorted out with a proper product manager and marketing or they risk customers jumping to easier to understand platforms that are good enough.
NASA "Force?" It sounds very similar to Space Force and Air Force and adds a militaristic tone to NASA. Maybe that is the intent. I know that NASA and the military are closely linked but the general brand of NASA is a the science-focused civilian side while something like Space Force would be the military side.
I've built a few transportation simulations where I started out with pathfinding methods like A* but the compute cost doesn't scale well with 10,000 or 100,000 agents running around. Pre-computing flow fields for common map destinations is one of those areas where you trade off storage for compute. The agents just look for the signpost telling them "this direction to destination x" instead of actually calculating a path.
Also, have massive layoffs every few months just to keep people on edge. AWS wants people to leave with RTO and badging policies, comp range shifts lower unless you have year over year ratings, and an obsessive push to force AI into every process. Top talent is leaving and will continue to leave AWS.
I made a decision to reframe "news" as the "fear network" so my brain had that context as I found out the news of the day. The article had an interesting perspective on information diets contributing to overall pessimism.
Marco Rubio needed this for his presidential run in 2028. Does this mean that Putin will look the other way for Maduro as long as Trump looks the other way when Putin captures or kills Zelenskyy? Have they officially agreed to divide the world as spheres of influence?
Learning to play individual notes from sheet music only helps you learn one song. The breakthrough for me was thinking in musical structure.
- There are 12 keys on the piano just repeated - A scale can start on any of those 12 keys - The "home" key of the scale get labels with a roman numeral one, I - The rest of the keys in the scale get roman numerals ii,iii,IV,V,vi,vii - The I,IV,V are all upper case to represent major chords, the lower case for minor chords - Most pop songs use I,IV,V from a scale. In C-major scale, C, F, G major chords. - You can start on any key on the piano and if you play the same sequence of I, IV, V, you'll get the same song, just transposed into a different key. (the scales are slightly different due to even temperament for advanced ears)
So, learn songs by the chord structure first. It is easier to remember and you'll start to recognize patterns in other songs and unlock them faster.
Color grading is pretty interesting. Please share good resources if you find some.
Technical:
- Understand how to deploy teams of agents effectively to accomplish significant goals
- Learn ECS/Dots in Unity to scale a system to hundreds of thousands of actors
Non Technical:
- Improve people management skills for leading technical teams with a target of helping each person grow in 2026 and level up the team
- Automate more of my personal finances to gain leverage from systems instead of hoping I make good decisions consistently
I do the same. It lets you see exactly what the LLM is using for context and you can easily correct manually. Similar to the spec-driven-development in Kiro where you define the plan first, then move to creating code to meet the plan.
When you get the answer you want, follow up with "How could I have asked my question in a way to get to this answer faster?" and the LLM will provide some guidance on how to improve your question prompt. Over time, you'll get better at asking questions and getting answers in fewer shots.
Doesn't MacOS already render as 2x resolution and downsize in order to do font smoothing? Looks are important to Apple and I think they are willing to add custom hardware capable of handling these type of effects without killing battery life and CPU cycles.
There was a time when everyone hand-coded HTML. Then came Macromedia Dreamweaver and Microsoft FrontPage which promised a WYSIWYG experience. No one would ever need to "learn HTML and CSS" because the tool could write it for them. Those tools could crank out a website in minutes.
When those tools created some awful, complex and slow output, only the people who knew HTML could understand why it wasn't working and fix things.
Vibe coding is in a similar place. It demos really well. It can be powerful and allows for quick iteration on ideas. It works, most of the time. Vibe coding can produce some really terrible code that is not well architected and difficult to maintain. It can introduce basic logic errors that are not easily corrected through multiple prompts back to the system.
I don't know if they will ever be capable of creating production quality systems on par with what senior engineers produce or if they will only get incrementally better and remain best for prototypes and testing ideas.
Self-learning opens new training opportunities but not at the scale or speed of current training. The world only operates at 1x speed. Today's models have been trained on written and visual content created by billions of humans over thousands of years.
You can only experience the world in one place in real time. Even if you networked a bunch of "experiencers" together to gather real time data from many places at the same time, you would need a way to learn and train on that data in real time that could incorporate all the simultaneous inputs. I don't see that capability happening anytime soon.
I just read Technofeudalism by Yanis Varoufakis which has an interesting perspective that "cloud capitalism" is replacing traditional capitalism and competition. A few players are assembling their own fiefdoms inside dominant web/mobile platforms. https://www.penguinrandomhouse.com/books/751443/technofeudal...
The internet doesn't have a clear, simple, micro-payment system that would allow people to reward value, so instead we have an attention based system where the number of likes and followers grants social status and financial opportunity.
English and other languages come with lots of ambiguity and assumptions. A significant benefit of programming languages is they have explicit rules for how they will be converted into a running program. An LLM can take many paths from the same starting prompt and deliver vastly different output.
Looks like an opportunity to sell a price negotiation AI to all consumers. Get paid on both sides of the AI wars.
One thing these models are extremely good at is reading large amounts of text quickly and summarizing important points. That capability alone may be enough to pay $20 a month for many people.
Some cruise lines, like Norwegian, have specific solo travel fares and cabins and even social events just for solo travelers
The reproduction cost for the 2nd copy of media is near zero just like software. Handmade or customized furniture is more expensive because it takes more labor for each copy. With media, the cost is fixed, even if it is large. Once the first version of handmade media has been created, the owner is incentivized to get as much value from it as possible. The optimal demand curve is probably not a few rich people paying as much as possible.
There is no regulation or legislation that is telling Amazon they can/can't or should/shouldn't post this information. Amazon appears to be responding to the wishes of the President out of fealty or fear. The threat of a negative response by the President is enough to change the behavior of the company.
I've been thinking about the need for something like this. Another person told me the idea was like Neo in the Matrix saying "I know Kung Fu." I think the most powerful idea is to create unique bundles of context for specific use cases. Context A + Content B = best context for Situation C.
Idiocracy in real life. The AI told me to drink Brawndo. Why? Because it has electrolytes. My AI said electrolytes are good.
HR departments are constantly looking at "market rates" for jobs which is a fancy way of saying they share salary data or get it from ADP and have much more information about what people are willing to accept.
At hiring time, they are willing to pay market rate (or some percentage of market rate) to get people in the door. Once you are employed, they don't care anymore and will let excellent people slowly fall behind market compensation with 1% to 2% raises.
When those employees get frustrated and leave for 20% bump in comp, the companies seem fine replacing them with a new hire at market rate. So now, they have a new employee making market rate, they have to train the new employee for months before they are productive and they've taken on the risk of an unknown vs. just giving the existing employee a raise to market rate. It doesn't make sense unless you want to telegraph the message that employees are fungible and you don't really care about people.
Ironically, if you are a contractor going through a staffing agency, when you are "fired" from a contract, you generally still get paid something by the agency and get "bench time" to train up your skills or work on internal projects. Sometimes contractors have better job security from a paycheck perspective.
If you do a lot of video calls, you can also consider how to light yourself so that you show up well on camera. The traditional three-point lighting setup is a good place to start. A key light which is the brightest, main light source, a fill light which is from the opposite side to soften shadows and a backlight to help you stand out from backgrounds.
Virtual video sound is rarely full duplex meaning the first person to speak blocks others and you have these really awkward pauses and latency that interrupt natural communication. Then, add in the lack of body language cues that are difficult to see over video and the pace of communication becomes slower. Ideas don't easily bounce around a team because the technology creates a one-person-at-a-time framework.
This museum, just outside of DC, is worth the visit if you enjoy encryption and learning about code breaking. They even have a pair of enigma machines that let you encode a message on one and decode on the other. It is small but packed with some unique artifacts including some of the earliest super computers.