bs7280

I definitely see things as the grey nuanced details they are, but ive made an effort to try to communicate with more decisive language. I've found the extra details confuses the non technical people, undermines or calls into question my knowledge or the authority of my idea to other people like execs. I joked with my boss once, "that's what I get for being honest" when he said some other vp's had doubts on my plan, even though I knew my plan was way ahead of the alternative because I was too detailed and honest.

I think a lot of people assume consciously or sub consciously that everything is a bit of a lie, and one's solution is worse than it's sold as.

Theres a distinction between built more rigorously and tested more rigorously. By many definitions, agent created apps are built more rigorously. They write more unit tests, attempt to handle more edge cases, write more lines of code, create features faster. For certain things, I do think agent created tools tend to be better.

But - I have found that every single model available is lacking in creating a great UX, even when you are aggressively using playwright in your agent loop. IMO this comes down to a fundamental limitation of how LLMs work. If an app has a button that is too small, or white text on white background, or inconsistent layouts, or non deterministic ux state due to network calls (looking at you spotify).... the agent will not notice or care. A playwright driving agent can click on a button wether its 1px or 1000px wide.

But ultimately, a good ux is about designing for the constraints that we as humans have to live with as mere mortals with all our imperfections. This is not easy to do, and requires a lot of craftmanship by the developers as well as buy in from the suites and middle managers.

In your windows vista example, the issue is almost surely the latter of my previous sentance - middle manager buy in. No one at microsoft leadership gives a shit about the user experience its all kpi seeking nonsense. The user having a good experience is not measured in their spreadsheet.

When you think about it, pretty much every single product we use helps address a problem related to our fundamental human flaws. Im typing this on an expensive ergonomic keyboard because my wrists hurt, youtube premium exists because I dont want to waste my precious time on stuff I dont want to see, we use operating systems with GUIs because our mortal minds cant easily comprehend thousands of lines of text in seconds.

When I use a horrible website I have a very intuitive, almost instinctive reaction similar to how I react to physical pain. An LLM will never experience this, so it will never create a ux as good as a human can on its own.

After a significant number of iterations of testing, and improving it. When we wrote code by hand, something like changing an api endpoint or a simple front end change would still involve dozens of "write code" then "retest manually". This would not only get the work item done, but you would often encounter edge cases or other issues you hadn't thought of when writing/reading the ticket.

When AI can one shot the implementation and testing, it likely only gets 1 round of human testing if you are lucky. Then scale this to an entire AI generated project, the ratio of features to manually run tests is astronomical. In the old days this ratio was inverted, and the tool has been battle tested before reaching any users.

I would argue we have a duty to do so, as a counter incentive to the network and streaming execs for making it so annoying and inconvenient to watch anything. I do not pay for YT TV w/ NFL Sunday Ticket, but there was one weekend last year where ESPN and Google were fighting, and no NFL games were shown on YT TV.

There have been times where I will pay to "own" instead of rent a tv show I really like, and then a year later that streaming service loses the rights to it, and so do I.

You just unknowingly pointed out the real problem - the system is setup so that we think we are in on it, but as a wise man said "It's a big club and you ain't in it".

Also "lose their personal savings" is a hyperbolic bad faith argument. If anyone's entire net worth is put into Meta stock, they get whats coming. Second, 93% of the stock market is owned by 10% of the population. The average American has more to gain from holding companies accountable than letting them rape and pillage our society so the richest people can have even more money.

The same reason it took many years for corperate america to get a real productivity increase from computers and the internet, all the old ways of doing things had to be redone. I think the largest companies are least equipped to take advantage of AI productivity gains. Agile no longer makes sense, Org charts as no longer make sense, etc...

It definitely can, but Its easier than ever to fork an older version and modify with claude, or just switch to a different OSS provider. With commercial tools there's no rolling back. In fact, I would pay money to go back to the pre Allen Dye liquid ass IOS update.

Thanks for sharing, now im wondering if they did in fact grant my wish since I messed with this in January. Im lucky my job does not require me to make any pptx presentations.

The way I see it, I am committing more time and hassle to setup my homelab now to save time, money, and hassle later when the SaaS alternate's inevitably do something hostile to me as the user.

Even before AI coding I did not trust big tech companies to make updates that have my best interests in mind. Many of these examples are weird feature or ux changes - like how spotify treats liked songs as a playlist (requiring more taps to like a song). Not the end of the world. But on the more extreme side, my gaming PC's windows installation has died via bad windows update twice now. I don't use it for anything important anymore. It's insane that my claude code driven homelab stack has allowed me to have better uptime than github.

Now in the AI coding era, its obvious to me that alot of the tools I am forced to use on a day to day basis are frequently shipped with slop, causing more bugs and all around jankier UX's. With my claude max subscription, im suddenly able to make tools perfectly tailored to what I want, and I can rest assured I wont wake up and find the equivalent of a bad IOS update.

What's ironic about this is, afaik there's no programmatic way to get a screenshot of a PowerPoint on windows other than opening up PowerPoint and screenshotting it. I only discovered this when I first tried out Anthropics official PowerPoint skill when it came out this winter.

If anything happens from this, I hope it forces Microsoft to address some of its long standing hostile to power users issues so they can stay relevant in the world of AI tooling.

I am subsidizing data centers via my power bill, all so they can over clock them more than really necessary. I use 1 billion claude tokens a month and understand the usefulness of AI, but id wager over 80% of AI spend comes from unsolicited interactions like on google searches, or AI summaries on emails, shitty AI videos on instagram and facebook, endless garbage youtube reels they love shoving down my face...

That is the visible part that leads most people to being anti data center. The average Joe thinks "why is my community, water supply, and money going to this thing I dont want"

What about my comment gave you the confidence to (incorrectly) answer your own question for me?

I want to see large investment in mass produced nuclear power plants, enabling us to do exactly what you are suggesting.

I hate that I am subsidizing the power bills for these data centers.

School start times should be pushed back an hour anyways. Students learn better, especially middle school + highschool age. The only reason we don't is because it conflicts with parents working schedules.

One of my biggest criticisms of DOGE that I have not heard elsewhere is - why can't we have a "Department of Government Transparency". Giving Elon the ability to judge, jury and executioner any congressionally approved spending he wants has its own legal issues, but I do think there is a lot of value in having a massive publicly available dataset on where the money goes.

I envision something similar to a massive sankey diagram like dataset in a public git repo that anyone could access and audit. There is certainly lots of waste, but there is also loads of government spending where the value is not obvious.

A while ago I was driving a loaner car - a brand new top tier Accura rdx and the infotainment system was truly the worst and most dangerously designed I had ever witnessed. It was essentially a lenovo thinkpad type touchpad by the cup holder, and the screen was far away. The first time I tried to use it, it was so distracting that I would of crashed the car if it weren't for the safety features.

My car is a 2018 with car play and physical scroll knobs and buttons, while awkward, I can operate it with my eyes on the road (realistically I can do everything I need from the steering wheel). This weird middle ground carplay was somehow the worst combo of buttons and touchscreen.

Executives will be more afraid of being sent to prison for criminal charges, than having someone else's money get spent on fines. We can do both - increase the fines and set a precedent of arresting executives when their company does criminal things.

Would you be open to sharing or open sourcing the underlying data source? I think there are some very powerful things you could do with this.

For example, I recently discovered this site https://bomwiki.com/ which attempts to make a giant dependency graph of all the parts used to make large machines. This would not be possible without it being a wiki.

There are already mechanisms in place to hold SUV and Truck drivers accountable and track them down (via their license plate).

The problem with ebikes is any unlicensed driver can get one, and go 40mph on a sidewalk without any practical way to hold them accountable.

I live near one of the busiest biking + walking trails in the country, and the egregious disrespect and recklessness of ebike's and scooters is insane to me. Even on the parts of the trail that are split into two or three sections (walking, running, and biking sections) I see people going 20-30mph weaving in and out of walkers. What's crazy is it would be safer if they were operating gas motorcycles, because atleast you could hear them coming.

I think the reasonable middle ground anthropic is trying to achieve is - let the organizations that make the most important and critical software get a head start on cybersecurity before they inevitably allow everyone else the same access.

Other commentors have made good points that these guardrails are counter productive for well intentioned cyber security, because I can't use it to test and harden my own software.

Remember this is the number the Government is measuring and reporting. The "real" inflation that every day people feel in their wallet is significantly higher.

I've been saying pensions should not exist, as they are contradictory to our political system. Some politician 40 years ago can promise everyone the moon, and never force the next generation to figure it out. I'm from Chicago which has a nightmare pension system that's keeping me from ever buying a home in the city I love, because my property tax increases just go to retired people who moved to Florida.

I really appreciate this perspective as It helps fill in gaps in my mental model of where our economy has gone wrong the last 50 years. Unrelated but - I've read an interesting paper on how allowing private banks to create money has led to the infinite profit growth goose chase...

The ability to quickly make an API connection + custom UX means that companies with a sub par website / app but a good data API are more valuable to me than the world class fancy website with a locked down API.

At work I have a great brain dump + TODO list tracker via custom API + MCP into confluence, using confluence pages as the app state. The website is so bloated it takes like 20 seconds to go from "idea" to getting it written down. Im now able to avoid all of that and make ~ MY ~ perfect UX while still being a good corporate employee.

You can go on bridge lift days during the fall + spring and see inside the engine room while the bridge goes up.

I should of emphasized it better in my comment but, the nuance of "read all my emails every day" as a prompt can yield a wildly different solution than "read recent emails every day". The first may literally read all your emails over and over, which is a ton of tokens for little gain. The latter is orders of magnitude fewer tokens with the same if not more productivity.

The difference here is just one word in the prompt, but serves as an example of how just a little but of deliberate thought in one's prompt can yield massive efficiency in outcome.

Whats wild to see both online and at work, non engineers given vibe code like tools will quickly show their ignorance to the importance of deliberate design and need for specific instructions one learns via coding. The "missing semi colon" meme is an example of the intuition we all developed early in our coding careers.

Many people are hoping AI can build and design for them, when in reality the deliberate design choices up front are as important if not more so than before AI.

This is a good example of doing it in a deliberate way that absolute is worth the tokens etc... especially when you are keeping tabs on the cost vs time saved.

The example I was thinking of would be a vibe coder having it "read my emails every hour" only for claude to read the same 1000 emails over and over...

I have ancedotal examples of claude code choosing a solution to a problem that is ridiculously token inefficient.

One example - was giving several agents different sub problems to solve in a complex ML / forecasting problem. Each agent would write + run + read a jupyter notebook. This worked ok, the notebooks would be verbose but it was fine... until one of them wrote out hundreds of thousands of rows to a cell output, creating a 500MB ipynb file. Claude tried several times to read it and it used my entire context limit.

The solution was to prescribe a better structure of doing the world (via CLI analysis scripts + folders to save research results to). But this required some planning, thought, and design work by me the operator.

When I see people spending $10k a month in tokens, I can only assume they are taking lazy hands off approaches to solving problems with the expensive hammer that is claude code. EX: have claude read all your emails every day... the lazy solution is to simply do that, but a smarter solution is to first filter the email body HTML to remove the noise.

I think the important point in the parent comment is "Burning a shit ton of tokens". Openclaw was built fast and loose, making it use far too many tokens for trivial things. I'm confident the next Claw can and will be engineered to be at least 10x as token efficient and more reliable.

People should not be getting into politics for the money. I want senators who are doing it because they want to see a change in the world.