underwater

The state of recipe sites is 100% driven by Google. Those sites add the fluff because that is what Google rewards. The actual recipe quality doesn't matter. If I use my example, the page contains:

* Bio of the author - Google SEO rewards content written by "real humans", so this is there to appease the bot.

* Giant images of overnight oats - This helps the page rank highly on Google Image Search.

* Fluffy paragraphs - "The beauty of overnight oats is that you can make them as simple or creative as you’d like. The base recipe is delicious, and filling all on its own. But if you’d like to spruce it up, you can add a variety of toppings and mix-ins, including fresh fruit, nuts, seeds, spices, jams, and more". Is that supposed to be text written for other humans? Google rewards original content, which means recipe sites need to include massive paragraphs of filler to appear like an authoritative source.

* Ads - Gotta make the money.

* More fluff "Here’s a few reasons why you should to whip up this recipe today…" - Again, this is not text written for humans.

* Repeat the fluff, pictures and ads approximately 5 times.

* Common questions "Do you have to use yogurt?" - Trying to appeal to Google Instant Answer searches.

* Recipe - Finally!

* Recipe rating - This allows the site to put a star rating and review count on Google, again, for SEO

So, 95% of the content and functionality of that page is an attempt to convince Google into sending traffic, or ads generating revenue, rather than serving the end user.

Content on the web has dramatically changed from 20 years ago because publishers are creating content for a broken machine and not for other humans. This negative feedback loop is hurting the web, and not just Google. This leaves the door open for upstarts like ChatGPT or TikTok to gain mindshare.

No, Google results are trash. Here are three examples from this week:

- I Googled "speed work running", wanting some workout suggestions. I visited four or five results that either didn't have concrete suggestions, were poorly written, or overly verbose. I ended my search with little faith that I was getting good advice. I typed the same thing into ChatGPT and got a suggestion of 6 different workouts that all seemed coherent.

- I asked Google Home "How do I make overnight oats". It replied, "I found a result for [...], should I read it?" then "The first ingredient is Nutella". That's it, that's all it said. I tried the same search on the web, and every result Google result was spam that was 20 paragraphs of inane blog content with a recipe tacked on the end. Again, ChatGPT then gave me a sane, baseline, recipe.

- I searched Google for "how much caffeine is in coffee". It gave me a calculator that said, "40 mg" and then a suggestion an alternative search for "Q: How much caffeine is in an average cup of coffee? A: between 80 to 100 milligrams". It turns out the calculator was normalizing the caffeine content to caffeine per 100mg of coffee.

I'm using Google exactly how their product is training me to use it, but the failure modes are all consistent. Google's AI features has not real understanding of the world, so their instant answers are frequently nonsensical. And Google clearly can't filter out SEO garbage.

Even if they could filter out SEO garbage, Google's early success killed the golden goose. 99% of web publishers are publishing content for Google, not for their readers and web publishing has become a cynically commercial affair. Individual publishers have by and large moved on from the web to other creator-focused platforms. So, the quality and experience of web content is absolute rubbish. Results are filled with cookie banners, ads, signup prompts, verbose SEO filler, poor writing, lack of authority, etc.

What a weird take. Religions ability to influence behaviour works because of our lizard brain, not despite it.

The idea of God judging people from afar is an extension of the important humans place on social conformity. We are hard wired to care about what other people think of us. If you remove the religious decorarions then that underlying trait still exists.

Establishment art is just technical proficiency mixed with good marketing. In fact, you can do what artists like Jeff Koons do and just skip the skip the technical proficiency step. I don't see how, at its core, it is any different from what this TikToker is doing.

That was the point the author was making. His popularity has come completely from outside the establishment.

I'm not surprised that this is the case (I've made that same decision myself). But there are some apps that I just done understand: airline ticketing apps, theme park apps, etc. How is handling offline usage not ofne of the first requirements.

Serious runners are a whole other thing to a casual runner. I did a 20-30k per week for a decade without problems.

When I doubled that - which required a very intentional effort on my part - and ended up with injuries. But was able to address the underlying issues without too much drama.

> Every API call was so form fit to the UI that nearly every UI tweak needed frontend and backend changes.

Having to change backend code is not necessarily a bad thing. Just don't structure your code as two seperate codebases built by two different teams.

Am I missing the sarcasm? If your brother in law is relying on Uber, then employment protections would mean he doesn't lose his sole income unfairly and without notice.

Uber argue that most gig workers are just doing it as a spur of the moment side hustle, and therefore the flexibility is necessary. But your anecdote is the opposite.

My beef with Maps is the UI. It's so skittish and modal I'm constantly fighting with it.

If I position the viewport ans run a query, it unhelpfully searches a different area. If I tap on a pin it moves the viewport again. If I dismiss the pin info it moves yet again to an arbitrary position. Same for a while other bunch of interactions.

It feels like a UX that thinks it's clever but doesn't respect me as the user.

Htmx is essentially how Facebook was built before React. Everything was server rendered in PHP and you could use special HTML tags to trigger partial updates. Ajax requests would return HTML fragments that would be injected into the correct part of the page.

It worked fine for simple use xases, but it was someone brittle and began to fail as soon as you wanted more than basic interactivity.

If not Doors, then what about Windows?

Some people think that Hyperloop was purely intended to sabotage the high-speed rail project by creating the false impression there was a cheaper, easier, option on the horizon.

The Avengers films did exactly that. They filmed characters with random costumes because they hadn't done the designs yet. Then replaced the costumes with CG later.

$8 is cheap. The relatively high cost is only because the final analysis wasn't run on more passages.

Exactly. Apple has a track record and fanbase that means they can sell a $3500 device. Meta know that even if they could build this device, there is zero chance anyone would buy it. They have to start with the low price point to make up for the lack of branch cachet.

It's the same as a one of those games with the wire and the buzzer, just with a different interface.

I've seen this backfire when someone breaks the configuration of the cache key. All of a sudden people start getting back usernames and personal information of random users.

MG is a Chinese brand reusing the branding of the old British version of MG.

Over the last half a decade they've become the fifth biggest brand in Australia ahead of Mitsubishi and Hyundai.

The mantra for this is "perfect is the enemy of done."

This is great. I'm relatively comfortable with command lines, but I think people who reflexively respond with "RTFM" are completely wrong. There are few other parts of my life where reading a manual are necessary to start using a product. Being apathetic to users' needs (or deliberately hostile towards users) is frowned upon in almost every field.

You should check out https://tldr.sh/ It gives you practical examples for common use cases.

Let's apply the same logic to people then. "John scammed his neighbour of his life savings, we don't need to lock him up because lots of people will refuse to associate with him or supply him with goods and services."

Making loss of business or minor fines the only mechanism for correcting behaviour, means that the leaders can view antisocial and unethical behaviour as a cost benefit tradeoff. And the lack of personal accountability means the company leadership has very limited downside, even if they completely screw up.

That's weird because the article also says the plaintiffs argument was that the autopilot drove the car I to the curb, not that it disengaged.

I don't see how a company can just have a blanket disclaimer that says a product can fail in a fatal way, and be in the clear.

When marketing you should be focused on your customer's problems, not your product.

What is it worth my time looking into this in depth? All you've said is that it's the best and it's got a good API. What's the hook?

Putting the same text prompts into each system seems like a flawed methodology. Instead they should have a general goal and tweak the prompt for each system.

After all, directives like "Warm glow, elegance" or "skeuomorphic style" are hacks that people have learned to use to specifically improve MidJourney's results.

If you double productivity then bosses will double the workload. I don't know of any software team that is in danger of completing their backlog.

No decent mobile app, no groups, no Messenger, no Instagram, no WhatsApp, no real advertising product. It isn't 35x better, but there is no way the product offering is worse.

The Facebook experience feels worse, but that's because the novelty wore off and people don't share anywhere as much as they did back in the heyday.

I was at Facebook shortly after this. Zuck was very present in internal comms, regularly posting on the internal Facebook and holding a weekly Q&A. To me it felt transparent and authentic, and I felt aligned with company strategy.

This email was famous, the clickbait subject gave some people a fright, but overwhelmingly people were strongly aligned with the sentiment and spoke about it in a positive way.

Now at another big tech co and I hear from my CEO and CVPs via mainstream news instead of internal channels. I feel very little connection to the culture and have no idea what our strategic direction is.

Right. Kids who answer the “wrong way“ from top to bottom got answers 1 and 100 incorrect, but correctly answered questions 2 through 99 and score 98/100.

Kids who read the last question and leave the rest blank get two questions correct for a score of 2/100.

On a typical test the blast radius for a question is limited. There is no convention that getting a question wrong alters the outcome for other questions, too.