pvtmert

14 hari yang laludiscuss

The comment from a 5-year old: But Google is doing that too!

  > The spokesperson added: "We remain concerned that ignoring the growing power of Google Cloud and Gemini will tilt the market in a harmful way."

pvtmert

23 hari yang laludiscuss

I am 100% sure there will be a `defaults write ...` command and/or a policy setting via Apple Configurator (mobileconfig profile) to disable it completely.

They will probably push hard on the "enabled by default" part. Since they are literally paying millions to Google...

pvtmert

25 hari yang laludiscuss

Ryanair et.al. advertising minimums that are not including anything larger than a purse or a small laptop bag.

When you go through the funnel, adding a cabin luggage (or anything basically) usually doubles/triples the price.

Most of regular airlines include a proper cabin item up to 8kg.

Furthermore, it is not possible to have a just cabin item option in Ryanair's booking website. It is always bundled with things like priority boarding (read: wait outside & standing longer while other people depart from the plane)

Nevertheless, these practices are misleading. Hurts customers more, creates unnecessary complaints and bureaucracy as well.

Because you can in fact pay a regular airline a reasonable price (40-60eur) and get the standard package (with cabin luggage) and get treated like a human being rather than a farm animal. (Btw adding cabin luggage bumps the ticket price 45-55 eur range in Ryanair's case. In a normal airline, you get a beverage & croissant for free while you must pay 5eur for 0.2L water in a Ryanair flight)

pvtmert

27 hari yang laludiscuss

I guess the article is targeting regular people. However combining terms like "Linux distro" with other non technical explanations like "useless components such as screens, cameras, etc" sounded a bit weird to me.

Also;

  For example, phones have a “low memory killer” daemon, which throttles memory-hungry applications.

I mean, this is also needed for server applications, more-so than personal use. Reason being in fact a runaway process would interrupt many users rather than a single user in server versus client use cases respectively...

pvtmert

27 hari yang laludiscuss

They need a distinction between pedal assist and fully electric (motor) bike.

I live in Luxembourg and the infrastructure is not as good as the Netherlands. Majority of offenders (in my experience) are delivery guys and teenagers. Where teenagers simply don't care & do reckless things like passing on a big intersection at full speed on a red light, delivery guys do the same things for extra tips.

Some of the fat-tire electric bikes are really fast, I happen to chase them on Kirchberg cycle road. I sustained 40km/h average behind delivery guys and they were still faster, eventually dropping me. (I was on a road bike).

Simply put, anything above 250watts does not belong to cycle lane.Includes humans :j

pvtmert

2 bulan yang laludiscuss

Given the advent of LLMs and agentic coding, I believe this article needs re-visiting as it makes it much more discoverable to compare individual files across commits.

pvtmert

2 bulan yang laludiscuss

...each PhoneTool profile contains relevant "awards" (titles) on people's profile. Things like "Blaze", "Thunderstorm", etc are awarded publicly. Any employee can also click to these awards, and see who else also been awarded the same thing. (Plus total number of people)

*) PhoneTool = Internal People Directory. Shows user-profiles of employees...

pvtmert

2 bulan yang laludiscuss

I was working at Amazon until recently. Number of internal documents (PRFAQs, 1-pagers, etc) having these sort of proses boomed since 2024.

Punchy titles are also part of the marketing speak. Before Claude or ChatGPT, it would be a delicious read, understanding how they come up with the initial idea for an internal system. Since then, most of reads like "It's not just X, ..." every other paragraph, making it dull...

pvtmert

2 bulan yang laludiscuss

Absolutely lovely article.

> Software development is about translating a problem into a solution that a computer can understand and automatically resolve. Preferably in a secure and scalable way.

True, meanwhile software engineering puts optional bit into the requirements bucket. (ie. Secure & Scalable)

---

For the problem description and gathering requirements sentiment; I don't think we'll _ever_ have a 100% proper way of doing this. If we did, we'd basically solve any and all problems in the world.

Nevertheless, I think AI can help with investigating and exploring the problem space. Especially when the problem is an already solved thing that the prompter hasn't gained enough expertise yet.

Moreover, I think (and keep mentioning) we will see different kind of models in the near future. Those would be more specialized per industry, per language (both programming and human languages), even per field.

Those will open up newer areas for employment & job market. Something like an "AI-trainer" but more of a knowledge-worker style. Although this can also be automated with LLMs, the limits on context length/size plus amount of compute required to re-train the models to iterate faster both are quite heavy.

pvtmert

2 bulan yang laludiscuss

Although I agree with the sentiment in the article, it smells very LLM~y. Especially the sections and punchlines. Such as: `That is not a rounding error. That is a line item that needs its own budget code.`

pvtmert

2 bulan yang laludiscuss

Not a hack at all.

In fact, this is how Kubernetes APIs work. Even the CLI asks YAML or JSON content type when `-o yaml` or `-o json` is specified.

Also, some other sites (albeit sometimes looking at the User-Agent header, hence more hacky) also doing the same. For example `ipinfo.io` returns HTML page when opened in a browser but a JSON (prettified) when requested via cURL.

pvtmert

2 bulan yang laludiscuss

  > But a representative for Amazon said that there is no such company-wide metric for AI usage, nor are there internal leaderboards where employees are measured against each other. Rather, employees are able to view their own AI usage on personal dashboards.

Such a bullshit statement. There is a _global_ dashboard that ranks Kiro/QuickSuite (formerly Amazon Q) usage per-employee based on tokens. The dashboard itself is in QuickSight (well, that also became part of QuickSuite anyway).

Not only the data is open to anyone, you can clearly sort by rank, daily/weekly/monthly/yearly usages. Current and former employees included. (By internal alias).

Moreover, there is an internal "awards" system that shows up in PhoneTool profile, each employee gets "awarded" of Kiro/AmazonQ/Quicksuite titles like "Blaze", "Thunderstorm", etc. You can see other recipients of the same award just by clicking on it.

  Note: PhoneTool is an internal profile directory where you can look up other employees as oneself...

---

On the side, I know several people who cannot produce proper code themselves, or integrate to anything on their own. Those who need constant hand-holding keep producing immense amount of stuff with Kiro/AmazonQ, over-ranking SDEs nowadays. (these are not SDEs, more of SysDev, Support Engineers and TPMs). This itself is not a specifically good or a bad thing. But once they stack-rank based on the token usage, I am pretty sure "good" engineers who put effort into writing "good" code will rank worse than people who does not put effort into "concise" solutions. Therefore, quality will eventually detoriate. And it will be too late once the leadership realizes what has been going on. (Well, they already seen the Amazon-Q/Kiro related outages and keep denying it...)

pvtmert

2 bulan yang laludiscuss

> But a representative for Amazon said that there is no such company-wide metric for AI usage, nor are there internal leaderboards where employees are measured against each other. Rather, employees are able to view their own AI usage on personal dashboards.

Such a bullshit. There is (was) Amazon Q (Now QuickSuite) leaderboard which is a Quick Sight dashboard. Moreover, each PhoneTool

pvtmert

2 bulan yang laludiscuss

https://archive.is/VLWWE

pvtmert

2 bulan yang laludiscuss

You don't need a (one huge) model to do everything. You need specialized & smaller models that are very good at specific tasks. Collaborating among themselves.

The fact that we see stagnation in terms of billions of parameters shows that efficiency does not scale linearly with the model size. More of an S shaped chart. The middle was Claude 3.5. Since then, it is more about integrating and collaborating with different systems.

pvtmert

2 bulan yang laludiscuss

I opened the article thinking that it's about the on-call scenario (paging). Something like, I got paged in the middle of the night and let the remediation/mitigation to the agent...

Which makes a lot of sense. Especially non-Tier-1 services.

  Note: Having previously worked at Amazon, certain shifts can be really busy. Busy as in 30-40 pages/incidents over the course of the shift. I sometimes wake up to a "ghost" page, although I left my position earlier this year...

pvtmert

2 bulan yang laludiscuss

> not affiliated with subq,

i see in the linked post they mention O(n) not O(1). O(1) would basically be impossible and instant. Something like no compute required, constant results...

The name subquadratic is actually good and makes sense to me. Because today's models are usually O(n^2) or worse. Anything equals or less than O(n^1) is basically sub-quadratic.

Meanwhile O(log n) would be logarithmic as the log name indicates. But we have a long way to go there. Maybe with double tokenizer plus extensive caching it may be possible...

What I mean here is tokenizing the user input; then capturing intent; caching intent -> response. So that next time once you get the intent, you don't need to do full transformer inference compute. This can be logarithmic complexity in terms of time complexity.

pvtmert

2 bulan yang laludiscuss

solution is simple: commits get shared privately. distros can build on these private patches. after some point those can be made public (ie commited at upstream)

pvtmert

2 bulan yang laludiscuss

The best part is: Copilot did not catch the impact/implications of this change in the (automatic) review.

Should we continue to keep trusting the AI review provided by Copilot?

pvtmert

2 bulan yang laludiscuss

AWS has been (blatantly) using Microsoft method of making their way in. Redis, Elasticsearch, whatnot, all follow the same procedure: 1. Here is a managed service. 2. Here is a fork of the managed service where we manage the server (you don't see) with 15% off in price/credits. Easier backups with clicks etc. 3. We are dropping support of managed-X, move to our fork. 4. Due to the market conditions, our forked service is now 50% more expensive. 5. Ah also, you cannot export/download your backups because they are in proprietary format. 6. Locked-in.

Submissions

Mounting git commits as folders with NFS (2023)

Networking changes coming in macOS 27

AWS S3 now includes NFS 4.1 support via Mount-Points

Tmate.io is Shut Down...

SSH Virtual Hosting: No Host Header, but Public Key

History of self-sustaining LLM agents in real-life workflows

Co-Routines in 1-page of C (2013)

Linux, Docker, and Binary Emulation: Curious Case of Multi-Arch Builds

The curious case of binfmt for x86 emulation for ARM Docker

Shai-Hulud Strikes Again, Again. (NPM Supply Chain Attack)

comments