anomaloustho

The problem is the background is often times doing a wave motion across the screen.

Then the foreground content is doing an in/out undulation on top. So you’re seeing an undulating in/out in every possible direction + the background. And the foreground animations are all at the same time. So it’s not that we’re emphasizing any one thing. We’re emphasizing all of it.

The key with animations is in what they’re trying to draw attention to, the character of the movement, and the timing of it. You usually don’t want everything to equally animate at once.

I would: • Use background movement that also isn’t a “wave” • Stagger the timing of foreground animations so the main content is emphasized, followed by a pause, followed by the sidebars • Change the nature of the animations so they’re not doing the same essentially thing “zoom and pan” - so have the center zoom and pan, but do something different for the sides.

I think you don’t want to anthropomorphize nature, not because you shouldn’t have empathy, but because it can cause you to draw the wrong scientific conclusions.

e.g. faithful bird conclusions when they aren’t always monogamous, ants forming societies and having loyalty when they don’t, alpha wolves when alphas aren’t really a thing.

Just noticed the “whataboutism”. I don’t have a particular take on the comment above but those countries do those things in their own parts of the globe.

The government of nations is anarchy and in anarchy the only rule is that “might makes right”. Some seem to have a view that there is a world government and that there are “rules” when in reality there are none.

Why do we use polytree in this context instead of DAG? Because nodes can’t ever come back together?

It’s already been said, but most companies already have those instant “alarms” that go off within minutes. 80% of the time, those alarms are red herrings that get triaged. At a lot of companies, they go off constantly.

As a company, you don’t want to declare an outage readily and you definitely don’t want it to be declared frequently. Declaring an outage frequently means:

• Telling your exec team that your department is not running well • Negative signal to your investors • Bad reputation with your customers • Admitting culpability to your customers and partners (inviting lawsuits and refunds) • Telling your engineering leadership team that your specific team isn’t running well • Messing up your quarterly goals, bonuses etcetera for outages that aren’t real

So every social and incentive structure along the way basically signals that you don’t want to declare an outage when it isn’t real. You want to make sure you get it right. Therefore, you don’t just want to flip a status page because a few API calls had a timeout.

Isaac Arthur has put it in a way that resonated with me. If you live in a universe that has FTL, it’s scarier in a lot ways. It means the really dangerous bad guys in the other galaxy can reach you. If you live in a world without FTL, it makes things happen really slowly and over generations. That’s not as fun and exciting, but it also severely limits the amount of bad guys that can get to you.

And any bad guy that can even reach you basically means you’re already dead if they so choose.

No, because you can only randomly measure the state of your particle and therefore the remote particle. (then lose the entanglement) But you can’t put the particle into a state.

I wrote elsewhere but I’m more interpreting this distinction as “RL in real-time” vs “RL beforehand”.

Looks like they added the link. But I think it’s doing RL in realtime vs pre-trained as an LLM is.

And I associate that part to AGI being able to do cutting edge research and explore new ideas like humans can. Where, when that seems to “happen” with LLMs it’s been more debatable. (e.g. there was an existing paper that the LLM was able to tap into)

I guess another example would be to get an AGI doing RL in realtime to get really good at a video game with completely different mechanics in the same way a human could. Today, that wouldn’t really happen unless it was able to pre-train on something similar.

If I was interested in getting started with Oxide, could I even do so? It seems like this is only tailored to large enterprise sales. The only other option is just using them as your AWS replacement.

Isn’t there a different observation for why planets tend to orbit in the same direction in a solar system?

I went to give a similar answer and realized that - in the instance that the two galaxies are sideways facing each other and with the text saying “it gets brighter” not “one side gets brighter” - I’m not sure if this is the actual answer.

I can’t figure out why the Doppler Shift would make the entire galaxy brighter. I assumed it would make the rotation side spinning towards us brighter. But also redshift the side spinning away.

Yeah, I thought the reason he was in jail was because he aired ads and sponsored arenas trying to convince folks to put money into his seemingly legitimate platform that he was actually using as his personal bank account.

Side note: I’ve found that it’s never quite so binary. I can enjoy having a beer with someone and shooting the breeze, while also talking about work and jobs.

Your gardener might be the person who finds you on the floor choking and saves your life.

Your coworker who you don’t even really know might be moved to donate an organ to you.

Humans are complex.

Just to add some context. The U.S. ranks very highly in healthcare innovation, top marks in drug discovery, and highly in patient centric healthcare. The main component that you’ll see drag the U.S. down on those indexes is cost of healthcare, and that’s particularly weighed down in private insurance as opposed to Medicaid/Medicare.

Whether that cost is entirely good or bad is up for debate, as U.S. doctors get paid twice as much and perform twice as many interventions. First place countries usually have costs around 4K per person on healthcare. Other top countries around 6-7k, and the U.S. is at 10k.

But it is worth coloring exactly what we’re talking about when we say “abysmal rankings” because some folks might weigh the innovation and patient centered scores higher than the cost score.

Started a business doing 1M annual recurring. I have plenty of reasons why it’s more advantageous.

• When times get tough, cofounders playing chicken to see who can take the lowest salary

• Giving yourself the lowest pay to prevent income loss of your employees

• Always having to hunt for sales to maintain your survival (which is typically nebulous and subject to when they feel like signing)

• HR and always trying to figure out how to not get sued

• Losing big customers or big customers going out of business

For any one of those bullet points there’s a coulda woulda shoulda, but again, just taking your 400k from a well established company that isn’t fighting for its survival every day is not inherently inferior to the heroics and cortisol required to operate a business.

I really want this general concept to be true from a “practicality sense” but am having trouble seeing it.

Expressed as a mathematical expression. Given a problem that arrives at timsort with galloping mode as the solution, how would you hear the initial problem, and then express the solution as timsort in order theory?

It feels like it would be awesome to do but also doesn’t feel like the logic of the algorithm would quickly be found simply by knowing order theory.

I think another angle that often gets overlooked is that building out these tools is a competitive edge and valuation multiplier.

Harkening back to Paul Graham’s “Beating the Averages” talk about Lisp - they’ve solved problems that their competitors are also trying to solve, which could give an edge.

From the valuation perspective, for that same reason, this is now an asset that can be acquired as a technology on its own - whereas using MySQL adds no intrinsic value.

We just built a JS-frontend-framework-agnostic library because we have Vue, React, and Svelte. We found that the trade off was that it worked poorly in all of them and we couldn’t take advantage of their individual strengths.

We’ve since switched to React-based and found that, because support for React is so high, they can be incorporated elsewhere more seamlessly than “agnostic” components.

The loudest part of the company that inflicted pain on people revolved around the background checks. We had a call center dedicated to taking those calls, e.g. “You have a DUI and your landlord doesn’t allow” / “You committed a felony and your landlord says they don’t allow felonies.” - everyone was encouraged to sit on those calls to know what it felt like and have empathy.

Most products centered on whether you actually got a lease or not in the first place (and it was mainly crime related). Once you got into a lease and were struggling to pay rent, anything else would have moved to a manual action taken by the landlord if they wanted to be a soulless jerk. (Not sure about whether Realpage has some kind of automated product for that but we didn’t)

As for competing YieldStar product. Our main objection was that if we were to build price projection software, it’d better reflect the actual market. The algorithm and heuristics like “raise rent on Christmas” didn’t sit well, so we raised the conversation and stalled the project for a few weeks. Eventually we were told we’d get to revisit our concerns in the V2 but we’d be overruled for the time being “just to get out this MVP”.

I worked on a direct competitor to YieldStar and we had very high parity to YieldStar before we were acquired by Realpage. At least, the discussions at my smaller company pre-acquisition about negotiating price was that one unfortunate mis-negotiation could result in a Fair Housing incident (legitimate or not)

For example - Tenant 1 of racial profile X walks through the door and is a good negotiator. Tenant 2 of racial profile Y walks through the door and doesn’t negotiate. Tenant 2 finds out about Tenant 1 and opens a discrimination case under FHA.

At least the culture at my smaller company was to do everything to steer the rental property away from potential Fair Housing incidents. However, we did learn while working on the competing YieldStar product that the simple act of removing the negotiations caused a big knock-on effect of creating a revenue increase. That kind of put a bad taste in our mouths, because we didn’t like the fact that it wasn’t the software that was causing the increase so much as the pre-requisite of stopping negotiations. We started experimenting with how to improve the algorithm even more and if it could create bigger gains that drove more product value than simply the “don’t negotiate” effect. But we were then acquired by Realpage.

There are other ways these algorithms discriminate indirectly. For example, these algorithms tend to dial up prices around holidays like Christmas. And they do that because anyone who wants to sign a lease around the holidays has a much higher percentage chance of having some sort of life turmoil. (like maybe a family fight broke out or abuse happened on Christmas that caused someone to move out) From a business standpoint, the rental property would argue, “Someone in a bad way has a statistically more significant chance of also causing undo cost increases or breaking leases early.” — so the algorithm cranks up the prices to make up for potential costs.

Dialed up at the level and scope of industry control that Realpage has gained over the years, then you encounter all kinds of other issues.

The other perspective that these products take is they look at the short term rental industry like hotels and AirBnB. The business approaches by wondering, “Why can’t long term rentals be as technologically sophisticated as the short term rental industry. Let’s create a product that brings long term rentals automation into this decade” and the issue you run into is that short term rentals have 30x-100x the data points that you have. So it creates a gravity towards reaching into as many data points as you possibly can in order to make the product half as compelling, which includes reaching into your own internal data.

You’ve pointed out a couple times that checkmarks are not clear for checkboxes. This statement feels a bit off in the sense that it’s like saying you shouldn’t put “cookies” in a “cookie jar” or any other item inside of a container named for item - could you elaborate on why checkmarks are not intuitive when placed inside of checkboxes?

Just because this is such a pervasive idea. Is it actually true that sugar correlates to any kind of hyperactivity? In the past I have seen studies concluding no correlation between sugar intake and energy. Also anecdotally, a few teaspoons of sugar has never given me the ability to run faster, or a little bit longer. It’s never made me become more active or wired. If it did, I’d be more prone to strategically use it to “gain an edge”.

That is to say, I’m even a victim of joking about this with other parents. But I do because I felt that it is a pervasive “truth” and I encounter no human that is willing to entertain the contrary. To the same extent, I’ve also never witnessed any adult sneak a tablespoon of sugar to get that hyperactivity boost.

Fat is also more calorically dense than gasoline. But I’ve also never seen someone bounce off the walls after consumption.

There are 2 different thoughts here I think. If using GitOps in Kubernetes, then application and set up (Pods) aren’t associated with Nodes (EC2). And both can be torn down and rebuilt without state issues. When state is required, then PVCs and Stateful Sets come into play.

For managed services like S3 and RDS, there are other GitOps tools like Crossplane.io which you can use for similar GitOps management. But the paradigm shift might also be that you add GitOps config to perform regular backups, and also add config to ensure that if it is being recreated, it restores from a backup.

I’m having trouble determining the effectiveness here. The article repeatedly uses the phrase, “no increased spending on temptation goods”.

I’m reading that as, “If the person had an addiction to meth, they still spent the same amount on meth, but they did it from the comfort of a new couch, in new clothes, in a new apartment. Sometimes they spent less on meth.”

Is that a fair interpretation?

Also, how do they know the person is spending money on temptation goods or not? Self-reported?

On mobile, about 70% of the screen real estate is dedicated to searching and filtering UI. At the bottom 30% of the screen, I can see 1 headline for an article.

When visiting HN about 90% of my mobile screen real estate is dedicated to content and I’m able to see the first 13 results.

Zero SR/F owner here. I was also put off by this before buying a Zero. They did back off of this a bit, but most of the stuff floating around online is from when they initially did it.

Now, if you purchase any premium Zero like the SRF and SRS, every software upgrade comes unlocked and is apart of your purchase as you’d expect.

If you purchase their cheaper tier models like the DS and SR, then the “upgrades” can be software unlocked. I’m not a fan, but it is a one time fee from what I understand. It’s not a subscription. (correct me if you have conflicting data)

That in mind, my SR/F came off the lot with everything already enabled. I checked the app and there are no “upgrades” that can be purchased.

Zero also had a reputation of being one of the less repairable bikes, but they’ve been slowly going in the other direction.

I walked into a dealer being dead set on buying an Energica bike and not buying a Zero. They had no problem selling it to me, but after updating me on where the current state of Zero was vs Energica, I decided to buy the Zero and didn’t regret it in the least.

My guess is this exploits an AirBnB loophole that allows a host to remove a review if the guest never stays.

Small technicality, but the review asks you to rate how clean the place is, what condition it’s in, etcetera — you probably wouldn’t have been able to rate those honestly having not stayed.

Some of this rides a thin line in the sense that you made it far enough through the booking to have technically left a review, but wouldn’t have been able to if you had the dispute earlier. In other cases (ones I’ve personally experienced), I’ve had hosts who were jerks and I canceled before ever getting close to the check in time. In that case, you can’t leave a review on AirBnB or Turo. The only punishment to the host is lost revenue and lost opportunity.

At the very least you know the host lost revenue for this. It can be difficult to get a new booking in such short notice.