efitz

I created a script that removes the "videos", "short videos", suggested searches, and "people also search for" on Google search results page (videos only suppressed when not searching for videos).

https://www.tweeks.io/share/script/c5d9b63419f64d64a107fb48

Redefining doing the thing into a complex process, where actually doing the thing is only one step of the process, and then other steps, is not doing the thing.

I haven’t seen this new google model but now must try it out.

I will say that other frontier models are starting to surprise me with their reasoning/understanding- I really have a hard time making (or believing) the argument that they are just predicting the next word.

I’ve been using Claude Code heavily since April; Sonnet 4.5 frequently surprises me.

Two days ago I told the AI to read all the documentation from my 5 projects related to a tool I’m building, and create a wiki, focused on audience and task.

I'm hand reviewing the 50 wiki pages it created, but overall it did a great job.

I got frustrated about one issue: I have a github issue to create a way to integrate with issue trackers (like Jira), but it's TODO, and the AI featured on the home page that we had issue tracker integration. It created a page for it and everything; I figured it was hallucinating.

I went to edit the page and replace it with placeholder text and was shocked that the LLM had (unprompted) figured out how to use existing features to integrate with issue trackers, and wrote sample code for GitHub, Jira and Slack (notifications). That truly surprised me.

Yeah, I was hoping that AI would make games more exciting, not play them for me.

On the other hand, it opens new worlds for farming…

This is not a criticism of the author of the article; I can feel their frustration and the pain of their loss but I don’t have anything to say that can ameliorate that.

I’m coming up on 60 years old. The mindset that many people have now seems to be to wait for someone from the government to help them- whether it’s ambulance, police, health care, food assistance, college, whatever.

Back in the 80s when I was coming of age the mindset among the people around me was different. The mindset of many people today seems more “I’m helpless”.

The government has never been good at anything; it doesn’t have the right organizational incentives and doesn’t give the right personal incentives to employees. I’m not criticizing government employees; there are individuals that do a great job, and from time to time I am delighted with an interaction, but mostly it’s just lots of waiting for mediocre service.

I don’t know how to change a widespread mindset of helplessness, but I suspect that it involves changing laws to remove all the obstacles that government and lawsuits have put in place to helping yourself and helping others.

Why does the successor to the terminal need to be text oriented at all?

Maybe it is an API. Maybe the kernel implements this API and it can be called locally or remotely. Maybe someone invents an OAuth translation layer to UIDs. The API allows syscalls or process invocation. Output is returned in response payload (ofc we have a stream shape too).

Maybe in the future your “terminal” is an app that wraps this API, authenticates you to the server with OAuth, and can take whatever shape pleases you- REPL, TUI, browser-ish, DOOM- like (shoot the enemy corresponding to the syscall you want to make), whatever floats your boat.

Heresy warning. Maybe the inputs and outputs don’t look anything like CLI or stdio text. Maybe we move on from 1000-different DSLs (each CLI’s unique input parameters and output formats) and make inputs and outputs object shaped. Maybe we make the available set of objects, methods and schemas discoverable in the terminal API.

Terminals aren’t a thing of the 80s; they’re a thing of the early 70s when somebody came up with a clever hack to take a mostly dumb device with a CRT and keyboard and hook it to a serial port on a mainframe.

Nowadays we don’t need that at all; old-timers like me like it because it’s familiar but it’s all legacy invented for a world that is no longer relevant. Even boot environments can do better than terminals today.

Why is this news?

Microservices embody a concepts that the article overlooks or downplays:

1. modularity - yes - but even better than what the article describes, there is little or no ability to cheat the modularity - the microservice has an api as a contract and is isolated in execution so there’s no trivial way to go around the api.

2. Independently committable/deployable. One reason to consider microservices is organizational- maybe you don’t want to or can’t share a repro with another team.

Now of course microservices have lots of downsides and are not a panacea and may be a bad fit for your project.

Building software is usually a craftsmanship task.

Software can be engineered, but It’s rare and expensive so it’s only built that way when the cost is justified, as when building life critical systems (manned aircraft/spacecraft flight controllers) or security critical components like ssl stacks, cryptographic algorithm implementations, etc.

Non sequitur.

If you are a software engineer, you are going to be expected to use AI in some form in the near future. A lot of AI in its current form is not intuitive. Ergo, spending a small effort on building an AI agent is a good way to develop the skills and intuition needed to be successful in some way.

Nobody is going to use a CPU you build, nor are you ever going to be expected to build one in the course of your work if you don’t seek out specific positions, nor is there much non-intuitive about commonly used CPU functionality, and in fact you don’t even use the CPU directly, you use translation software whit itself is fairly non-intuitive. But that’s ok too, you are unlikely to be asked to build a compiler unless you seek out those sorts of jobs.

EVERYONE involved in writing applications and services is going to use AI in the near future and in case you missed the last year, everyone IS building stuff with AI, mostly chat assistants that mostly suck because, much about building with AI is not intuitive.

There is a general problem with rewarding people for the volume of stuff they create, rather than the quality.

If you incentivize researchers to publish papers, individuals will find ways to game the system, meeting the minimum quality bar, while taking the least effort to create the most papers and thereby receive the greatest reward.

Similarly, if you reward content creators based on views, you will get view maximization behaviors. If you reward ad placement based on impressions, you will see gaming for impressions.

Bad metrics or bad rewards cause bad behavior.

We see this over and over because the reward issuers are designing systems to optimize for their upstream metrics.

Put differently, the online world is optimized for algorithms, not humans.

Ubiquiti is awesome, but their IPv6 support leaves something to be desired.

I have two ISPs, one with IPv6 (Starlink) and one without (Frontier).

I want to use Frontier for all IPv4, with IPv4 failover to Starlink, and I want to use Starlink only for IPv6.

UniFi networking won’t let you configure this, and I’m not going to SSH in to my UDM to manually set routes, that will be lost at next boot.

They absolutely saw it coming and warned against it, but couldn’t figure out any durable way to prevent it.

So there is no other way of reporting such abuses? This one web site that nobody had heard of before today, was our sole way of hearing complaints of this type?

Why is it that eliminating one particular web site is somehow a failure of the US Constitution?

Yes, Congress is dysfunctional. Welcome to the post 17th amendment world. Repeal that and make the House truly proportional instead of artificially limiting it to 435 members and you’ll go a long way towards fixing a lot of the current problems. Eliminate PACs and donation caps and enforce KYC for donations and we can see who is actually buying our legislators.

But on the main topic, the left in the US is seeking judicial intervention to block nearly every single action that the administration takes, and district court judges are handing down nationwide injunctions against the president on a weekly basis. If this is such a crisis, then go judge shopping and get an injunction.

Yeas ago, I thought- why not both?

Use a catapult/trebuchet/cannon to launch the drone as close as possible to the target area - maybe use a discardable biodegradable sheath to improve aerodynamics. At the optimal distance and height, switch the drone from ballistic to powered mode and complete the delivery. Maybe increases payload (the drone doesn’t have to take off carrying the payload). And definitely increases range, as we are not having to use onboard power for flight for part of the laden delivery, so there is more energy available for the unladen return trip.

Personally I would love to see cannon launches but most localities are unlikely to approve (maybe Texas?)

Lots. Anything involving standing at the least, but also many of the “unauthorized people changed the rules without the authority of the state legislature”.

https://election-integrity.info/2020_Election_Cases.htm

You’ll find similar lists on Wikipedia and at the American Bar Association; they slant their interpretations differently than I do here. Look into a couple of the “standing” or “no merit” cases and see if you really think they were without merit or that there was no “case or controversy” as defined in the Constitution.

Yes, a couple were kooky, particularly around voting machines, but most were “you changed the rules in a way that weakens election security and you are not Constitutionally allowed to do that, since you aren’t the state legislature and the written law is at odds with your rules”.

Great, so I’m sure that you will not complain at all if Democrats lose in the future where a single company controls so many voting machines, unless they are able to prove fraud beyond a reasonable doubt in mere days after the election, right? And you’ll not complain if the vast majority of the suits filed are dismissed on procedural grounds such as standing, right? Because voters and candidates and state attorneys general obviously have no case or controversy in a contested election?

Yes there were a very small number of kooky cases in 2020. The vast majority did NOT get a fair hearing at all.

+1000

I should not have to vote on judges and dog catchers and stuff. I like officials being accountable, but voting for an unopposed “nonpartisan” candidate has negative value - it wastes time and resources and lends legitimacy to an essentially non democratic process with uninformed voting. Better to have an easy recall mechanism.

At the very least, put the federal stuff on a federal ballot, the state stuff on a state ballot, and the local stuff on a local ballot, and have them 4 months apart. Then we can get back to hand counting and election night results.

Two thoughts:

1. It is crazy that we are using machines in any way in the voting process.

2. Which is it? The MAGA people tried many lawsuits and many appeals to voting authorities for investigations. The unanimous response “safest election ever”. Ok fine, then no one should have a problem with whoever owns the voting machines, because there’s so little risk, only crazy people would even ask for investigation.

Which is it?

Ofc there is a problem with a single company or organization controlling a nontrivial segment of the voting machines used in the US. And ofc it was a problem in 2020 as well. The solution is to get easy-to-tamper-but-hard-to-detect stuff out of the voting process. Pen and paper and video recorded hand counts in front of witnesses. Same night results. It is not rocket science and most of the rest of the world does it this way.

I definitely got a strong vibe of “engineered to be viral”; I didn’t consider that it might be AI. I don’t dispute that. But I want to raise one point:

Em dash does NOT mean “AI”. Em dash is trivial to generate in most word processors- type a word, then a hyphen, then type another word. Microsoft Word and Google Docs both convert the hyphen to an em dash. I have intentionally done this for years. I also try to lead each paragraph with my point of the paragraph; that’s from outlining.

3 bullets, more than once: it’s an AI. Same emotional hook at the end of each paragraph, yep.

But just seeing an em dash does not mean it’s AI.

I’m flagging this post; the discussion went off the rails. I don’t understand how “government ends dubiously legal program that was abused and weaponized” can in any way be perceived as tyrannical.

Are you so nakedly partisan that you can’t accept a win at face value?

Gabbard et al were targeted by Quiet Skies. It was mentioned many times during the campaign and during the administration. I don’t think Noem is Machiavelli; the administration clearly signaled an intent to end this and it doesn’t surprise me that they did so and I think there were valid reasons given and I accept that anything politicians do may ALSO have a political goal in mind but doesn’t make their stated reasons insincere.

For truly colossal amounts of data, fedex has more bandwidth than fiber. I don’t know if any cloud providers will send you your stuff on physical storage, but most will allow you to send your stuff to them on physical storage- eg AWS snowball.

There are two main reasons why people struggle with cloud restore:

1. Not enough incoming bandwidth. The cloud’s pipe is almost certainly big enough to send your data to you. Yours may not be big enough to receive it.

2. Cheaping out on storage in the cloud. If you want fast restores, you can’t use the discount reduced redundancy low performance glacier tier. You will save $$$ right until the emergency where you need it. Pay for the flagship storage tier- normal AWS S3, for example- or splurge and buy whatever cross-region redundancy offering they have. Then you only need to worry about problem #1.

The lack of backups makes my blood boil. However, from my own experience, I want to know more before I assign blame.

The very first "computer guy" job I had starting in about 1990/1991, my mentor gave me a piece of advice that I remember to this day: "Your job is to make sure the backups are working; everything else is gravy."

While I worked in that job, we outgrew the tape backup system we were using, so I started replicating critical data between our two sites (using 14400 bps Shiva NetModems), and every month I'd write a memo requesting a working backup system and explaining the situation. Business was too cheap to buy it.

We had a hard drive failure on one of our servers, I requested permission to invalidate the drive's warranty because I was pretty sure it was a bad bearing; I got it working for a few weeks by opening the case and spinning the platter with my finger to get it started. I made sure a manager was present so that they could understand how wack the situation was- they bought me a new drive but not the extras that I asked for, in order to mirror.

After I left that job, a friend of mine called me a month later and told me that they had a server failure and were trying to blame the lack of backups on me; fortunately my successor found my stack of memos.

Reinventing? No. Using? Yes, for a lot of good reasons.

LLMs are expensive. Spending tokens to do something in bulk that is well suited to existing tools and algorithms, is wasteful and slow. And the main reason is that, using LLMs, the original author indicated only a 60% success rate for the task. Why spend many times more time and money and energy just to use an LLM on a well-understood preparatory task that it sucks at, when you can get much better results more inexpensively with off-the-shelf tools, and feed their results to the LLM for its unique value.

This was exactly my thought. Rather than feed the table directly to the LLM, build agents that extract the data and have the LLM act on the extracted data items. Then it’s a preference issue.

The author didn’t see much more than 60% accuracy which is not very useful for many (most?) real world tasks.

I believe that there is a lowest common denominator for accented English, and it sounds like “sung” English - which sounds like non-regional American English if the singer is singing and doesn’t naturally have a very pronounced accent. This is partially due to cultural effects but partially due to the mechanics of speaking with an accent.

By “lowest common denominator“, I mean omitting affectations like non-rotic R (Boston) and dropped g (urban) and exaggerated length of selected vowels (US southeast) and constant syllable length and other kinds of things. There are lots of videos on YouTube about how accents work.

But mainly phonetics is about eliminating silent letters, using the most unambiguous letter when there is a choice (eg K or S instead of C), etc.

Anyway I’m rambling but you probably get my point.

I don’t think it’s etymology and irregularity that make English “fun”; I think that it’s English’s large number of concepts because we freely appropriate words and their ideas from other languages (“schadenfreude”, “Deja vu”, etc), the small alphabet, the lack of diacritics, and the flexibility that has been forced onto it primarily by American speakers abusing grammar- verbification, etc.

I think that simplified English is the most likely candidate for a universal human language. The “Unicode of languages”, if you will.

I want to see English become mor fonetik and mor regular.

As I watched my son learn to speak, then later to read and write, I paid attention to his misspellings and they all made perfect sense if you approached the language phonetically.

The first sentence he wrote for me was:

“my daddy and i tocd on d woki toki” (“My Daddy and I talked on the walkie talkie.”)

At first confusing, but it omits all the silent, or effectively silent because native speakers usually omit the sound, or abbreviate it so much that non-native speakers would miss it if they didn’t know it was there. Like “l” in “talked”.

And don’t get me started on irregular conjugations.