Preloader

News

admin published in News

Tutoring company tells parents to save their money and 'use AI instead'

Maani Truu
Education correspondent
Updated

An Australian tutoring company has told parents to invest in Gemini or ChatGPT rather than pay overpriced human tutors to give their children an academic edge, as students increasingly turn to artificial intelligence for exam preparation and immediate study feedback.

Dymocks Tutoring and Talent 100, which has five centres across Sydney, will close at the end of the week after telling customers that technology has rendered its service obsolete.

Loading…

EmailLinkedInTwitterFacebook

Subscribe to gift this article

Gift 5 articles to anyone you choose each month when you subscribe.

Subscribe now

Already a subscriber?

License article

Read More

Fetching latest articles

    Source: Hacker News

    • no reactions
    0
    0 Share
    admin published in News

    Judge orders Trump to restore CNN, MS NOW, Politico press access

    Sept. 24 (UPI) — A federal judge early Thursday ordered the Trump administration to reinstate CNN, MS NOW and Politico’s White House access, temporarily blocking President Donald Trump’s ban of the three media organizations.

    The Trump administration has taken a hard-line stance against media outlets whose coverage it criticizes and has taken numerous actions during its second term to restrict their access, including taking control of the press pool that has access to Trump.

    Trump banned the three news outlets on Friday, calling the outlets “FAKE NEWS” on his Truth Social media platform, prompting the trio to sue, alleging their constitutional rights had been violated.

    Judge Timothy Kelly of the U.S. District Court for the District of Columbia ruled in the news organizations’ favor early Thursday, issuing a temporary restraining order that directs the Trump administration to reinstate their hard-pass press credentials for 14 days, unless the court orders otherwise as litigation continues.

    In his ruling, Kelly, a Trump appointee, said the news organization were likely to succeed in showing that they were banned “without constitutionally adequate due process.” Because substantial interests are at stake, the government must provide clear standards for conduct that would lead to ban and give reporters fair notice of both the prohibited conduct and possible penalties.

    “None of defendants’ arguments convince the court that plaintiffs are not likely to succeed on their due process claim,” Kelly said, adding that the Trump administration had not shown why a ban was urgent as the reporting it identified were published months and years ago, and the activity it cited was “routine.”

    Kelly also said that the court was skeptical of the Trump administration’s defense that the ban was in the interest of national security, saying it had offered little evidence to support its claim.

    “For one thing, nothing in the record that predates this suit suggests that the revocation of plaintiffs’ hard passes was motivated by national security concerns,” Kelly said.

    “Certainly, that is not what President Trump said when he announced that he was ‘banning’ plaintiffs from the White House — instead, he focused on the alleged lack of truthfulness and negativity of plaintiffs’ reporting.”

    Ted Boutrous, a lawyer representing the outlets, told CNN that they “greatly appreciate” the court’s swift order.”

    “This is a strong ruling vindicating freedom of the press, due process and the rule of law,” Boutrous said.

    This week in Washington

    President Donald Trump speaks to more than 100 hunters and fisherman at a dinner in the Rose Garden of the White House on Thursday. Photo by Jim Lo Scalzo/UPI | License Photo

    Read More

    Source: U.S. News

    • no reactions
    0
    0 Share
    admin published in News

    The Year of Internal Tools

    Back to Code and Coordinates

    Code and Coordinates

    engineering at Geocodio

    The year of internal tools

    How Geocodio went from bash scripts to full internal apps, the system that keeps them maintainable, and where we draw the line between building and buying.


    By Mathias Hansen

    Engineering

    AI

    At Geocodio, it has always been important to us to automate processes and build our own tools and helpers, so we can save time and be more efficient. For almost ten years, Geocodio was just the two of us. Automation is a big part of what made that possible.

    Those tools have saved us countless hours over the years. Local bash scripts that set up a development environment or download the specific data files we need for geocoding. Infrastructure scripts that check the health of the load balancer or sync some data down. Small scripts that make the on-premises release process faster and more efficient, with fewer manual steps and less room for human error.

    One of the bigger ones came along well before the AI era. We built our own deployment tool. It visualizes the deployment process, so we can easily see how far along a deploy is and what is live right now. If something goes wrong we can see it and roll back. It handles deployment freezes and a bunch of other things that would be a lot more cumbersome and a lot less transparent as a manual process.

    The deployment tool, showing the public cluster with every node on the latest release and the deploy dialog open

    The deployment tool. Every node in the public cluster is marked with the release it is running, and the deploy dialog lets you pick a scope and post the result to Slack.

    All these small scripts and helpers and various tooling have been helpful over the years and have had a positive impact. But it pales in comparison to where we are at in 2026 and what we are able to do now.

    What changed

    This is the year of internal tools. We have gone from creating small bash scripts here and there, artisan commands in Laravel, small isolated repos with code to generate reports, to being able to really raise our ambitions and create full-fledged internal apps. Mobile friendly, good UX, fantastic test coverage.

    What allowed us to do that is the advent of AI and frontier models that let us build at a much higher level than we have ever been able to before.

    I must admit that in the beginning I was hesitant to start taking on these larger projects. The thing that always got in the way of the bigger, more ambitious ideas was not the building. Even pre-AI you might be able to knock out some kind of awesome internal tool over a couple of days. But then you have the maintenance burden. You cannot keep building tools if you do not think about the fact that you have to maintain them, keep them working, keep them up to date, fix the bugs you find.

    That is the other side of the coin, and it is the side AI has also allowed us to solve. I am not worried about building all these internal tools and maintaining them, because I can use AI to keep up with the maintenance as well.

    We have a system for it now

    We have gotten to the point where we have specific systems in place that let us quickly scaffold and create a new internal tool. We recently launched our console-ui package, which includes Tailwind CSS tokens as well as shared React components, so all these internal apps share one component library instead of each one growing its own.

    We have also built up enough experience with AI that we have some solid principles for how to give these tools a strong foundation, so they are built for the long term:

    • A significant amount of planning before we take on the job of building anything.

    • A lot of research, including engineering spikes that prove out a concept first.

    • Numerous Grill Me sessions, using Matt Pocock’s wonderful Grill Me skill, which interrogates a plan until the weak parts fall out.

    • Iterating on UI mockups, which answers a surprising number of questions and resolves problems before we go and build the real thing.

    • Using powerful models like Fable in the planning process, to make sure what we are building solves the problem we are trying to solve, in a sustainable and well-engineered way.

    On top of that, we have also built up high quality internal standards for these projects. Testing, CI, static code analysis, best practices for authentication, that kind of thing.

    That is also where the time goes. The planning and the spec take weeks. The implementation takes hours or days. Some of these apps we had been thinking about for years before that, so the spec was mostly a matter of collecting what was already in notes and meetings and turning it into a project. Atlas is the clearest example. It had been in my head for years, and once the spec existed, building it was the fast part.

    Every tool ships with its own documentation

    Each of these tools has a full documentation subsite. A real guide on how to use the tool and how each feature works, with screenshots and examples.

    The Bullpen documentation subsite, with a user guide and internal docs section and a page explaining what Bullpen is

    Bullpen’s documentation subsite. Every tool has one, split into a user guide and internal docs.

    A lot of it is AI-generated, at least the initial draft. Even so, having that consistency across every tool is really nice. It communicates what a tool does and how to use it much better than a README ever did.

    It also solves a problem I did not expect. When you have been sitting there writing big specs and plans for weeks, you sometimes lose touch with what you built. Writing the documentation brings that back into focus.

    Every one of these repos has a rule in its CLAUDE.md saying that any change affecting the documentation requires Claude to go and update, add, or remove the relevant parts. It still needs a human eye here and there, but it is a solid base rule that keeps things current. Documentation that is not up to date is useless.

    This is the rule from the Atlas repo, lightly trimmed:

    CLAUDE.md

    ### Keep both sections in sync with the app

Whenever you make a material change you MUST update the documentation in the same change.
A user-visible change lands in the user guide; a change to architecture, dev
workflow/tooling, infrastructure, deployment, integrations, a pipeline/job/schedule, a
scheduled command, a deploy/secrets change, or an Environment API contract bump lands in
the internal docs. Many changes touch both.

- **Changed behavior** → edit the relevant page(s) so they describe how the thing actually
  works now. Don't leave stale screenshot markers, steps, class names, paths, commands, or
  claims.
- **New feature or new maintainer-facing surface** → add a new `.md` in that
  section's directory and register its slug and title in the matching `DocsManifest`
  constant. The manifest is the single source of truth for which pages exist, their order,
  and their sidebar grouping.
- **Removed capability** → delete the page **and** its manifest entry, and any cross-links
  to it from other pages.
- **Renamed/moved** → update the slug in both the file name and the manifest, and fix
  inbound links.

    What we have built

    We are taking on bigger challenges now, like building our own customer support platform.

    • Atlas for customer support. It is built, and we are rolling it out now.

    • Bullpen for managing, organizing, and planning our sprints.

    • Yak, a coding agent for papercuts. It picks up small tasks from Slack, Linear, Sentry, and GitHub and comes back with a pull request to review. Yak is open source, so anyone can download it and run it against their own repos.

    • Our deployment tool.

    • A compliance tool that covers the parts our compliance platform cannot do, and then pushes the results back into it.

    • Treehouse, which gives every git branch its own worktree and its own docker-compose stack, so several branches or agent sessions can run at once without colliding on ports. Also open source.

    Yak showing a completed task: the request from Linear, a summary of the change, the activity log, and a recorded walkthrough

    Yak, after finishing a task that came in from Linear. The request, the summary of what changed, the step-by-step activity log, and a recorded walkthrough of the result.

    Our infrastructure engineer is also working on a tool for managing infrastructure, handling CVEs in our dependencies, and making regular maintenance easier with CD jobs on top of GitHub Actions. That one is a good illustration of the maintenance point above. Some of the tools exist to keep the other tools healthy.

    Why the support platform is worth building ourselves

    The big benefit of building our own is that we can make it extremely well integrated with our workflows and our data.

    Right now, helping one customer can mean going into three or four different tools to get the information you need. Our current support tool. An admin tool, and we have two of those, one for self-service and one for enterprise. The billing tool. Analytics. Having both self-service and enterprise customers makes the picture more complex for us than it would be for most companies.

    The whole point of building the support platform ourselves is to bring all of that into one place. The person replying to a customer can see exactly what is going on with that customer, what kind of account they have, and what has happened recently, in a UI that is streamlined for the exact workflow we have at Geocodio. They get the information and the actions they need at their fingertips, which means we can solve the problem much faster.

    What we are not going to build

    None of this means we are going to replace every SaaS product we use with our own tool, or that we are going to go build dozens of new internal apps. There is a limit, and there is a fine line between what should be in-house and what should be off the shelf.

    Email stays with Bento, and we are happy with it. Running our own email infrastructure is a responsibility that makes no sense for us to take on. The risk is too high, and it is not something we have any experience with. Twilio keeps the communications work. Nobody here is writing a Slack. There is plenty of good tooling out there.

    Three questions sort it for us:

    1. Do we have real experience in this domain, or would we be learning it on the job?

    2. What happens to the business if this is down for a day?

    3. Does the value come from being integrated with our own data and workflow, or would an off-the-shelf tool do the job just as well?

    Just because tools are easier to maintain now does not mean we can take on fifty of them. We have to be deliberate and thoughtful about the way we build them. This is also why we build them on shared infrastructure, shared UI, and shared principles, so they are easier to maintain and share as much code as possible. We are not trying to create a micro-internal-tool-services architecture either. We want tools that cover a whole feature set, rather than inventing a new service every time we have an idea.

    Every tool you build is a responsibility

    That trade-off is real each time, and the part that is easy to forget is uptime. Internal tools need to be reliable too. What would happen to our business if we could not reply to customer support requests for a day because we shipped a change and broke something?

    We got a small taste of this recently. We have an internal ETL tool for importing and working with address data. We did a massive upgrade of it, and then it was unavailable for a day while we ironed out a few issues with the deployment. In this case it did not affect keeping the lights on. If the same thing had happened to Atlas on a busy day, it would have been a different story.

    That day was the result of a trade-off we made on purpose. These tools do not have a staging environment. A staging environment for every internal tool is one more thing to keep running and keep in sync, and for tools that only we use, we decided that cost was not worth paying. Changes go to production, and local QA, test coverage and CI carry the weight that staging would otherwise carry. The ETL outage is what that decision looks like when it goes wrong, and we accepted that when we made it.

    The other half of the responsibility is access. Atlas can see customer accounts and billing, so who can reach it matters as much as whether it is up. One big benefit of hosting these tools ourselves is that they live entirely inside our private network. They are not reachable from the public internet at all. You have to be on our VPN before the sign-in page even loads, and only then do the usual authentication rules apply.

    Where this leaves us

    I do not know what the future will bring. We have to be careful not to end up maintaining tons of internal tools, or creating tools for the sake of creating tools.

    But if these tools help us do our work better, make us more efficient, and reduce the chance of human error from manual processes, then we can build a better product, we can give our customers a better experience, and we can improve the quality of life for our team.

    If you are thinking about this at your own company, start small. Find the process where you already work around the gap between two tools by hand every week, and build the thing that closes it. That is where we got the most value the fastest, and it does not require you to believe that anyone should rebuild a SaaS product from scratch.

    Previous
    We’re Hiring: A Software Engineer and a Technical Support Engineer

    Get new posts in your inbox

    We write about what we’re working on, thinking about, and getting so excited playing around with that we accidentally stay up a bit too late.

    Thanks for subscribing!

    Source: Hacker News

    • no reactions
    0
    0 Share
    admin published in News

    Police ID man found dead in 1982, cite possible Tylenol murders link

    Sept. 23 (UPI) — Authorities in Idaho have identified a man found dead from cyanide poisoning in a Boise church more than four decades ago and believe he may be connected to the infamous Chicago Tylenol murders of 1982.

    The man known as the Unknown Wanderer has stumped detectives for years. He was found dead Dec. 4, 1982, inside Sacred Heart Catholic Church with no identification, an envelope containing 19 $100 bills and a typed note signed Wm. L. Toomey and two key rings with nearly 30 keys. His hair had been dyed, his nails were finely filed, his skin was unseasonably tanned and his pants were either brand new or recently pressed.

    The case went cold until 2019 when Ada County Detective Tim Cooper reopened the case after a member of the church asked him about it while he was sitting in his car drinking cheap gas station coffee in a parking lot.

    On Wednesday, Cooper identified the Unknown Wanderer to reporters during a press conference as Dr. Mathew Betkouski.

    Cooper said the break in the case came from one the keys they found on Betkouski’s person, which indicated it was from a Jacksonville, Fla., hardware store. Etchings on a second key led detectives to Jacksonville’s Mills Court road.

    After acquiring the property deeds for the street going back to the 1970s and 1980s, one person stood out to detectives — a man who had been pronounced dead in absentia in 1989. The Ada County Sheriff’s Office contacted the decedent’s only living relative who said that his brother had gone missing in 1982.

    A DNA sample collected from the man showed that there was a “very high probability” that they shared the same father, Cooper said.

    “So, the Unknown Wanderer was determined to be Dr. Mathew Francis Betkouski,” Cooper said, adding that amid the investigation they learned that he had worked at two drugstores during his undergraduate studies and had a history of volatility, including a documented fascination with cyanide and explosives.

    In the late 1970s, while Betkouski was working for a flavors and fragrances company in the Jacksonville area, colleagues became concerned about him after he discussed “the perfect crime” and how “cyanide could be used as a less messy means of taking one’s life,” Cooper said.

    “Dr. Betkouski was observed talking to himself on occasions and had fears that he was being tracked,” Cooper said.

    Writings by Betkouski found by detectives show that he tried to manage schizophrenia and depression and feelings of euphoria brought on by imaginary thoughts of violence.

    In late 1979, he was hired by Lockheed where a fellow employee recalled Betkouski stating that poison could be introduced into over-the-counter capsules, according to Cooper, who said Betkouski was fired in March 1982 after creating an unspecified disturbance at a Las Cruces, N.M., area pharmacy as he tried to acquire some kind of pills or capsules he said were for an experiment.

    Betkouski’s whereabouts after that date are not fully known and Cooper said they have not been able to place him in Chicago in September 1982, when at least seven people, including a 12-year-old girl, in and around the Windy City died after consuming potassium cyanide-laced Tylenol.

    Detectives found pictures of Betkouski in Chicago, but they believe they were likely taken in December 1981. He was last known to be in Las Cruces on Oct. 15 and was next found dead at the Boise church.

    “So, where this leaves us is while our office may have identified some possible connection between Dr. Betkouski and the 1982 events, we’re not naming him as a suspect and we cannot say he was in Chicago during the final week of September 1982,” Cooper said.

    “More information does exist in this case that we won’t be releasing today.

    While Cooper was restrained, Dr. Park Dietz, renowned forensic psychiatrist who was consulted during the initial Tylenol murders investigation and who was brought into the reopened case in 2021, was confident that Betkouski was likely responsible.

    “It’s an extraordinary thing to have been around at the beginning in 1982 and now, today in 2026, see very likely closure, that the identity of the Tylenol killer from 1982 can be shared,” Dietz said.

    Dietz told reporters that they presented information about Betkouski to groups of forensic psychiatrists, forensic psychologists and criminal profilers, who overwhelmingly considered Betkouski to be a viable suspect, while some “were as convinced as I that Detective Cooper had solved the most important unsolved murder of my lifetime, the Chicago Tylenol murders.”

    The Ada County Sheriff’s Office said it has shared its findings with the Chicago Police Department and is calling on members of the public with information about Betkouski to contact the office as investigators seek to determine his whereabouts from Oct. 15 to Dec. 4, 1982.

    The developments and yet-to-be-released connections between Betkouski and the Tylenol murders are to be the subject of an Unsolved Mysteries documentary titled Unmasking the Tylenol Killer, which is to air on Roku Oct. 9, saidAda County Sheriff Matt Clifford said during the press conference, explaining that authorities are hoping the exposure will generate more leads.

    “Detective Cooper has done amazing work, but there’s still a lot of questions we don’t have answers to,” Clifford said.

    “And that’s part of why we’re standing here today. I’m not going to beat around the bush with everybody, but this isn’t over.”

    Read More

    Source: U.S. News

    • no reactions
    0
    0 Share
    admin published in News

    Ideas on modernizing the open-source desktop

    By Joe Brockmeier
    September 23, 2026

    Akademy

    Scott Jenson has been working on user interfaces (UIs) and user experience (UX)
    for many years at Apple, Google, and other companies. Now, he’s trying to convince
    open-source projects to experiment more and drive the desktop beyond the age-old “windows, icons, menus,
    pointer    ” (WIMP) model. At Akademy 2026, KDE’s annual developer
    conference, he shared his complaints and ideas in a talk aimed
    at convincing those in attendance to take the lead on desktop design.

    He introduced himself as someone who has been doing UX for a long time, much
    of that time at Google. He said that over that time he had noticed a cultural shift,
    “because back in 2005 when I joined, we weren’t evil; we were really trying to do the
    right thing    “. He had worked on the Chrome team for a while, which he said was
    “filled with people who wanted nothing but open-source, open-web”
    projects. Things had changed radically, which led him to leave Google, “but I’m
    trying to atone for my sins working for these companies    “.

    This LWN.net subscription-only content has been made available to you by
    an LWN subscriber. To see more of this content, please take advantage of
    the following special offer.

    Free trial subscription

    Try LWN for free for 1 month: no payment
    or credit card required. Activate
    your trial subscription now     and see why thousands of
    readers subscribe to LWN.net.

    He now does work for Mastodon and Home Assistant, “just kind of trying to
    squeeze a little bit more good UX design into open source    “. Jenson said he did
    not want to be confrontational, but he thought that open source needs a lot of help
    in the area of UX design. That is challenging, because open-source projects are
    understaffed.

    Innovation

    He was at Akademy to discuss his ideas because he had given a talk
    similar to this at Ubuntu’s conference last year, which proved to be popular. “It
    got more than half a million views on YouTube—which by YouTube standards is
    trivial, but by Scott standards is enormous.    ” That told him that the topic of
    desktop UX was something that people wanted to talk about. This time around, he also
    wanted to show a few examples of prototypes he had to demonstrate his ideas.


    [Scott Jenson speaking]

    He thinks that desktop UX is very different from other kinds of UX. Jenson related
    a story about working on the Macintosh’s Finder application, which is a file
    manager. The team “wanted to do ellipses” in file names when showing a file in
    the Finder that had a name too long to display in its entirety. The developers wanted
    to put the ellipsis at the end of the file name, but he thought that would lose
    too much information, and suggested that the ellipses should be in the middle of the
    file names. “It’s now considered to be one of those things that’s an exciting
    little attention-to-detail issue the Macintosh does.    “

    Another UX detail that he worked on is the way that clicking and dragging files
    from one window to another works on the Mac. A mouse click action has two parts: when
    the user depresses the button (mouse down) and when the user releases the button
    (mouse up). On the Mac, item selection happens on mouse down, but the window only
    raises on the mouse up action. In other words, if a user drags an icon from a Finder
    window into another window, the first window is not raised.

    On most Linux systems, though, if a user is trying to click and drag a file from a
    window in the background it will cause the window to raise as soon as the user does
    the “mouse down” action. That complicates trying to click and drag a file from one
    window to another. That is a fundamental action that means that users can’t move data
    into applications quite as easily on Linux as they can on the Mac.

    He said that he had talked about that issue two years ago on Mastodon, which
    raised a response from a KDE developer who thought it was a good idea to separate the
    mouse down and mouse up actions when dragging data between windows. Then they
    implemented it a few hours later.

    “I forgot who that was. So if anybody here did that“, he said, but before
    he could finish the sentence a voice from somewhere in the room responded, “you’re
    welcome    “. Jenson yelled “thank you! I’ve been wanting to meet you for two
    years!    ” That was an example, he said, of how open source can be awesome:
    someone can just say “that’s a good idea, I’ll just do it” and then rapidly
    implement and push the feature to the project.

    Stabilized

    The point he was really trying to make, though, is that the desktop UX
    had not really changed in 20 years. There were plenty of new ideas and improvements in
    the 1980s and 1990s, “and then it just stabilized, and not much has really
    changed    “. But people are doing much more with their computers today, and he
    thinks that there are ways for desktop UX to grow.

    Much of the UX for Linux systems was originally copied from Windows and the Mac,
    “KDE even says it’s for Windows users, that it wants to be similar to Windows for
    users    “. There’s nothing wrong with that, but he felt that the Linux community
    “in general” had been leaning on Microsoft and Apple to do all of the UX research
    and testing, “and frankly, [make] the mistakes“.

    Some might point out that GNOME has conducted
    user research     for some of its design changes, and KDE’s human-interface guidelines (HIG) were also developed based on research by its
    contributors. However, it would be fair to say that these are exceptions rather than
    the norm; certainly, no one is funding or conducting regular UX design research and
    development for the Linux desktop overall.

    He said that it is smart to let Apple and Microsoft do these things first “so
    we can draft behind them    “, but the problem is that “they’ve just kind of given
    up to an extent    “. Apple has been letting its desktop UX, and desktop hardware,
    “languish for years and years and years” because it had shifted its focus to
    the iPhone and iPad devices. “I mean, like, Apple wanted the Mac to die. They
    wanted the world to move over to the iPad. Guess what? It
    didn’t.    “

    Apple has since turned its attention back to the Mac, but the things they’ve done
    since—such as the Liquid
    Glass design language    —have not been really popular with users. The features
    that Apple tends to add, he said, are “basically about tying it closer to the
    iPhone, tying you closer into their ecosystem    “. That isn’t about improving the
    desktop, it’s about increasing their moat.

    Microsoft “is just making one mistake after another“. He cited the
    company’s attempts to force users to store files using its OneDrive cloud-storage service, the
    privacy-invasive Windows
    Recall     feature that drew widespread criticism, and “ads everywhere” on the
    Windows desktop. His point was not to bash Apple or Microsoft, though the audience
    didn’t seem to object to this, it was to point out that the Linux desktop could not
    follow their example any longer. “The whole Linux community has been just waiting
    for Apple and Microsoft, more or less, to take all the risks, and they’re not taking
    any more risks.    ”

    Three pushbacks

    Jenson emphasized that he didn’t want to change everything, but he wanted to see
    experiments and additional development aimed at improving the desktop
    experience. However, he said that he gets pushback when he tells people that desktop
    innovation has stopped, and that the Linux desktop “can’t just patch and bug fix
    our way into the future, we have to start taking leadership ourselves    “.

    He grouped the pushback into three categories. The first objection he receives is
    “mobile won; the desktop is old, just do what mobile does“. He dismissed that
    by saying that mobile has indeed won the consumer market, social media, and other
    areas, but “it didn’t win productivity“. People still do their work on
    computers, because “desktops are silently awesome; they have keyboards and big
    screens, and they can do amazing things    “. However, people have lost sight of that
    and don’t really appreciate how good desktops are. “I think we should understand
    why they’re good, and then what we can do to make them better.    “

    The second objection he hears is that there are only so many ways that the WIMP
    model can work. “Just stop trying to fix it; it’s done.” He disagreed with
    that as well, and thought that there was more to be done. The final, particularly
    strong, objection from the tech community is “don’t touch my stuff“. People
    get angry when they have worked hard to get their desktop environment just as they
    like it and something changes to disrupt that. “And I get that. As I said before,
    I don’t want to change everything.    ” But the world is changing, “and we need to
    grow our way gently into some new things.    “

    What is UX?

    Thinking about where to go next with desktop design gets to the question “what is
    UX?” He said that term was misunderstood and that people, especially manager types,
    “they think it’s just pixels, but UX is so much more than pixels“. It helps to
    think of UX as having layers, he said. There is style, which consists of white
    space, iconography, color schemes, and yes, pixels. But there is also structure, he
    said, which is how the user moves through the application, the navigation
    model.

    Strategy is another part of UX; understanding who the target user is, and
    determining what they want. Deciding what is important, and what is not; what will
    the development team prioritize and what will it ignore? “The fastest way a UX
    designer can save your team time is to help you say no.    “

    There is also a layer he called “stuff“: the lower-level limitations of the
    underlying technology that a platform has which influence UX. For example, he said, if
    an application is being created for DOS, “you’re not going to write a nice
    Macintosh application. The technology dictates what you can do    “. The desktop has
    been evolving for 40 years and has “historical cruft” that carries expected
    standards.

    We just have a bunch of stuff that we just accepted. And I think if
    we’re going to fix the desktop, we have to understand this lower-level limitation,
    and how do we want to fix that? We just accept that mobile and desktop has stuff
    today, and we just don’t think about fixing it.

    People forget, he said, that the original Macintosh screen was “a whopping 342
    pixels high    “. Naturally, windows overlapped a lot. Since then, “we just
    assumed that Windows have to overlap    “. But many people have “giant, giant
    screens    “, and the existing window-manager model does not work so well for those
    devices. “I can understand why people use tiled window systems and so forth,
    because this existing model just does not work very well.    “

    Understanding the future

    Early in the talk, he had referenced a saying by Alan Kay, who had led design of the
    first WIMP interface at Xerox PARC: “Perspective is worth 80 IQ points.”
    Jenson now circled back to Kay: “Who, as you can tell, I really like a lot. […] He
    talks about three steps to understand stuff    “. The first step, he said, was to
    think about the present, “what is happening right now?” The second is
    “don’t forget the past, know what work has happened before“, and the final
    step is to ensure that “your questions are really clear“. Too many people rush to
    a solution, he said, but they should be rushing to questions. If the questions are
    clear, then the answers will be better.

    He did not want to make the talk about AI, he said, but he had to address
    it. “I have to talk about AI, and it pisses me off because there’s so many cool
    things we could be talking about, and AI just sucks up all the oxygen, and it’s all
    you can talk about    “. But, he conceded, there are many things happening with AI on
    the desktop, albeit in a weird way.

    He explained some of the current efforts to integrate AI with the desktop as
    “trying to sneak AI in” while keeping the UI exactly the same. Other efforts
    try to use AI to understand what is on the screen and use chatbots to interact with
    the interface, “which I think is kind of missing the point here“. That is,
    until last summer when Anthropic introduced Claude Code. “It completely changed
    things, because was looking down into your filesystem    ” and drew context from the
    content of a user’s files. That allowed it to do more interesting things and interact
    with the system.

    After mentioning Claude, he paused to say he was only speaking about frontier
    models because everyone else is discussing them. “I have a lot of personal
    problems with the frontier models. They are ethical and environmental disasters. I am
    not very excited by them    “. He said he was excited by Apertus, which is billed as a fully open,
    responsible model for sovereign AI created by the Swiss AI Initiative. “I do think it’s
    possible to talk about ethically trained small language models running locally, and
    you don’t have to sell your soul to the devil.    “

    The curse of direct manipulation

    The key point he wanted to make was not about generative AI uses, but about how
    giving it access to the filesystem improved its usability. It now had context and
    could work with the user’s data, but it was working at the filesystem level and not
    using the desktop UX.

    He talked about the desktop building blocks, basically the familiar WIMP model
    plus the desktop clipboard, as the way that users move data around. Expert users
    can move data around quickly, he said, but there’s a problem: the desktop is
    stateless. He called this the curse of direct manipulation; there is no working
    memory to the desktop. “If you copy a few too many things to the clipboard, oh
    sorry, it’s gone    “.

    Jenson theorized that the reason there are so many desktop environments, window
    managers, and Wayland compositors for Linux—as well as alternative clipboard
    managers—is because the desktop isn’t quite working right for users. He thought
    it was really about working memory: “How do I manage my data in such a way that I
    can remember it across all of these things and actually use it.    ” His perspective
    shifted to asking “what would we do to working memory to improve the desktop
    UX?    “

    AI may have given him the idea, but he added that he wanted to solve the desktop
    UX for people first. “If the AI can use it, great. I don’t care. That comes
    later.    “

    Remember your past

    There is prior art in trying to create systems that help provide users with more
    context and working memory related to their data. He referenced a paper called
    “Lifestreams: a storage model for personal data”. That paper made the case that
    desktop systems are “are ill-equipped to manage the electronic information and
    events of the typical computer user    “, and introduced the Lifestreams idea as a
    metaphor for dynamically organizing a user’s computer workspace.

    That paper influenced the ill-fated WinFS project from Microsoft, which
    was an attempt to merge data storage and management based on relational database
    systems. It was demonstrated in 2003 and scheduled to ship sometime later, but
    ultimately the project was shelved in 2006. On the free-software side of the house,
    there was the semantic
    desktop     project NEPOMUK,
    which was a research project funded by the European Union.

    A version of NEPOMUK was included
    in KDE 4    , with the idea was that it would allow applications “to use
    information from all over the desktop, the web, other devices, and combine it into
    one coherent interface    “. It was later removed and replaced with the Baloo file-indexing and search framework
    for the Plasma desktop, which has some overlap with NEPOMUK’s functionality but with
    a reduced scope of features.

    Jenson said that he is concerned that people see the failure of those projects as
    proof that the ideas were wrong. He argued that is not the case, “I would say that
    it is the hardware and systems at the time let the vision down. I think it’s time to
    rethink these projects    ” in light of newer hardware and better systems.

    Prototypes

    Having visited the present and the past, as Kay recommended, he had come up
    with a few questions that he thought worth exploring related to desktop UX. The first
    was “what would happen if we only designed for large monitors?” What would it
    unlock if there were desktop designs that ignored laptops and only focused on
    widescreen monitors? Another question he had was how could the desktop capture a
    user’s intent over time?

    Those questions led to the prototypes that he said he was a little nervous to show
    the audience “because technical people love to find mistakes“. The demos are
    about 29 minutes into the video
    of Jenson’s talk    . His slides have not yet been published but he told me at the
    event that they would be at some point.

    The first demo was to show how a desktop might make better use of a widescreen
    display. He noted that the center of a widescreen monitor is good for working
    with an application, but it becomes more difficult to work on the sides of the
    monitor. “It’s good for peripheral vision, but not good for working“.

    He demonstrated a desktop layout, a screenshot of which is below, that would put the
    focus on the window or windows in the middle of the monitor. He had included a grid
    in the background that tapers off from the middle of the screen to simulate a
    widescreen monitor. The idea was that a widescreen-first desktop would emphasize
    information in the middle of the screen, and then use an Exposé-like tiling model to arrange the
    rest of the desktop’s windows. This would make the windows that were not in use
    easier to interact with, he thought, and be a better model than virtual desktops for
    managing lots of windows. “I’m not against virtual desktops. Virtual desktops are
    used by very organized and intelligent people. I know my audience.    ” However, he
    said, most people can’t deal with virtual desktops.


    [Window manager widescreen demo]

    He also demonstrated dragging a music-player window all the way to one side of the
    screen, which he called “the stash area“. As it moved to the edge, it
    transformed into a widget with just a play button instead of showing the full
    window. He thought that it would be possible to do the things he was demonstrating
    with Wayland as it is today.

    Jenson put up another demonstration of an improved clipboard model that copied
    ideas from the Obsidian editor’s Canvas feature. Canvas allows users to drag and
    drop images, text, and files to create a visual layout of information. A user could
    create a document and drag information from a web browser into the document, drop a
    file in from another source, “the idea here is that this is now effectively a
    collection of stuff that I have gathered for this document. And if I close it and
    then come back tomorrow, it’s all still there    “. He added that a local AI would be
    able to come in and organize the information a user had gathered, “oh, look,
    you’ve got a bunch of hotels, let me organize them for you    “.

    The final idea he discussed was a way of gathering “privacy-preserving
    data    ” to help give users a better visualization of things like web-browsing
    history. A web browser will only give the user a view of their browsing history that
    “is not very useful” he said, so he tried to gather data “not using AI,
    just using simple math    ” that would use attention signals—such as how long
    the user was on a web page—to tell a story of where they spent their time on
    the desktop. “I basically came up with this spatial associative episodic memory
    prompt    ” to organize data in a helpful, interesting way. He thought that working
    memory was a good way to explore new ideas for desktop UX.

    So the goal with this talk is to say, let’s start to take this apart. And I really
    hope that my prototypes cause you guys to come to me and say, We forgot about this,
    and let’s talk about that. […] So I just want us to try, because no one else is
    going to try, and I think someone’s got to start.

    There was time for a single question. An audience member asked how Jenson’s model
    would protect against all of the privacy concerns that Microsoft faced with Windows
    Recall. “How are we going to make sure that this history remains only accessible
    by the user and cannot be hacked in like it could happen with Windows
    Recall?    “

    Jenson answered that his idea was to gather telemetry data, “not interesting
    information    “, which would significantly reduce the attractiveness of the data to
    would-be attackers. He said he did not want to underplay the issue, though, and
    wanted to build prototypes to see if the idea was good first, and then explore ways
    to encrypt or protect the data. He agreed that even this data could be valuable
    information. He added that the biggest problem with Windows Recall was “how
    stupidly they protected it    ” and he was sure others could do a lot better job.

    [I would like to thank the Linux Foundation, LWN’s travel sponsor, for its
    assistance with my trip to Graz, Austria for Akademy 2026.]

    Index entries for this article
    Conference Akademy/2026

    Source: Hacker News

    • no reactions
    0
    0 Share
    admin published in News

    OpenAI agent hacked Australian government website, PM says

    Cachella Smith
    Live reporter

    A headshot of Australian Prime Minister Anthony AlbaneseEPA

    As discussions took place at the UN General Assembly in New York about the opportunities AI provides – and how best to manage it – Australian Prime Minister Anthony Albanese revealed that the country’s universal health insurance scheme had been hacked by an OpenAI agent.

    In fact, the hack happened in June, the company became aware of it in August, and then sent an email to a government agency’s general inbox in September.

    “It took the company way too long to inform the government,” Albanese said, adding that the method of notifying officials was also “unacceptable”.

    Believed to be the first known breach of a government system by rogue AI agents, the hack accessed Medicare’s statistics portal which contains private, but not sensitive, data.

    Our cyber correspondent, Joe Tidy, writes that experts believe the systems were poorly protected – but more important is that this is not the first instance of AI agents ignoring laws on accessing online information.

    Tech reporter Tom Gerken explains AI itself has no true understanding of what it is being asked – while tech editor Zoe Kleinman asks if AI is in its “move fast and break things” era.

    The Australian government has launched an review of AI laws and governance – but this is a global issue.

    AI might have dominated discussions at Unga this week, but a consensus among world leaders on how best to harness opportunities while tackling the risks still feels a way off.

    Our live coverage is closing now – read our main story here. Thanks for joining us.

    Source: Hacker News

    • no reactions
    0
    0 Share
    admin published in News

    Hackers say they stole thousands of FBI employee records

    Sept. 23 (UPI) — A criminal hacking group claims to have stolen thousands of sensitive records “on almost ALL FBI Agents and individuals who filed an application with the FBI for a job.”

    The group, called ShinyHunters, posted the online message Tuesday, saying it breached the FBI’s online jobs portal to obtain the information, which includes agent and applicant names, addresses, phone numbers, spouse names, medical details and other data. That website remains offline Wednesday.

    “We have compromised the FBI,” the message said. It was addressed to FBI Director Kash Patel and agency cybersecurity leader Brett Leatherman.

    In a statement issued Wednesday, the FBI said that it is aware of the claims but that the point of the breach is still undetermined.

    “We are actively and aggressively investigating this matter and working closely with those third-party providers that support FBIJobs.gov to mitigate any and all risk,” the statement said.

    ShinyHunters said that it targeted the FBI because of an advisory the agency issued in May. In that advisory, the FBI warned that the group was known to harass hacking victims or family members with “real or exaggerated” claims and extortion attempts.

    The group demand the FBI “correct or simply remove” the advisory, which the FBI issued after it identified ShinyHunters as responsible for a hack of the online learning system Canvas.

    “We have been severely offended,” the group said in its statement. It gave the FBI a week to change or remove the advisory, although it didn’t say what measures it would take otherwise.

    The New York Times reported that the group sent it and other news organizations a sample of what it claimed is the leaked information. “Cross references of some of that sample data against publicly available information online indicated that the materials are most likely legitimate,” The Times reported.

    Read More

    Source: U.S. News

    • no reactions
    0
    0 Share
    admin published in News

    U.N. General Assembly: Iranian president accuses U.S. of terrorism, illegal attacks

    Sept. 23 (UPI) — Iranian President Masoud Pezeshkian blamed the United States and Israel for causing global instability and accused them of launching illegal attacks on his country Wednesday in an address to the U.N. General Assembly.

    One day earlier, President Donald Trump told the United Nations that he must make a deal with Iran or “annihilate the Islamic republic.”

    “Do I drive them into hell with no chance of survival and no hope of future greatness or generations?” Trump wondered aloud.

    The Iranian military said the threats from Trump are “a sign of strategic desperation.”

    Pezeshkian’s speech contradicted what Trump said about Iran being desperate for a peace deal. Trump said the deal will come right after the U.S. midterm elections on Nov. 3.

    Trump didn’t attend the United Nations General Assembly on Wednesday. He was in Washington preparing for Chinese President Xi Jinping’s state visit.

    Pezeshkian began his speech by showing photos of children that he said were killed by U.S. and Israeli bombs. He told the United Nations that Iran will fight back “until our last breath.”

    “The United States president described us as terrorists. We have been the victims of terrorism,” Pezeshkian said.

    The only representative of the United States in the hall walked out as Pezeshkian spoke of the war’s civilian casualties.

    “Are we the sources of instability? We only seek to live independently,” Pezeshkian said. “We are the ones who cause instability? Those who are committing terrorism and committing crimes, are they the protectors of peace and stability?”

    It was the first time any country addressed the United Nations while being actively at war with the United States, The New York Times reported.

    In March, a Tomahawk missile hit a primary school in Minab, Iraq. It killed more than 150 people, including about 120 children.

    The United Nations’ Independent International Fact-Finding Mission on Iran found earlier this month that the United States may have committed war crimes in the attack.

    Tuesday afternoon, Trump told reporters that U.S. officials had a good meeting with Iranian envoys at the United Nations.

    Iranian officials gave “clear and transparent” conditions for diplomacy with the United States during those talks, Iranian Foreign Ministry spokesman Esma Baghaei said in an X post.

    He said those terms are: “Cessation of aggressive U.S. actions, including the naval blockade, economic terrorism, ending the war on all fronts, release of Iran’s frozen or restricted assets [and] acceptance of a safe shipping route in the manner agreed upon between the two coastal governments.”

    Iran has said it won’t give up the right to develop peaceful uses of nuclear technology.

    “ln order to defend our country, we do not seek anyone’s permission,” Pezeshkian said at the United Nations. “We will not bow our heads and give up the right that is inherently ours. We say it clearly: No nuclear weapons and no limitations to new peaceful nuclear technology,” he said.

    “For our country, peaceful nuclear technology is part of our inherent progress,” Pezeshkian said.

    Scenes from the 81st U.N. General Assembly

    President of Argentina Javier Gerardo Milei speaks at the United Nations General Assembly at U.N. Headquarters in New York City on September 23, 2026. Photo by Peter Foley/UPI | License Photo

    Read More

    Source: U.S. News

    • no reactions
    0
    0 Share