> AI is absolutely imperfect, just like teams of human developers.
The old "nothing's perfect therefore everything is equally imperfect" fallacy. It's not a binary. While everything is flawed, somethings are more flawed than others. Welcome to the world, it's complex.
> when we talk about AI code, it's always compared against some idealized
Have you ever seen a development mailing list? Seems like when human code is scrutinized it's held to a high idealized standard. "Technical debt" is a concept that originated from looking at human code. How then can it be true that applying it to AI code is setting a higher standard? It's setting the same standard. These things existed and were applied to human code before AI exploded.
The whole "many eyes" thing can be quite brutal. We don't always get the many eyes but when we do... wars are waged. It's brutal out there for anything getting scrutiny. Currently, AI code getting a lot of eyeballs. That's a good thing. Don't wish it away just because you're butthurt about your new pet tech being held to a standard. That's how it gets better.
The "problem" of AI being held to a supposed higher standard isn't a problem. It's a free pony.
I don't understand what you think should happen here. I honestly don't think that the EFF has shifted nearly as much as people have in this hyper partizan environment. The trouble with being "center" is that you get pulled around by the most extreme. The flag is tied to the center of the rope right? If the right pulls away should the EFF compromise their values just to be seen as less leftist?
When being the center is the principle value, you stop being defined by your own values. If you're the flag, you don't get to have a say. One side could hook a tractor up to their end of the rope. The flag has no agency.
Do I think the EFF should have more outreach to the right? Sure. But that outreach can't be: we compromised our principles to chase the moving target loosely defined as "the center" of the moment.
Of course the EFF had more allies on the right during the Obama years. They were suing the Obama administration! There is always going to be a nontrivial amount of tribalism going on. How do you think suing the Trump administration has affected the left? They are eating it up!
No, the EFF should stick to their principles and try to pull people out of their tribalism rather than cater to it. Suing the "your team" administration should not automatically be seen as "look how other team they are!"
That's the thing: They choose leftist allegiances over their ostensible job. For example, search for "twitter files" on the EFF's Twitter account. Nothing. Blatant government effort to censor people. Zip, zilch, nada.
Now they abandon X that's become more free, and head for Bluesky and Mastodon, which are basically recreations of the stifling atmosphere of pre-Musk Twitter.
Freedom for their favoured people to do what they like, perhaps. But for me and others? Nah, not on the program.
It's obvious to me that EFF should sue every administration. I started working at EFF during the first George W. Bush administration, and I worked there throughout that administration, the Obama administration, and the first Trump administration. EFF sued all of them, I worked on all of those cases, and I supported all of those cases.
As you correctly point out, people who liked each of these administrations were often unhappy when we sued them, and often assumed that we were politically biased against them. I always encountered people who effectively said "my party is using power appropriately for good purposes, and you should not question how we use power; that only helps my opponents". Part of the civil liberties framework and something that EFF has done well (including since the time it's become significantly left-leaning) is questioning how every administration uses power.
So, I'm absolutely not suggesting that EFF should praise or celebrate the Trump administration or not sue it.
> No, the EFF should stick to their principles and try to pull people out of their tribalism rather than cater to it. Suing the "your team" administration should not automatically be seen as "look how other team they are!"
Seems inconsistent to say "truth be told" before "probably". These sorts of vibes based statements cloud conversation and confuse the scientifically illiterate who are quick to attach truth to belief.
I'd be happier if they stopped using percentages. When percentage points don't consume close to an equal amount of time, don't use them. It's just complete nonsense at that point. It doesn't give the user any useful information just false hope that it might finish soon.
Of course if Windows update wasn't so horrible maybe it wouldn't matter as much.
> companies need to attract customers - i.e. make their lives better.
Companies need to sell the *idea* of making lives better. They don't actually have to make a customer's life better and many don't despite selling lots of product. Profitability is not a strict function of quality. If it were, we'd not give the slightest fuck about monopolies. It takes a great amount of effort to get companies to compete on "making lives better." If bettering lives is not the most profitable it is very often eclipsed by what is.
It's much more complicated than the simple lie that "you get what you pay for."
No, but it's very probable li2uR3ce bought things that were portrayed as being much better than they actually were. These days "profitable" for the seller rarely equates with "good value" for the buyer.
Feels like a false dichotomy? Brand loyalty fluctuates over time and by brand, market, etc. For example, cars are probably the largest repeated purchase in America, and car brand loyalty is nowhere as high as it once was:
People get convinced in the heat of the moment that they want something, when if they had time to reflect/research/etc. they might very well conclude that their money is better spent elsewhere. Marketing in a capitalist system is highly motivated to make you think you want or need something more than you actually do.
One tactic that some people find helpful when they want to buy a discretionary item is to wait a few days/weeks, and make sure that it's not just a whim that they'd regret.
Watch TV commercials over the years. The older ones are laughable. They steadily get more sophisticated as the years go by. That's because the customers get more sophisticated, too. You might call it evolution in action :-/
Political campaigns also get more sophisticated. Scott Adams has an interesting podcast where he breaks down the latest manipulative persuasion techniques used.
Hostility, it turns out is "engagement" which is profitable. We went from people organically paring themselves to algorithms steering people towards "engaging" encounters. Where we used to say "don't feed the trolls" we should now be saying "don't feed the algorithm."
I suppose we shouldn't feed either one but right now "the algorithm" is the bigger threat. I mean, I can usually get a troll to work for me with a bit of patience and perseverance... algorithms? They'll only work for me if it's profitable for someone else. It's so soulless.
I don't think that's true though I know a lot of people on HN do. People just respond to ragebait. Look on here and on Mastodon. Angry posts still generate the most engagement. I mean, forums used to have everlasting hellthreads that were so bitter that owners would cordon them off onto mega 300 page long threads. Usenet had and still has hellthreads and individuals can choose how they sort their "feeds"!
But your post is pretty typical of the viewpoint you see in these technical circles. I know some non-technical small business owners and while dealing with "the algorithm" can be stressful, many of them respect the power of the algorithm to generate awareness of their products. That's what I mean. Niche spaces online have become tribal and niche just like they used to be when they were offline. The brief open period of the web was just the time period when the old ingroup was joined by everyone else.
Altering the deal is SOP where law permits it. Read any privacy policy or user "agreement" and note that they all have "Darth Vader" clauses where they get to alter the deal when ever it suits them. It's even your job to spam refresh on the page and use a diff tool to figure out what changed.
The difference is that these customers potentially can defend themselves with lawsuits and switching to a competitor.
In a sense, those clauses seem like the embodiment of a best alternative to a negotiated agreement. They reserve broad leeway to account for possibilities that might not have been anticipated; for the fact you can never anticipate everything. They are tolerated to the extent to which they are applied conservatively for this implicit, reasonable purpose.
The problems arise when some cowboy executive comes in and reads the letter of the law without regard for the spirit of the law and starts thinking they can get away with radical innovations that destroy the stability of the business ecosystem. Business partners in a contract may grant broad leeway conditioned on the trust that it will not be abused, but now that trust has been destroyed, how can you continue to do business with such a counterparty?
Despite the gutting of desktop usability for the sake of being mobile compatible, many environments still have zero presence on mobile. Can I PLEASE have my fucking scrollbar back now?
I once spent a lot of my time and the time and the time of a developer trying to find a setting because there was no indication that a window had more content (a checkbox) to scroll down to. Something that would have been obvious before the onslaught of hidden scrollbars. The trouble is that having my pointer over the navigation pane--practically a guaranteed position--causes the scrollbar on the other pane to be hidden. Without the visual cue of a scrollbar there was no reason to move my pointer over to the other pane to discover there's more. Hell you might not even know it's a separate pane now that we've gotten rid of every defining border. I shared a screenshot with the developer, assured them that I was using the current version, only to have him say "scroll down." No doubt, I'm the fucking idiot (/s).
Just like on mobile, you're supposed randomly interact with every UI element in hopes of discovering how it works only to have that learned skill be unique to one fucking app. Tap it, slow tap it, slow tap it for a different amount of time, tap it faster, spam it... "google it"... oh, this time you're supposed to drag it to something that doesn't even look like a UI element. Stupid grandpas!
"Is the checkbox checked?" was never as ambiguous as "is the slider switch on?" Also, the checkbox uses less screen space! I'd argue that they optimized for neither screen space or user friendliness. It's optimized for a look and you can even make it worse by making it flatter. Go ahead make it look like two squares! Is the darker area the switch part? Who cares! It looks so clean and distraction free! I was so distracted by knowing what state the switch was in.
Sorry, time for my meds. I usually make it half way through the day.
I feel you. There is nothing today quite like _Inside Macintosh Volume I_ laying out out in drawings and prose (!) what the elements of a UI are, what they do, and how they work... almost like people had never seen a UI before. People must have had to think about it, they wrote a book for F sake. /s
They devote ink and paper to declaring that modes are to be avoided, and why. The do's and don'ts of UI on page 70 are worth repeating:
Do:
* Let the user have as much control as possible over the appearance of objects.
* Use verbs for menu commands that perform actions.
* Make alerts self explanatory.
* Use controls and other graphics instead of just menu commands.
Don't:
* Overuse modes (again!).
* Require keyboard / mouse when the operation would be easier with the other.
* Change the way the screen looks unexpectedly, especially scrolling.
* Redraw objects unnecessarily.
* Make up your own menus and give them the same name as standard ones (they define the standard ones in this book, you know: About, File, Edit as well as what goes in them. yes, yes, this is where it all started).
We've meandered into a bullshit local minimum where there is the One True UI and it's different for every app, but the same for every user. Meanwhile in industrial control where a $50,000 piece of equipment has its own app, used by maybe one person or three if it's operated 24 hours a day, the responsive mobile interface is as easy to lay out as slides in a slide deck and takes about as long to do. Hell, a customizable dashboard is a widget.
If the cloud made shoes there would be different shoes for grass and concrete, but they'd all be the same size and you'd have to cut off toes if your feet were too big or stuff them with prostheses if they were too small.
"In Greek mythology, Procrustes ... was a rogue smith and bandit from Attica who attacked [read: killed] people by stretching them or cutting off their legs, so as to force them to fit the size of an iron bed."
Eh, Apple's ideas are not to be taken as gospel. This is the company that fielded UI that's as bad (or worse) than everything we're complaining about here, decades earlier.
For example: secret alternate menus. You can actually press modifier keys on the Mac while a menu's open and sometimes you get totally different menus. These are not indicated anywhere. So theoretically every menu on a Mac may have... let me do the math here... eight sets of contents using Control, Option, Shift and all combos of those. So according to Apple, you should open every menu and mash every combo of modifier key to see what's in each... and memorize them.
Another Apple menu defect is to start every entry with the same word:
VIEW
Show meters
Show clips
Show this
Hide that
Hide the sense
Show WTF the point is
This makes the first word of every entry nearly useless, and massively degrades the usability of the View menu. You have to sit there and parse the first part of every line, which only has two options... both of which are four characters, BTW, and thus visually the same size.
You don't do this; you use CHECKMARKS, which we learned decades ago. Some Mac apps do this, but many Apple ones still have this asinine convention of "show" and "hide" repeated over and over.
VIEW
• Meters
• Clips
This
• That
• The sense
WTF the point is
But that brings us to another classic Mac menu defect: The misuse of the Window menu. This menu is supposed to show names of open windows in an MDI-type situation. But Mac apps often bury View options in the Window menu, apparently expecting the user to guess that whatever they're looking for has been implemented as a window. Why would I go into the Window menu to activate audio meters, for example?
And most of the time, whatever the option is has NOT been implemented as a window.
The way Microsoft office handles this is pretty wonderful. Alt reveals the keyboard shortcuts and they work in a very intuitive way. All pretty discoverable once you learn the one "alt" key trick.
(Side note, for mouse-work, They're now starting to f** with their ribbon a bit and the hidden/simplified version is just terrible, but they haven't forced that upon us yet.)
To handle the lack of vim modes in other apps, I just use a keyboard with my own custom QMK firmware that makes my keyboard have modes. That works well enough for 90% of what I'd be doing in VIM.
Microsoft in particular seems hellbent on putting touchscreens in everything. They REALLY want to make it happen but seemingly no one is having it. There's ONE person I know who owns a touchscreen laptop and genuinely touches its screen on purpose.
On your scrolling complaint — well, there's this desktop-specific thing that OP doesn't mention but that becomes very apparent once you start looking for it in old-school desktop UIs. Controls never, ever scroll. Only content does. If controls don't fit into a window, you don't make it scrollable — you split it into tabs or you put the extra controls into a separate window that opens via a button. This appears to be universal at least for Windows and macOS.
I've had touchscreen laptops for about 7years now. The only time that the touchscreen gets used is when somebody is showing me something, doesn't realize it's a touchscreen, and accidentally borks whatever we were looking at.
It's strange - I love to optimize my flow, but I just haven't yet hit a situation where my hands leaving keyboard and touching the screen is faster. (I'm a die-hard and proficient track point user,and my hands never leave keyboard, so maybe I'm a special weird case?)
> Microsoft in particular seems hellbent on putting touchscreens in everything. They REALLY want to make it happen but seemingly no one is having it. There's ONE person I know who owns a touchscreen laptop and genuinely touches its screen on purpose.
When I was doing a lot Android development, it was nice to be able to test touch without loading on a phone. It was equally nice to be able to markup screenshots with the pan. I switched to Linux from Windows a couple years ago and the new laptop doesn't have touch (it does have a giant 17" panel and all day battery), so I keep a tablet that has a stylus in the bag for doing UI markup and bug reports. I may switch to the 16" Gram because it does have touch and stylus... and I won't have to carry the tablet.
I have to admit that using a touchscreen laptop with a decent pen has made my university life a breeze. Ability to whip out a laptop and have all the notes, from all the years, searchable, with drawings just as good as when they were drawn and even better, because I can fix them and annotate them anytime... makes me feel like a terminator machine.
And it is the main reason I am stuck with Windows, despite also enjoying the 'normal' pc experience on linux much more than Windows bloat.
> You actually want autolayout, so that different languages only need translations and otherwise don't need almost any extra work.
What kind of extra work DO they need? I can only think of RTL layouts but I don't remember whether win32 "dialogs" loaded from resources automatically mirror the layout for RTL languages.
> Linux has, I think, proved, that fixed window sizes are needless.
Desktop Linux is an unfixable dumpster fire UX-wise. Don't even get me started.
A serious problem with many open-source GUIs, but especially those on Linux, is that they're built backwards: you first write the code, then build the UI. Your UI ends up being shaped by the underlying implementation of the thing it controls. When in reality you want to do it the other way around: you'd formulate user requirements ("they need to be able to do X and Y"), you'd think through all possible scenarios that the UI must accommodate, you'd make a rough outline of what a UI satisfying all these requirements would look like, and only THEN would you start actually writing any code.
Slide switches are just awful. A skeuomorphism that doesn't work. Checkboxes are far better, assuming there aren't double negatives in the label text (e.g. a check means something is disabled.... the developers who do this should be shot at dawn).
I was with you, until you got to the slider switch part. Now I'm with you AND my blood is boiling. Nice to know it's not just me though - I find the ambiguous slider switches in way too many apps now!
And what the fuck is with the wording on check boxes these days?
On top of not even knowing if the sliding circle is actually on or off, half the time I cannot figure out which of on or off I actually want. Double negatives all over the place. Weird wording. No indication of actual impact.
> I once spent a lot of my time and the time and the time of a developer trying to find a setting because there was no indication that a window had more content (a checkbox) to scroll down to.
We've got some innovation from having everyone invent their own user interfaces for every app, but it's been a very high cost to pay vs standard OS GUIs.
The old "nothing's perfect therefore everything is equally imperfect" fallacy. It's not a binary. While everything is flawed, somethings are more flawed than others. Welcome to the world, it's complex.
> when we talk about AI code, it's always compared against some idealized
Have you ever seen a development mailing list? Seems like when human code is scrutinized it's held to a high idealized standard. "Technical debt" is a concept that originated from looking at human code. How then can it be true that applying it to AI code is setting a higher standard? It's setting the same standard. These things existed and were applied to human code before AI exploded.
The whole "many eyes" thing can be quite brutal. We don't always get the many eyes but when we do... wars are waged. It's brutal out there for anything getting scrutiny. Currently, AI code getting a lot of eyeballs. That's a good thing. Don't wish it away just because you're butthurt about your new pet tech being held to a standard. That's how it gets better.
The "problem" of AI being held to a supposed higher standard isn't a problem. It's a free pony.