sReinwald · HackerTrans

4 дня назадdiscuss

> at a breakneck pace for no apparent reason, lately.

This isn't surprising to me at all.

The World Cup is on, and it draws attention away from politics. This has been a pretty common observable pattern for as long as I can remember.

sReinwald

3 месяца назадdiscuss

You clearly haven't played WoW in a decade, judging from this comment.

> Even if you buy the expansion just to get the "feel" on how it was, it's impossible.

You don't buy previous expansions after a new one launches - they roll into the base subscription. After Shadowlands released, buying BfA separately wasn't even an option.

> MMOs like GW2 and even SWTOR do it way better

“Keep every expansion fully relevant forever” sounds nice until you think through what that actually means for an MMO like WoW. You would either fragment the player base across twenty years of content or turn gearing and balance into a complete circus.

Imagine your best-in-slot trinkets from the current raid and Siege of Orgrimmar, your tier set from Dragon Soul, the weapon from Hellfire Citadel. Try organizing a group when other classes need gear from Icecrown Citadel, The Everbloom, Argus and Ahn'Qiraj.

The point of "current expansion content is relevant" is that it funnels the player base into a fairly narrow area of the theme park. That is important, because if you spread out the population over 20 years worth of content, you risk making the world feel incredibly empty, which is a death sentence for a theme park MMORPG.

Blizzard’s actual approach is much more sane: older content comes back in controlled ways. Timewalking reopens older expansion content with scaling and relevant rewards, and Mythic+ seasons already rotate older dungeons into the current endgame pool. Midnight's Season One, for example, features dungeons from Wrath of the Lich King, Warlords of Draenor, Legion, and Dragonflight.

sReinwald

3 месяца назадdiscuss

Yeah, totally valid for things like D&D - I split that out too, for the same reason.

Work and homelab overlap constantly for me though. I make heavy use of periodic notes, so even something like recipes would get linked in weekly logs for meal planning or tracking. The connections don't have to be deep to be useful - sometimes it might just be 'I made this thing on this day' and that's enough context to be worth having in one place.

sReinwald

3 месяца назадdiscuss

I keep one vault for almost everything because the cross-domain links are where the interesting stuff happens. I studied sociology, political science, and media & communication before moving into IT, and I'm still actively interested in all of it and especially how these topics intersect. Where does a note on Baudrillard's concept of hyperreality go? The social science vault? The IT vault? Both? With a single vault I don't have to answer that question - the note exists, and I can use it to make connections to AI, social media, or cultural trends.

Multiple vaults break this. You end up duplicating notes, or linking across vault boundaries (which adds friction), or just forgetting where you put something. That's trading one of Obsidian's strongest features for... tidier folders.

The one case where I do split things out is genuinely isolated project knowledge - like a D&D campaign I used to GM. 99% of those notes have zero connection to anything else. Separation makes sense there precisely because there's nothing to link.

One more thing: a lot of people solve the "big vault gets messy" problem with elaborate folder structures, but Obsidian's search (especially with Omnisearch) makes most of that unnecessary. I don't really organize my notes into careful hierarchies. I write them, link them, and search for them. The mess is fine. I am one with the mess.

sReinwald

3 месяца назадdiscuss

Third-party MCP servers create at least two different security problems. One is prompt/context injection through the tool output. The other is the much more conventional risk of executing untrusted code with transient dependencies on your machine (which is how the recent litellm compromise was discovered).

Containerization only helps with the second one, not the first, but that still matters. If you’re going to run random third-party MCP servers, isolating them from your host and any sensitive local data is still an obvious improvement over no isolation.

sReinwald

4 месяца назадdiscuss

"It's not a complaint about water. It's a complaint about the wetness."

If capitalism requires constant vigilant government intervention to prevent monopolistic practices, anti-competitive markets, and asymmetric bargaining power, then it seems to me that this is absolutely a complaint about capitalism. If anything, your comment just makes the indictment more damning.

sReinwald

4 месяца назадdiscuss

Any single LLM-edited comment reads fine in isolation. The uncanny valley kicks in when you read thirty of them in a row and they all use the same "it's not X, it's Y" construction. The problem isn't that LLM prose sounds inhuman but that it sounds like one human writing everything. Homogeneity at scale becomes an uncanny valley.

This happens because most people just paste a draft and say "make this better" with zero style direction. The model defaults to its own median register, and that register gets very recognizable after you've seen it a hundred times.

But this is a usage problem, not a fundamental one. I actually ran an experiment on this — fed Claude Code a massive export of my own Reddit comments, thousands of them across different subreddits, and had it build a style guide based on how I actually write and argue. The output was genuinely good. It sounded like me, not like Claude. The typical Claude-isms were just about gone.

I wouldn't expect most people to do that. But even a small prompt adjustment makes a real difference. Compare "improve this email" to something like:

    Your job is to proofread and edit the following email draft. 
    Don't make it longer, more formal, or more "polished" than it needs to be. 
    Fix anything that's actually wrong (grammar that changes meaning, tone misreads). 
    Leave stylistic roughness alone if it fits the voice. 
    If the draft is already fine, say so.

That preserves voice way more than the default "Hello computer, pls help me write good" workflow.

But if we're being honest, most people don't care about preserving their voice. They need to email their professor or write a letter to their bank, and they don't want to be misunderstood or feel stupid.

sReinwald

4 месяца назадdiscuss

> Overall this was a worthwhile assist. I believe (totally understandable) anti-AI animus is coloring a lot of these replies.

That, and hindsight bias. People know the second version came from an LLM, so it's automatically "flat." But if that edited comment had just been posted, nobody would've blinked. It reads fine.

IMO, there's a distinction worth drawing here: "AI edited" and "AI generated" are not the same thing. If you write something to express your own thinking, then use an LLM to tighten the phrasing or catch grammar issues, that's just editing. You're still the one with the ideas and the intent. The LLM is a tool, not an author.

The real failure mode is obvious enough: people who dump raw model prose into threads without critical review. The only one who "delved into things" was the model - not the human pressing send. That does flatten everything. But that’s a different case from a non-native speaker using a tool to express their own point more clearly.

The "preserve your voice" argument also smuggles in a premise I don't necessarily share - that everyone should care about preserving their voice. I'm neurodivergent. Being misunderstood when I know I've been clear is one of the most frustrating experiences there is. For some of us, being understood sometimes matters more than sounding like ourselves.

sReinwald

5 месяцев назадdiscuss

Interestingly, when I apply the "simply repeat the prompt" technique [1], Sonnet 4.6 on the website got it right every time, both with and without extended thinking.

Not repeating the prompt got a mix of walk and drive answers.

I love how prompt engineering is basically techno-alchemy

1: https://arxiv.org/pdf/2512.14982

sReinwald

5 месяцев назадdiscuss

You are essentially amortizing a single dose over 6 months by micro-dosing and receiving a clearly sub-therapeutic dose. While that may work for your specific usage, it doesn't change the unit economics for someone who needs the standard therapeutic dose, which in my case at 10 mg is roughly €400 every 4 weeks.

It’s a bit like arguing that a Porsche GT3 RS is an 'affordable' car because the monthly payment is low, provided you finance it over 30 years. The sticker price hasn't changed, you've just engaged in extreme creative accounting to make it fit a monthly budget.

sReinwald

5 месяцев назадdiscuss

I suspect a big reason for why Mounjaro is still fairly expensive here in Germany (I pay nearly €400 for a 10mg Qwickpen - a 12.5mg Qwickpen is nearly €500) is due to health insurance not being allowed to cover them for anything but diabetes treatment.

If health insurance companies would be able to cover these drugs, there'd have to be negotiations between Eli Lilly and the insurance companies, and insurance companies have a bigger lever than individual patients who pay out of pocket. Self-payers are just price-takers. We pay whatever Eli Lilly wants us to pay.

sReinwald

5 месяцев назадdiscuss

It would probably be interesting, but if you are not overweight, the appetite suppression will likely make this not a very healthy or very fun experiment. I started at 5mg in October, and even on that smaller dose I had to force myself to eat even just ~800kcal a day - especially in the early weeks. When you have a lot of weight to lose, that's a pretty welcome effect. When you are already at a healthy weight, not so much. That caloric intake would put most adults into a pretty deep caloric deficit.

I'd suspect if the effects on non-weight indications check out in studies, we might see drugs that could specifically target those effects without also slowing down your digestive tract. Addictions like nicotine and alcoholism and their consequences cost health insurance companies (and us as a society) billions of Euros/Dollars each year, so there'd be a strong incentive to pursue this.

sReinwald

5 месяцев назадdiscuss

One thing I have noticed on Mounjaro is a (at least subjectively) significant decrease in impulse spending / buying random crap off Amazon. I have ADHD and that has been a real problem for quite a while - even with ADHD medication (Elvanse/Vyvanse in my case).

The part of Mounjaro that regulates the craving side of the weight loss equation (like reducing 'food noise' and the desire for sugary and fatty foods) seems to also affect other behaviors due to Mounjaro's effect on the brain's reward circuitry. I believe there are also early preliminary studies that indicate it can help with addictions like alcoholism.

Those drugs really are quite something. Shame they're so damn expensive. Insurance here in Germany is unfortunately legally prohibited from covering GLP-1 class drugs for weight loss unless you have a diabetes diagnosis.

sReinwald

5 месяцев назадdiscuss

Like I said, the first examples are fairly trivial, and you absolutely don't need an LLM for those. A good agent architecture lets the LLM orchestrate but the actual API calls are deterministic (through tool use / MCPs).

My point was specifically about the news filtering part, which was something I had tried in the past but never managed to solve to my satisfaction.

The agent's job in the end for a morning briefing would be:

  - grab weather, calendar, Todoist data using APIs or MCP  
  - grab news from select sources via RSS or similar, then filter relevant news based on my interests and things it has learned about me  
  - synthesize the information above

The steps that explicitly require an LLM are the last two. The value is in the personalization through memory and my feedback but also the ability for the LLM to synthesize the information - not just regurgitate it. Here's what I mean: I have a task to mow the lawn on my Todoist scheduled for today, but the weather forecast says it's going to be a bit windy and rain all day. At the end of the briefing, the assistant can proactively offer to move the Todoist task to tomorrow when it will be nicer outside because it knows the forecast. Or it might offer to move it to the day after tomorrow, because it also knows I have to attend my nephew's birthday party tomorrow.

sReinwald

5 месяцев назадdiscuss

> But can't you do the same just using appropriate MCP servers with any of the LLM providers?

Yeah, absolutely. And that was going to be my approach for a personal AI assistant side project. No need to reinvent the wheel writing a Todoist integration when MCPs exist.

The difference is where it runs. ChatGPT Tasks and MCP through the Claude/OpenAI web interfaces run on their infrastructure, which means no access to your local network — your Home Assistant instance, your NAS, your printer. A self-hosted agent on a mac mini or your old laptop can talk to all of that.

But I think the big value-add here might be "disposable automation". You could set up a Home Assistant automation to check the weather and notify you when rain is coming because you're drying clothes on the clothesline outside. That's 5 minutes of config for something you might need once. Telling your AI assistant "hey, I've got laundry on the line. Let me know if rain's coming and remind me to grab the clothes before it gets dark" takes 10 seconds and you never think about it again. The agent has access to weather forecasts, maybe even your smart home weather station in Home Assistant, and it can create a sub-agent, which polls those once every x minutes and pings your phone when it needs to.

sReinwald

5 месяцев назадdiscuss

You could, but Claude Code's memory system works well for specialized tasks like coding - not so much for a general-purpose assistant. It stores everything in flat markdown files, which means you're pulling in the full file regardless of relevance. That costs tokens and dilutes the context the model actually needs.

An embedding-based memory system (letta, mem0, or a self-built PostgreSQL + pgvector setup) lets you retrieve selectively and only grab what's relevant to the current query. Much better fit for anything beyond a narrow use case. Your assistant doesn't need to know your location and address when you're asking it to look up whether sharks are indeed older than trees, but it probably should know where you live when you ask it about the weather, or good Thai restaurants near you.

sReinwald

5 месяцев назадdiscuss

Disclaimer: Haven't used any of these (was going to try OpenClaw but found too many issues). I think the biggest value-add is agency. Chat interfaces like Claude/ChatGPT are reactive, but agents can be proactive. They don't need to wait for you to initiate a conversation.

What I've always wanted: a morning briefing that pulls in my calendar (CalDAV), open Todoist items, weather, and relevant news. The first three are trivial API work. The news part is where it gets interesting and more difficult - RSS feeds and news APIs are firehoses. But an LLM that knows your interests could actually filter effectively. E.g., I want tech news but don't care about Android (iPhone user) or MacOS (Linux user). That kind of nuanced filtering is hard to express as traditional rules but trivial for an LLM.

sReinwald

5 месяцев назадdiscuss

Not saying the frontier models aren't smarter than the ones I can run on my two 4090s (they absolutely are) but I feel like you're exaggerating the security implications a bit.

We've seen some absolutely glaring security issues with vibe-coded apps / websites that did use Claude (most recently Moltbook).

No matter whether you're vibe coding with frontier models or local ones, you simply cannot rely on the model knowing what it is doing. Frankly, if you rely on the model's alignment training for writing secure authentication flows, you are doing it wrong. Claude Opus or Qwen3 Coder Next isn't responsible if you ship insecure code - you are.

sReinwald

5 месяцев назадdiscuss

IMO the value and differentiating factor is basically just the ability to organize them cleanly with accompanying scripts and references, which are only loaded on demand. But a skill just by itself (without scripts or references) is essentially just a slash command with metadata.

Another value add is that theoretically agents should trigger skills automatically based on context and their current task. In practice, at least in my experience, that is not happening reliably.

sReinwald

5 месяцев назадdiscuss

I don't see how that makes it uniquely viable in France. Germany has something very much like this too. And we've had it for nearly 13 years.

> Since 31 August 2013 companies which operate public petrol stations or have the power to set their prices are required to report price changes for the most commonly used types of fuel, i.e. Super E5, Super E10 and Diesel “in real time” to the Market Transparency Unit for Fuels. This then passes on the incoming price data to consumer information service providers, which in turn pass it on to the consumer.

As a consumer, there is no direct API by the MTS-K that you can use, but there are some services like Tankerkoenig which pass this data on to you. I have used their API in Home Assistant before I switched to an EV.

https://www.bundeskartellamt.de/EN/Tasks/markettransparencyu...