brumka: (Default)
brumka ([personal profile] brumka) wrote2025-09-02 06:55 am
paserbyp: (Default)
paserbyp ([personal profile] paserbyp) wrote2025-09-01 12:52 pm
Entry tags:

The End of Programming as We Know It

For decades, programming has meant writing code. Crafting lines of cryptic script written by human hands to make machines do our bidding. From the earliest punch cards to today's most advanced programming languages, coding has always been about control. Precision. Mastery. Elegance. Art.

But now we're seeing a shift that feels different. AI can write code, explain it, refactor, optimize, test, and even design systems. Tools like GitHub Copilot and GPT-4 have taken what was once a deeply manual craft requiring years of hard-fought experience and made it feel like magic.

So, the question on everyone's mind:

Is AI the end of programming as we know it?

The short answer is yes, but not in the way you might think.

To understand where we're going, we must look at where we've been as an industry.

Early computing didn't involve keyboards or screens. Programmers used punch cards, literal holes in paper, to feed instructions into machines. It was mechanical, slow, and very fragile. A single misplaced hole could break everything, not to mention a bug crawling into the machine.

Then came assembly language, a slightly more human-readable way to talk to the processor. You could use mnemonic codes like MOV, ADD, and JMP instead of binary or hexadecimal. It was faster and slightly easier, but it still required thinking like the machine.

High-level compiled languages like C marked a major turning point. Now we could express logic more naturally, and compilers would translate it into efficient machine instructions. We stopped caring about registers and memory addresses and started solving higher-level problems.

Then came languages like Python, Java, and JavaScript. Tools designed for developer productivity. They hid memory management, offered rich libraries, and prioritized readability. Each layer of abstraction brought us closer to the way humans think and further from the machine.

Every step was met with resistance.

"Real programmers write in assembly."

"Give me C or give me death!"

"Python? That's not a language, it's a cult!"

And yet, every step forward allowed us to solve more complex problems in less time.

Now, we're staring at the next leap: natural language programming.

AI doesn't give us a new language. It gives us a new interface. A natural, human interface that opens programming to the masses.

You describe what you want, and it builds the foundation for you.

You can ask it to "write a function to calculate the temperature delta between two sensors and log it to the cloud," and it does. Nearly instantly.

This isn't automation of syntax. It's automation of thought patterns that used to require years of training to master.

Of course, AI doesn't get everything right. It hallucinates. It makes rookie mistakes. But so did early compilers. So did early human programmers. So do entry-level and seasoned professional engineers.

The point is simple. You are no longer required to think like a machine.

You can think like a human and let AI translate.

AI is not the end of programming. It's the latest and most powerful abstraction layer in the history of computing!

So why do so many developers feel uneasy?

Because coding has been our identity. It's a craft, a puzzle, a superpower. It's what we love to do! Perhaps for some, even what we feel we were put on this Earth to do. The idea that an AI can do 80% of it feels like a threat. If we're not writing code, what are we doing?

Thankfully, this isn't the first time we've faced this question.

Assembly programmers once scoffed at C. C programmers once mocked C++, Python, and Rust. Each generation mourns the tools of the past as if they were sacred.

Here's the uncomfortable truth: We don't miss writing assembly, managing our own memory in C, or boilerplate code.

What about API glue? Or scaffolding? Low-level drivers? We won't miss it one bit in the future!

Sure, you may long for the "old days," but sit down for an hour, and you'll quickly thank God for the progress we've made.

Progress in software has always been about solving bigger problems with less effort. The march to adopt AI is no different.

For the last 50+ years, we've been stuck translating human vision into something that machines can understand. Finally, we are at the point where we can talk to a machine like it's a human and let it tell the machine what we want.

As programming evolves, so do the skills that matter.

In the world of AI-assisted development, the most valuable skill isn't syntax or algorithms, it's clarity.

Can you express what you want?

Can you describe edge cases, constraints, and goals?

Can you structure your thinking so that an AI, or another human, can act on?

Programming is becoming a conversation, not a construction.

Debugging becomes dialogue.

System design becomes storytelling.

Architecture becomes strategic planning, done in collaboration with AI and your team to align vision and execution.

In other words, we're shifting from "how well can you code" to "how well can you communicate?"

This doesn't make programming less technical. It makes it more human.

It forces us to build shared understanding, not just between people and machines, but between people and each other.

So, is AI the end of programming as we know it?

Absolutely.

Syntax, editors, or boilerplate code no longer bind us.

We are stepping into a world where programming means describing, collaborating, and designing.

That means clearer thinking. Better communication. Deeper systems understanding. And yes, letting go of some of the craftsmanship we once prized.

But that's not a loss.

It's liberation.

We don't need punch cards to feel like real developers.

We don't need to write assembly to prove our value.

And in the future, we won't need to write much code to build something amazing.

Instead, we'll need to think clearly, communicate effectively, and collaborate intelligently.

And that, perhaps, is the most human kind of programming there is.
mark: A photo of Mark kneeling on top of the Taal Volcano in the Philippines. It was a long hike. (Default)
Mark Smith ([staff profile] mark) wrote in [site community profile] dw_maintenance2025-08-31 07:37 pm

Code deploy happening shortly

Per the [site community profile] dw_news post regarding the MS/TN blocks, we are doing a small code push shortly in order to get the code live. As per usual, please let us know if you see anything wonky.

There is some code cleanup we've been doing that is going out with this push but I don't think there is any new/reworked functionality, so it should be pretty invisible if all goes well.

paserbyp: (Default)
paserbyp ([personal profile] paserbyp) wrote2025-08-31 01:43 pm
Entry tags:
denise: Image: Me, facing away from camera, on top of the Castel Sant'Angelo in Rome (Default)
Denise ([staff profile] denise) wrote in [site community profile] dw_news2025-08-31 12:28 pm

Mississippi site block, plus a small restriction on Tennessee new accounts

A reminder to everyone that starting tomorrow, we are being forced to block access to any IP address that geolocates to the state of Mississippi for legal reasons while we and Netchoice continue fighting the law in court. People whose IP addresses geolocate to Mississippi will only be able to access a page that explains the issue and lets them know that we'll be back to offer them service as soon as the legal risk to us is less existential.

The block page will include the apology but I'll repeat it here: we don't do geolocation ourselves, so we're limited to the geolocation ability of our network provider. Our anti-spam geolocation blocks have shown us that their geolocation database has a number of mistakes in it. If one of your friends who doesn't live in Mississippi gets the block message, there is nothing we can do on our end to adjust the block, because we don't control it. The only way to fix a mistaken block is to change your IP address to one that doesn't register as being in Mississippi, either by disconnecting your internet connection and reconnecting it (if you don't have a static IP address) or using a VPN.

In related news, the judge in our challenge to Tennessee's social media age verification, parental consent, and parental surveillance law (which we are also part of the fight against!) ruled last month that we had not met the threshold for a temporary injunction preventing the state from enforcing the law while the court case proceeds.

The Tennesee law is less onerous than the Mississippi law and the fines for violating it are slightly less ruinous (slightly), but it's still a risk to us. While the fight goes on, we've decided to prevent any new account signups from anyone under 18 in Tennessee to protect ourselves against risk. We do not need to block access from the whole state: this only applies to new account creation.

Because we don't do any geolocation on our users and our network provider's geolocation services only apply to blocking access to the site entirely, the way we're implementing this is a new mandatory question on the account creation form asking if you live in Tennessee. If you do, you'll be unable to register an account if you're under 18, not just the under 13 restriction mandated by COPPA. Like the restrictions on the state of Mississippi, we absolutely hate having to do this, we're sorry, and we hope we'll be able to undo it as soon as possible.

Finally, I'd like to thank every one of you who's commented with a message of support for this fight or who's bought paid time to help keep us running. The fact we're entirely user-supported and you all genuinely understand why this fight is so important for everyone is a huge part of why we can continue to do this work. I've also sent a lot of your comments to the lawyers who are fighting the actual battles in court, and they find your wholehearted support just as encouraging and motivating as I do. Thank you all once again for being the best users any social media site could ever hope for. You make me proud and even more determined to yell at state attorneys general on your behalf.

paserbyp: (Default)
paserbyp ([personal profile] paserbyp) wrote2025-08-30 03:21 pm
Entry tags:

Shadow AI

AI adoption in the enterprise is no longer theoretical. It is already happening, whether organizations are ready or not. Employees are using publicly available AI tools to complete real work. They are summarizing documents, writing emails, generating reports, translating materials, producing code, and answering questions. They’re doing this not because someone told them to, but because the tools solve real problems quickly and effectively. The people closest to the work have moved forward. The only question is whether leadership has noticed.

Bans and delays are not stopping this behavior. The tools are too accessible, and the benefits too obvious. Banning AI use at the enterprise level only removes visibility and control, ensuring that employees will use AI without security, governance, or organizational alignment. Enterprises that believe they are not using AI because they haven’t approved it are wrong. Shadow AI use is widespread. It is resulting in skyrocketing data risks, employees downloading malware in the guise of useful AI tools, and intellectual property leakage. Avoiding it through policy or silence does not reduce risk: It increases it.

The question that should be on leadership’s mind is: How do we enable people to adopt and use sanctioned AI tools across all levels of an organization?

Successful adoption begins with clarity. AI is not a tool to be rolled out in one step. It’s a capability that becomes embedded in how people work. As such, it develops in stages, with each stage building on the success of the one before it. Enterprises that attempt to skip ahead or impose top-down mandates consistently fail to generate value.

The first phase is user adoption. It is also where the most critical missteps happen.

To succeed in this phase, leadership must offer employees access to AI in a way that is secure, supported, and aligned with policy. The goal is not training; it is personal utility. Can the tool summarize a document, draft an email, or extract key information effectively? If it can, users will adopt it organically. If it requires training, installation, or configuration, they will not. If no sanctioned tool is available, they will find their own. This is the foundational phase. Without broad, voluntary use of approved AI at the individual level, no enterprise AI strategy will gain traction.

Once users find value and begin incorporating AI into daily work, the organization moves into the second phase: individual productivity enhancement. Here, AI becomes part of how people complete tasks. Drafts are written faster. Notes are summarized more effectively. Data is processed more consistently. Repetitive work is reduced or eliminated. These impact of these individual gains compounds quickly. Hundreds or thousands of users saving small amounts of time each day adds up to a significant shift in output. More importantly, usage becomes measurable. The organization begins to see what’s working, where the friction is, and which use cases are emerging as the most valuable.

The third phase is user-driven process enhancement. At this stage, users begin linking multiple AI capabilities together to complete more complex workflows. A single employee might use AI to extract structured data from a document, analyze it, format a summary, and generate a customer-facing report. AI shifts from assistant to collaborator. This phase often catches leadership by surprise. It reveals just how fast power users can innovate when they are given access and autonomy. These workflows should not be dismissed. They should be monitored, validated, and prepared for formalization.

The fourth phase involves optimization through business-driven process enhancement. AI becomes embedded in systems and workflows. It is not something a user opens. It is something the process depends on. Models support classification, triage, prioritization, routing, and forecasting. Human review becomes the exception rather than the default. Efficiency gains are no longer isolated to individuals. They are systemic. AI becomes a business capability, not a personal productivity tool. It is supported by governance, monitored for performance, and managed like any other part of the operational architecture. This phase cannot be reached unless the first three are executed correctly.

Despite the clarity of this progression, many organizations struggle to begin. One of the most common reasons is poor platform selection. Either no tool is made available, or the wrong class of tool is introduced. Sometimes what is offered is too narrow, designed for one function or team. Sometimes it is too technical, requiring configuration or training that most users aren’t prepared for. In other cases, the tool is so heavily restricted that users cannot complete meaningful work. Any of these mistakes can derail adoption. A tool that is not trusted or useful will not be used. And without usage, there is no feedback, value, or justification for scale.

The best entry point is a general-purpose AI assistant designed for enterprise use. It must be simple to access, require no setup, and provide immediate value across a range of roles. It must also meet enterprise requirements for data security, identity management, policy enforcement, and model transparency. This is not a niche solution. It is a foundation layer. It should allow employees to experiment, complete tasks, and build fluency in a way that is observable, governable, and safe.

Several platforms meet these needs. ChatGPT Enterprise provides a secure, hosted version of GPT-5 with zero data retention, administrative oversight, and SSO integration. It is simple to deploy and easy to use. Microsoft Copilot is embedded in Word, Excel, Outlook, and Teams. It is particularly effective in organizations already standardized on the Microsoft stack. Google Workspace Duet AI offers similar benefits across Gmail, Docs, and Sheets. Claude from Anthropic provides a high-quality alternative with strong summarization and long-context capabilities.

Each platform has strengths and tradeoffs. What matters is not finding the perfect solution, but selecting one that users will adopt immediately and that the organization can govern responsibly. The platform must be extensible. It must allow the enterprise to move beyond Phase 1 without needing to rip and replace. But most of all, it must be usable on day one. If the tool is not helpful, if it is not trusted, or if it cannot be accessed without friction, adoption will stall before it starts.

Phase 1 is not about pilots or proof-of-concept exercises. It is about enabling the entire workforce to gain exposure to AI in a structured, monitored way. It is about helping users discover value in their own work and allowing the organization to observe where adoption is strongest. Everything that follows depends on this foundation. Productivity gains, workflow redesign, process optimization — none of it matters until employees are using AI tools to complete real work. The faster that happens, the faster the enterprise begins to understand where to invest and how to scale.

Adoption does not begin with a roadmap. It begins with access. When users have tools that are simple, safe, and useful, they will adopt them. When adoption is visible and measurable, the organization can plan for what comes next. This is not innovation theater. This is operational readiness. Enterprises that wait will fall behind, not because they lacked vision, but because they failed to enable action.
paserbyp: (Default)
paserbyp ([personal profile] paserbyp) wrote2025-08-28 02:47 pm
Entry tags:
paserbyp: (Default)
paserbyp ([personal profile] paserbyp) wrote2025-08-26 08:20 am
Entry tags:
denise: Image: Me, facing away from camera, on top of the Castel Sant'Angelo in Rome (Default)
Denise ([staff profile] denise) wrote in [site community profile] dw_news2025-08-26 12:24 am

Mississippi legal challenge: beginning 1 September, we will need to geoblock Mississippi IPs

I'll start with the tl;dr summary to make sure everyone sees it and then explain further: As of September 1, we will temporarily be forced to block access to Dreamwidth from all IP addresses that geolocate to Mississippi for legal reasons. This block will need to continue until we either win the legal case entirely, or the district court issues another injunction preventing Mississippi from enforcing their social media age verification and parental consent law against us.

Mississippi residents, we are so, so sorry. We really don't want to do this, but the legal fight we and Netchoice have been fighting for you had a temporary setback last week. We genuinely and honestly believe that we're going to win it in the end, but the Fifth Circuit appellate court said that the district judge was wrong to issue the preliminary injunction back in June that would have maintained the status quo and prevented the state from enforcing the law requiring any social media website (which is very broadly defined, and which we definitely qualify as) to deanonymize and age-verify all users and obtain parental permission from the parent of anyone under 18 who wants to open an account.

Netchoice took that appellate ruling up to the Supreme Court, who declined to overrule the Fifth Circuit with no explanation -- except for Justice Kavanaugh agreeing that we are likely to win the fight in the end, but saying that it's no big deal to let the state enforce the law in the meantime.

Needless to say, it's a big deal to let the state enforce the law in the meantime. The Mississippi law is a breathtaking state overreach: it forces us to verify the identity and age of every person who accesses Dreamwidth from the state of Mississippi and determine who's under the age of 18 by collecting identity documents, to save that highly personal and sensitive information, and then to obtain a permission slip from those users' parents to allow them to finish creating an account. It also forces us to change our moderation policies and stop anyone under 18 from accessing a wide variety of legal and beneficial speech because the state of Mississippi doesn't like it -- which, given the way Dreamwidth works, would mean blocking people from talking about those things at all. (And if you think you know exactly what kind of content the state of Mississippi doesn't like, you're absolutely right.)

Needless to say, we don't want to do that, either. Even if we wanted to, though, we can't: the resources it would take for us to build the systems that would let us do it are well beyond our capacity. You can read the sworn declaration I provided to the court for some examples of how unworkable these requirements are in practice. (That isn't even everything! The lawyers gave me a page limit!)

Unfortunately, the penalties for failing to comply with the Mississippi law are incredibly steep: fines of $10,000 per user from Mississippi who we don't have identity documents verifying age for, per incident -- which means every time someone from Mississippi loaded Dreamwidth, we'd potentially owe Mississippi $10,000. Even a single $10,000 fine would be rough for us, but the per-user, per-incident nature of the actual fine structure is an existential threat. And because we're part of the organization suing Mississippi over it, and were explicitly named in the now-overturned preliminary injunction, we think the risk of the state deciding to engage in retaliatory prosecution while the full legal challenge continues to work its way through the courts is a lot higher than we're comfortable with. Mississippi has been itching to issue those fines for a while, and while normally we wouldn't worry much because we're a small and obscure site, the fact that we've been yelling at them in court about the law being unconstitutional means the chance of them lumping us in with the big social media giants and trying to fine us is just too high for us to want to risk it. (The excellent lawyers we've been working with are Netchoice's lawyers, not ours!)

All of this means we've made the extremely painful decision that our only possible option for the time being is to block Mississippi IP addresses from accessing Dreamwidth, until we win the case. (And I repeat: I am absolutely incredibly confident we'll win the case. And apparently Justice Kavanaugh agrees!) I repeat: I am so, so sorry. This is the last thing we wanted to do, and I've been fighting my ass off for the last three years to prevent it. But, as everyone who follows the legal system knows, the Fifth Circuit is gonna do what it's gonna do, whether or not what they want to do has any relationship to the actual law.

We don't collect geolocation information ourselves, and we have no idea which of our users are residents of Mississippi. (We also don't want to know that, unless you choose to tell us.) Because of that, and because access to highly accurate geolocation databases is extremely expensive, our only option is to use our network provider's geolocation-based blocking to prevent connections from IP addresses they identify as being from Mississippi from even reaching Dreamwidth in the first place. I have no idea how accurate their geolocation is, and it's possible that some people not in Mississippi might also be affected by this block. (The inaccuracy of geolocation is only, like, the 27th most important reason on the list of "why this law is practically impossible for any site to comply with, much less a tiny site like us".)

If your IP address is identified as coming from Mississippi, beginning on September 1, you'll see a shorter, simpler version of this message and be unable to proceed to the site itself. If you would otherwise be affected, but you have a VPN or proxy service that masks your IP address and changes where your connection appears to come from, you won't get the block message, and you can keep using Dreamwidth the way you usually would.

On a completely unrelated note while I have you all here, have I mentioned lately that I really like ProtonVPN's service, privacy practices, and pricing? They also have a free tier available that, although limited to one device, has no ads or data caps and doesn't log your activity, unlike most of the free VPN services out there. VPNs are an excellent privacy and security tool that every user of the internet should be familiar with! We aren't affiliated with Proton and we don't get any kickbacks if you sign up with them, but I'm a satisfied customer and I wanted to take this chance to let you know that.

Again, we're so incredibly sorry to have to make this announcement, and I personally promise you that I will continue to fight this law, and all of the others like it that various states are passing, with every inch of the New Jersey-bred stubborn fightiness you've come to know and love over the last 16 years. The instant we think it's less legally risky for us to allow connections from Mississippi IP addresses, we'll undo the block and let you know.

paserbyp: (Default)
paserbyp ([personal profile] paserbyp) wrote2025-08-25 05:43 pm
Entry tags:
paserbyp: (Default)
paserbyp ([personal profile] paserbyp) wrote2025-08-24 12:59 pm
Entry tags:

Microsoft or Macrohard?

Elon Musk is creating a direct rival to Microsoft through a new company called “Macrohard.”

“It’s a tongue-in-cheek name, but the project is very real!” Musk tweeted on Friday(More details: https://x.com/elonmusk/status/1958852874236305793).

The CEO of SpaceX and Tesla plans to take on Microsoft by harnessing AI. Musk describes Macrohard as a “purely AI software company” that’ll be tied to his other startup, xAI.

“In principle, given that software companies like Microsoft do not themselves manufacture any physical hardware, it should be possible to simulate them entirely with AI,” he added.

Musk made the announcement weeks after xAI registered the Macrohard trademark with the US Patent Office. Last month, he also said he was creating a “multi-agent AI software company” that would use xAI's Grok chatbot. (In 2021, he also tweeted: “Macrohard >> Microsoft.”)

The goal is to spawn “hundreds of specialized coding and image/video generation /understanding agents all working together,” he wrote. The same AI agents can then emulate human users “interacting with the software in virtual machines until the result is excellent.”

"This is a macro challenge and a hard problem with stiff competition! Can you guess the name of this company?” he wrote at the time.

So, it sounds like Musk is betting AI can replicate and pump out high-quality software, rivaling the Office programs from Microsoft, a company that's betting heavily on generative AI. Last year, Musk also mentioned his plans to use artificial intelligence to create video games.

To develop Macrohard, Musk seems to be leveraging the growing Colossus supercomputer at xAI’s Memphis facility. According to Musk, xAI will buy millions of Nvidia enterprise-grade GPUs as rival companies, including OpenAI and Meta, do the same in their pursuit of cutting-edge AI.
paserbyp: (Default)
paserbyp ([personal profile] paserbyp) wrote2025-08-23 12:24 pm
Entry tags:
paserbyp: (Default)
paserbyp ([personal profile] paserbyp) wrote2025-08-22 02:37 pm
Entry tags:
brumka: (Default)
brumka ([personal profile] brumka) wrote2025-08-22 12:57 pm

СЯУ

Слова "сенатор" и "сенильный" происходят от общего "senex" означающее "старейшина, старец".
paserbyp: (Default)
paserbyp ([personal profile] paserbyp) wrote2025-08-20 04:24 pm
Entry tags:
paserbyp: (Default)
paserbyp ([personal profile] paserbyp) wrote2025-08-19 01:44 pm
Entry tags: