Modeling the ideas and incentives of others to determine what they would want based on their mission and perspectives helps us fine-tune our own!

i want to cry

It's important to first understand how something was done and why before removing or shifting away from it; if the point isn't understood, the rework or reconstruction is likely to be just as bad - if not worse - because it will surely have an overlapping set of flawed and exploitable issues.

Good design!

If I want to express myself I should make it myself!

We build what demand exists for, which in America is a terrible local optimum - I have a car so I want a road! I have no public space so I want to own a private backyard! - etc.

I disagree with the America-centric view of this piece. It's absolutely correct that the US has homogenized too early and without culture or taste - we found so much money so late as we ascended to a world power in the past 100 or so years without setting real expectations for how a life should be lived to last. This is not to say that these same trends are not perpetrated basically everywhere in the world; new buildings, designs and businesses are homogenizing for the internet regardless of where they are or what they're doing, and the fact that American businesses have a disproportionate impact on the world stage is only part of the issue here. The internet is breeding homogenization far faster than any individual country's culture is able to.

I'm sorry to hear that someone feels this way! Eating is an opportunity to enjoy yourself three times a day, and we'll only ever be able to have a specific number of meals before death. Every meal should be valued and the food treasured. Please do not reach for the microwave next time.

Given the number of platforms that these logos are made for, companies are optimising their logos and brand names to display in a clear fashion on every and every platform.

The good: homogenizing logos emphasizes the words behind them. The bad: Everything looks the same! It's boring! Where is the expression of fashion design? Of design at all? What user testing told these people to look just like everyone else with a few letters' difference?

Is the issue the translation of the problem domain into such formulas rather than the formulas themselves? Ideally if we have some way of formulating an expression in a deterministic way s.t. it represents the problem that is also in our brains, surely we can find some satisfying assignment(s) within the constraints that classically solve the problems.

How do we avoid "being lazy"? In the thirty years since this opinion was written, "classical AI" - or deterministic symbol pushing - has seen little development and advancement, whereas neural networks and statistical methods have undergone a "winter" and an entire retransformation with the advent of GPUs and application of an incredible amount of computing power to training such systems.

What gives deterministic methods power here, then? How do we scale these to fit problems we haven't yet encountered?

Far easier to build a tight-knit, narrow system to solve problems than a loose, unproven system that can generalize. Expert systems are, most of the time, far too narrow-minded to generalize to unknowns.

That is, not coupled to neural networks or some abstract statistical methods, or to the symbol pushing of expert systems; rather, AI is a problem that has a variety of solutions, each of which suits different use cases. Example: NLP needs statistical modeling to keep up with the evolving nature of language and to accomodate slight variation in dialect. Routing problems require symbol manipulation tactics because we require a precise answer regardless of complexity, and symbols let us 'cheat'

This is what I've aimed to do with JOSS! These actions can and should be taken at 'edit time', rather than compile time or runtime. It's particularly interesting to see how the code will compile so that users can worry about optimisation if they so choose.

50 years later, they still do!

Narrative strategy is the practice of writing prose for marketing, proposing and describing ideas that are used for business purposes. It's important for these narratives to tell consistent stories, have consistent messages and personalities, etc - all drawing back to this mission or thesis that the business attempts to portray, even when this mission is evolving.

It's so, so easy to find solutions to problems that don't exist (classically: social inconvenience? let's make an app for that!) by approaching the world with a solutions-founded mindset.

It's in the interests of the shareholders to grow the company, but not necessarily that of the employees. I'd argue that the role of the employee is to sustain the underlying goals of the company by producing valuable work - and the job of the company management is to expand the potential of the company to do more and be more. To do this they hire more people to serve more roles or reallocate existing people to new positions.

Why we supposedly need this government and social contract - without a common power there is no place for industry or collaboration because there is no trust! Modern ideas of a "trustless society" fall flat in this way - there is no way forward without depending on community wellness and trust to accomplish work, no matter how hard computer libertarians work on doing everything themselves.

I wish this were viewable as an article.

Oh well: - Cameras - and thus facial recognition - is absolutely everywhere. They track people anywhere and everywhere. - 2.5 billion facial images stored at a given time in Fujian (a quite small province in China!) - Many of these cameras are present in private businesses in private spaces. - "Cameras should go in places where people fulfill their most common needs." - WeChat collects everything and can associate global data with local data. Trackers in real life imitate cellular and wifi signals to pick up cell and wifi pings, checking on phone communications. Many use proxies for real networks to indirectly track information. - Social Media -> Contact Company -> Device ID -> Look up position of device -> Apps installed, phone information. They look for Uyghur to Chinese dictionaries installed as apps on their phones to identify the minority! Frightening. T-Mobile is doing this as well... - China is collecting as much biological information - fingerprints, DNA, speech - from everyone, in order to analyze and track voice prints, voice recognition, etc. These are associated with personal identifiers, iris patterns, etc.; longer-term tracking solutions for people throughout their lifetimes.

Why do we decide to make it traumatic? Why is it important to imbue such manual labor in the pre-rigorous stage with examples and memorised formulae?

We don't need intelligence to facilitate people talking to each other > : )

I'm not sure how to feel about this! On one hand, this scales incredibly well today - we have to make all of these stupid manual customer service calls to robots, etc. Generally speaking, given that most companies have reasons to expose natural language interfaces to their problems (I may want to ask a bank a question about my tax circumstances, or how much money I have, or what benefits I'm eligible for, for example), it might make sense to expose entire APIs as simply speech act interfaces - that is, the computers can talk to one another in natural language just as easily as a human can! This makes it far easier to develop consistent interfaces, though you do lose lots of precision to the statistical models necessary to model language. (We need a predictive, statistical model because languages have edge cases, users don't always speak in a canonical fashion, and what the canonical fashion at a given time is changes over time - so we can't just bake some definitions in the model! It has to adapt).

Hey!!!

Computers are decaying...

https://danluu.com/input-lag/

This reminds me of my preferred approach to UI development. I can't get behind using design tools to completely architect solutions - I'd rather write the HTML myself and explore different elements, then pick the best ideas that emerge and continue with them! In this way embracing the idiosyncrasies of the web platform is a wonderful experience.

However, I'm not sure how we tolerate system specific peculiarities in other ways - it just doesn't make sense to rewrite every program for every device everywhere, which is why we have abstractions and standards. Business logic should always work the way it's designed!

Build technology that is acutely aware of its own limitations. How much memory does this program have access to left? How much power is this computer receiving? Will it run out of battery soon? Can we still serve this website?

Data like this can be presented to end users of the software - if broadcast over a server kind of system - or can provide immediate feedback to the computer - say, a scheduler - to decide what tasks to run when if there are many batch jobs to run queued in the background in order to optimize resource usage.

The question becomes, then, how to ensure all of this reading is done in a performant fashion and whether it's worthwhile to evaluate processes based on their resource consumption rather than performing some random OOM killing nonsense to preserve the device.

The best way to handle this really depends on the computer, but it's very clear that we can't morally build technology - or, specifically, write code - without acknowledging its resource consumption.

It's okay for documents to be documents and text editors to edit text without embellishment. I love turning animations off on my computer. They're "fun", but at the cost of what?

Why? I don't understand the fascination with it. I love squares. I love user interfaces that feel clean and straightforward and fine-lined and don't require processing my feelings at every single beziered corner.

I'm more used to interfaces like this - but they are unfortunately all I have been exposed to, really!

I do miss the old idea of opening a directory that contains some files with music in them and playing that music. A music streaming app is really just a file system browsing app that has tags (playlists) in addition to folders (artist, then album), file previews and live streaming of files from Spotify's computer to yours.

They actively work to deprecate streaming APIs. Why? I just want to use a silly ncmpcpp interface or a thing that lets me navigate lists and folders. Businesses should be allowed to do whatever they want with user interfaces, but they should allow users to make and choose their own user interfaces as they see fit.

I love the text formatting here!

Hovering is generally undesirable unless an action is triggered instantly - otherwise, it slows the user down! Hovering should only be used to hint at what a user could do with the item hovered - i.e. shading a button to make it appear clickable - it should never be used to allow a user to do things that they haven't been able to without hovering for some period of time.

Is HN a good place to provide feedback to iterate on products? It might be worth spending some more time on!

Why do applications change without informing the user?

I suspect that many of these "silly" or "fake" bug fixes are to cover up serious engineering errors - if the app admits flaws or faults, people will be less likely to use it going forward. Strict improvements are fine to post in "what's new", but bugs? Given that end users can't do anything about them, it makes sense to obfuscate them - not to avoid overwhelming consumers, but rather to save face and prevent loss of sales.

Sorry, but the angle here likely isn't to obfuscate changes from the user - at least in this specific message and others like it. It's to obfuscate frightening flaws from terrible software development practices from end users. (Sometimes, also, teams just don't know what was wrong but found a fix for it in a roundabout fashion - it crashes daily? cron job! - and I would honestly rather not know how often this is done).

The snacks!!!!!!!!!!!!!!!!!!

I love the idea of electronics teardowns. These are done so tastefully!

This is the correct take

What's happened to this culture? How do we bring it back?

Diversifying work outputs decreases reliance on resource bottlenecks, thereby enhancing productivity. Now we don't have to rely on overarching product decisions to iterate on parts of smaller products!

Every new software company that tries to "fix things" ends up taking this approach, and I'm not entirely sure it's the wrong one. The problems will absolutely not be solved directly, but through creating a unique work culture, you have the opportunity to create a community that may not be perfect, but that acts as a unique draw to the company or a lever of commitment to the organization from those who enjoy these different organizational values.

That said, for this to work those values must be very good. If lazy and tacked-on, there isn't any hope here.

Make software that feels playful but works with you as your skills develop! It should be able to "scale" to advanced, professional work from simple and silly ideas.

This website has too much - the cursor is so big and has transparency, so it doesn't show where the user can click! Absurd. I think "how much" is a poor question as well - we also need to ask what kind of humor we introduce and whether it compromises existing functionality or design inferences that users make.

Have a single customer to test against first. You can't be that customer because you'll compromise on issues for various reasons.

Don't compromise - provide your user the features that they want!

I don't really like the highlights on this website - I find them distracting rather than augmenting. I'd rather provide the emphasis myself through an external service!

If you aren't happy with the tools you must use - even a little bit - build your own! It'll be a wonderful learning experience for everyone involved - and you'll be able to build beautiful software that can meet precisely the needs of your users - all of them.

The machine code - and the disassemble command that allows the introspection of it - is fascinating! We should be able to introspect every aspect of our computers in this fashion.

Well put. All work we do is predictive, in a sense, as we can never anticipate future need - the best we can do is allow an end user agency, and this can be accomplished either implicitly with adaptive statistical methods or explicitly by allowing users to rework their products and fill in the gaps with their own ingenuity.

Yep, figures. Lojban can't predict today's English though, and interpreting language now is inherently a prediction problem. What do we do when we create a model in the present to fit an ever slightly changing future? We use statistical methods!

This is legitimate - statistical methods specific to interpreting language are really a way of abstracting over an evolving system and admitting that no 'expert system' will be able to capture and model perfectly all of the rules and exceptions in language. Perhaps if we all spoke Esperanto or Lojban we could perfectly model language, but when a rapper in vogue can alter the structure of speech overnight, it's a practically impossible task to continuously model new changes in a social construct that constantly evolves.

'Science' is the operative term here - is science the deterministic collection of facts to draw conclusions? Is it right to package up the results of experiments and experience with statistical methods to draw conclusions? This is the way experimental science has gone, so it is clearly not so far gone to use these same methods to draw more abstract, intermediate conclusions to aggregate data rather than after performing this data aggregation.

I sense that Chomsky has a more 'purist' perspective of science than many practicing, wet lab researchers have today, though - one that attains mathematical and linguistic perfection through exploring and designing human-controlled systems and using them to explain away phenomena rather than working 'experiment-first'.

Either way, statistical models are clearly necessary for evaluating models for fit with reality, especially as we recognize that we will never be able to perfectly capture many phenomena in isolation. Using these methods is as simple as reducing noise in the data we receive to evaluate our models and predictions.

How do we alleviate these concerns and this hesitation? How do we encourage people to continue to work on these things?

Is the problem rooted in missing primitives? Debugging tools? Visual editors? Or is it impossible to solve this in any technical fashion?

Find the energy for this first and it motivates you!

I'm less pessimistic and descriptive than this - if something will change then I will change it.

The service they provide is to create paywalls that require a social signal of interest to pass - no different from these cryptocurrency communities or buying that big app. You don't just pay for an article subscription or for a monkey - you're paying to participate in a subculture, to join a unique and exclusive internet community, and - in terms of the latter - you're utilizing this asset as a public display of wealth.

Article generally postulates that when everyone everywhere knows and uses everything, Internet culture is beaten down to the most basic and widespread forms of content. Today TikTok is far more important than geocities.

Now it's primarily a way for large companies to ship products, monolithic web apps and the works. Users of the web don't seem to have much agency at all!

Unfortunately, America has a lot more to ask of its people now than do European countries... the social fabric is so, so different, and radical southern and eastern cultures, religious and otherwise, and a distrust of government (as an intrusive force that potentially obscures agency) is rampant here. Many European countries don't have these issues or have not faced them in the same radical ways; they pose very different roadblocks to policy success than exist in Europe today.

What are the politics of aesthetics?

Where the game ends, the user's work begins! Users should be presented with complete systems but also feel empowered and supported to hack, to play around with the system, to dig into its guts and make it something that better suits their own needs and interests. Software shouldn't be presented as a monolithic, complete product - instead, the code should be presented as one potential realization of an idea that can be infinitely tweaked to amuse the owner of the software.

The fact that cryptocurrencies are, without regulation and by definition, pyramid schemes is frightening. At least the stock market has some basis in reality - crypto's footing is Elon Musk's Twitter account.

This is absurd! The fact that the counterclaim must be stronger than the claim itself in cases without serious consequences seems damaging and incorrect. This seems immoral esp. given modern consequences

How necessary is this? It feels like we have good enough subscription primitives already for these databases - are they not efficient enough?

What happened? Where can I find this history?

Everything can be treated as a database! We shouldn't have to worry about "replicating" the data - backups are good, but storing temporary sources of truth in some places with more permanent sources of truth in others is a little silly... what gives them that privilege and why is all the data constrained to the backend?

I will stop sharing ideas prematurely; I often don't let them bake in my brain for long enough before sharing them with others. I need to wait a day, take a nap, sleep on them - anything that gets me to wait, really - until I find the crux of the idea and can summarize it in an elegant, cogent way to peers who can provide me with feedback.

Only then will I be able to properly evaluate the value of an idea, as I will have analyzed it objectively rather than with some sort of fleeting passion in mind.

Is language the primary issue here?

I don't really understand this perspective. How was an internal language lost? Surely it was more difficult to learn the "external" language of English after immigration - but a first language as an internal monologue can remain the same without issue.

I've never felt attached to a place, to a life, in this way. How can this be achieved?

As the western hemisphere's lingua franca, English often consumes the lives of immigrants; such a mediocre language to claim so much attention!

The ability to choose writing as a career is an action that can come from a place of privilege; independent of the quality of these essays, our perspective of the immigrant experience here will be tinted by the relative, presumed privilege of the authors.

"Doff" is new to me!

Purchase only fresh, whole foods. Make large meals with these foods. Leftovers can be used to bridge gaps in meals when there is no time to cook. Fast food is simply a mistake!

Command (via imagemagick) to create a contact sheet from many images

I will reserve some time for starting to write. I only have so much time for practical software development, but I have good ideas; writing and sharing cohesive work publicly is the best way for me to share these.

This is already done regularly. For everyone who has experienced complex flexible, modern software: DAWs like Ableton start you off with a complete track in the editor window. The Pocket Operators have rhythms built in that can be played, with visual indications to demonstrate what buttons can be pressed to stimulate what sounds to allow for music making without considerable instruction. Notion starts with a series of preset pages that you're encouraged to customize.

The only software that doesn't really do this right is programming languages - and, by extension, programming systems. How do we build computer systems that encourage exploration? How to I make it easy and fun to dig into systemd, or system logs, or the JVM, or where my programs get their memory and CPU allocated? How can we design user interfaces to 1. make these processes more explicit, and 2. encourage users to play with them?

This is what seems to make so much big tech - especially Google - so inefficient - the reward systems for compensation don't align with the incentives of the business! They're used to prey upon people at the company, dangling options carrots in front of everyone ad absurdium, instead of actually aligning with the work that people do.

When booms and busts are not ground in concrete demand for resources, they're far more common - because they're fads at the whims of people with the income to invest in new technology!

This is what public transportation is for - should be for! Private company price hikes make developments like this unsustainable for the long run, and the needs of consumer aren't properly met as Uber scales costs to price out those who can't afford to move in otherwise.

How?

These vignettes are very Borges-esque - though much simpler and more straightforward in narrative.

He's a UI designer!!!

Was this not observed elsewhere? This is a record of Khan identifying some small thread of potential corruption, some danger to the empire, throughout his empire

Each is a different optimism - one is the reprieve of land, another is the freedom of sea

So Zora is just a "mind palace"? A mnemonic?

Sicily analogy

Roman, Italian, Sicilian... Islands that have lasted for thousands of years, through many conquests, etc etc.

When is this true? "Sail the farm", "Walking", etc. suggest otherwise; likely people that are missed, not city

I'm currently not sure what rewriting this requires, exactly- how tightly coupled are these companies to cloud platform? Or is this just a measure of the cost to roll new code to servers? Why not just use systemd lol?

I appreciate the approach to measurement in this article - concretely measuring the potential of businesses by quantifying cloud costs and directly weighing against evaluations and (more subjectively) development time.

Why?

Making a ditherer would be a rewarding project!

This connects to the need to learn to market yourself - if you can make something exciting to others - anything - then you can do anything, because you can encourage others to do it too.

Stay away from commitments with long time and/or financial scales - these consume more time and money insofar as they hinder social and financial flexibility. Favor fixed costs up front, minimize daily expenses, and focus on funding that directly benefits the individual without incurring (time-based, financial, social) debt.

It's interesting that so many so-called 'intellectual' workers eventually journey to explore themselves in physical space - when you think of yourself as an observer for so long, or a tool for the computer to use, or a mechanism that socializes on legs, it's difficult to exercise, to live, to feel yourself.

Daniel's mentioned that he doesn't like photos because he is "uncomfortable with his physical manifestation" - I wonder where the insecurity ends and the genuine discomfort (a sort of out-of-body but in-body experience?) begins.

Can't wait to get started with hypothes.is!

Welcome to Hypothes.is! I'm trying this out as a comment system for my wiki and am curious to see how it pans out.

This is a page note on one facet of my locally hosted wiki.

The AUR is a well-known user repository for Arch projects! This is a test note.