Share good ideas and conversation.   Login, Join Us, or Take a Tour!
veen's comments
veen  ·  5 days ago  ·  link  ·    ·  parent  ·  post: Pubski: February 13, 2019

Well at least now I know what Hyundai named its car after! Hawaii seems like a magical place.

kleinbl00  ·  5 days ago  ·  link  ·  

Dude. Kona coffee is something else.

veen  ·  10 days ago  ·  link  ·    ·  parent  ·  post: Pubski: February 6, 2019

We had few shenanigans this time around! We’re a very RP-heavy party, and most of this session was about one of us revealing their Tragic Backstory(TM).

Someone did carve the head of a statue off, and tried to distract a gigantic spider with a mirage though.

veen  ·  12 days ago  ·  link  ·    ·  parent  ·  post: Pubski: February 6, 2019

Hey Pubs. Wednesday eve, Thursday is only four minutes away. Just came back from a fun DnD sesh. Ski lessons are going pretty well, although it's been occupying the 'swim' place in my list of priorities, so I haven't gone swimming in a while. Gonna fix that tomorrow.

Work's at a good pace right now. I've been automating and programming a bunch, and there's new interesting things on the horizon. I'm not too busy and am pretty good at keeping it that way.

Tomorrow we're going with fifteen colleagues to a pétanque bar, which like the name suggests is a hip new bar where you can also play pétanque like you're on a French campsite or something. The invitation had tactically placed Comic Sans and there's bound to be some wine so I think it's gonna be fun. I'm hoping it's every bit as cheesy and hip as I imagine it'll be.

KapteinB  ·  12 days ago  ·  link  ·  

So what shenanigans did your D&D group get into? :-)

veen  ·  10 days ago  ·  link  ·  

We had few shenanigans this time around! We’re a very RP-heavy party, and most of this session was about one of us revealing their Tragic Backstory(TM).

Someone did carve the head of a statue off, and tried to distract a gigantic spider with a mirage though.

veen  ·  12 days ago  ·  link  ·    ·  parent  ·  post: Pubski: February 6, 2019

Dang, what do you get out of the book? Model names, prizes, complications?

I would totally be interested in the average price and standard deviations per complication, maybe regressed to how expensive the brand is in general. Hmu if you need some data science help!

veen  ·  12 days ago  ·  link  ·    ·  parent  ·  post: Cal Newport on Why We'll Look Back at Our Smartphones Like Cigarettes

I think it very often looks that way because lookalike audiences have gotten really good. The real issue is that many, many websites have some analytics connection to Facebook, so ads have a lot of data to target you with.

veen  ·  12 days ago  ·  link  ·    ·  parent  ·  post: Tech Is Splitting the U.S. Work Force in Two

I feel a bit torn on the subject of automation. There seems to be a very long list of boring, dangerous and demanding tasks that I can see becominig automated in the coming decade or two. Matter of fact, I have been busy the last weeks with automating a bunch of repetitive webapp tasks that can be fully automated with a REST API. It'll almost save my company (so, me and the four colleagues now doing this work manually) months of dull manual work in the coming years and I'm all the happier for it, since we can then use our hours to do more interesting, value-adding work.

So I tend to extrapolate that experience to the bigger scale - automation will free us from the tasks we don't have to do so we have more room to tackle the bigger, more difficult issues. But I am also very aware of the problem that it's very often not the same person that gets to do the new, more interesting thing. My experience may be good for me, but the average automation case is that a dozen low-skilled workers are replaced by one higher-skilled worker, and that nobody gives a crap about those that are left behind. Retraining only gets you so far - who's gonna hire the 50-year old retrained-but-unexperienced worker over the cheaper younger person, for example?

I don't know if this already happened on your side of the Atlantic, but I see more and more fast food chains doing away with people behind the counter for orders, and having instead large touch-screen based ordering. Still don't know if I should cheer that on or not.

user-inactivated  ·  12 days ago  ·  link  ·  

    Retraining only gets you so far - who's gonna hire the 50-year old retrained-but-unexperienced worker over the cheaper younger person, for example?

Not only that, but the older someone gets the more obstacles there are to overcome. Ageism being the first to come to mind, but there might be other problems as well such as work restrictions due to old injuries, poor credit score, previous run ins with the law. All of these add up and can greatly diminish a person's chances of finding new employment. Just by playing the odds game, the older someone gets the more likely they're going to have issues like these happen in their lifetime.

    I don't know if this already happened on your side of the Atlantic, but I see more and more fast food chains doing away with people behind the counter for orders, and having instead large touch-screen based ordering. Still don't know if I should cheer that on or not.

The common narrative here in The States is that jobs such as fast food or retail aren't "good jobs." They don't pay well, there's not a lot of room for advancement, the skills learned don't necessarily transfer to other industries, what have you. Therefore their loss shouldn't be mourned. The flip side of that coin is though, those jobs help put food on the table, a roof over your head, and a reason to wake up in the morning. Even if the wages might be inadequate, all of those things are arguably good. If we take those jobs away, replace them with automation, and don't offer any safety nets for the people who've jobs are lost, aren't we essentially robbing the world of the goodness of food, shelter, and purpose?

Additionally, what kind of message do you think the world is sending these people when all it takes is a $200 tablet to threaten their ability to be employed? How do you think they can end up seeing themselves, both as individuals and as members of society?

veen  ·  12 days ago  ·  link  ·    ·  parent  ·  post: Cal Newport on Why We'll Look Back at Our Smartphones Like Cigarettes

Replace "Messenger" with "Whatsapp", and you have about half the globe with the same problem. Literally all my communication since like 2015 has gone through WhatsApp for me and everyone I know, replacing texts swiftly. And since Facebook bought WhatsApp, it's effectively also Messenger. Potato potato - effectively, there's no escape from Facebooks grasp. Even if you don't have an account, Facebook profiles you.

ffatty  ·  12 days ago  ·  link  ·  

What does everyone think of rumors that messenger targets advertisements based on private conversations?

Facebook denies it but there's a lot of anecdotes (my own included) and experiments on YouTube that say otherwise

veen  ·  12 days ago  ·  link  ·  

I think it very often looks that way because lookalike audiences have gotten really good. The real issue is that many, many websites have some analytics connection to Facebook, so ads have a lot of data to target you with.

veen  ·  12 days ago  ·  link  ·    ·  parent  ·  post: What This Seat Bracket Says About the Future of Automotive Manufacturing

I might very well have my terms mixed up - I'm by no means well-versed in this type of engineering. (Yet?) My understanding of parametric is that it is a relational method of design; thing A's size depends on thing B. It can be made much more complex than that, and it can be types of "inputs A should lead to shapes B constrained by strenght model C and solution space D", but fundamentally it's about the relationships between design elements. Could've made that clearer, I suppose, or I'm missing something.

Generative design as I have dabbled with is "given input parameters A that each can vary this amount, and given model B that takes these parameters and generates a design, measure set of result-based KPIs C and iterate over A to optimize C."

kleinbl00  ·  12 days ago  ·  link  ·  

In mechanics, you're talking about parametric design; Solidworks and Fusion will do it all day long. "The distance from A to B is 1/2 the distance from B to C." Move C, A will move. Those interdependencies allow you to change things up pretty easily if you set your model up right.

Generative design in mechanics runs exactly as you describe, but the goals are different:

"Given torsion forces A, B and C, compressive forces D, E and F, shear G H and I and clamping forces J and K, generate a continuous cross-section between pins Alpha-Beta and holes Gamma-Theta that minimizes the weight of the assembly. Presume a Young's Modulus of x, a shear modulus of y, a density of z and a factor of safety of five now GO."

Things had to get computationally intensive to do this. you're goal-seeking through finite element analysis which, back when I had to do it, took ten minutes of server time on a DEC Alpha just to model a bike frame made of constant-diameter tube.

veen  ·  16 days ago  ·  link  ·    ·  parent  ·  post: What This Seat Bracket Says About the Future of Automotive Manufacturing

Generative design rests firmly on parametrically designed products - e.g. if a watch is 42mm, the hour hand length should be less than half that, so

    hourHandLength  = watchDiam * 0.4

The demo example that I've been playing with is that of three towers. Each tower has an x and y coordinate for its centerpoint and a height parameter. The three towers together have a width and length parameter. Based on those simple inputs, you can have the program design three towers, adding for example the constraint that the middle tower must be 20% higher than the other two or that they have to comply to the NYC setback.

The goal in this example is to maximize inner volume while also minimizing the outer surface area. I think the idea was to limit the amount of glass needed on the outside. So a good strategy is to have the towers overlap, but how much and in what formation? Basically, with the variable inputs (x,y,height,...), the parametrically defined shapes and constraints, and the desired outputs to optimize for(MaxVolume, MinSurfaceArea), the software will try many variations on the input variables and score them on the output variables.

Of course, many different goals / KPIs can be considered, as long as they can be calculated based on the resulting design shape. So with this bracket example I am pretty sure that they also ran the computer-generated designs through some strucural simulation models in Fusion to see how strong it would be against certain loads.

I think the interesting thing is that it is a very flexible framework for creating wayyyy more alternative solutions to a problem. It also makes the consequences of your design choices much more visible - you can actually see what it means to choose sustainability over profitability, or volume over surface area, or any set of evaluation KPIs over another set of KPIs. Currently, those design choices and consequences are hidden from view, or neglected, or made on gut feeling / expertise. If I can sit down with relevant stakeholders and say "here's 200 designs that are all up to code but each score differently on these factors you value, now what do you guys really want?", that's gonna be incredibly valuable I think.

kleinbl00  ·  15 days ago  ·  link  ·  

Yeah but the watch hand length isn't parametric. It's preferential. Parametric is "there's this much torque from the main barrel, reduce material until only the necessary structure withstands the force on the jewel holes:

Autodesk's archetypal example - the one that's at all the trade shows - is the motorcycle swing arm:

veen  ·  12 days ago  ·  link  ·  

I might very well have my terms mixed up - I'm by no means well-versed in this type of engineering. (Yet?) My understanding of parametric is that it is a relational method of design; thing A's size depends on thing B. It can be made much more complex than that, and it can be types of "inputs A should lead to shapes B constrained by strenght model C and solution space D", but fundamentally it's about the relationships between design elements. Could've made that clearer, I suppose, or I'm missing something.

Generative design as I have dabbled with is "given input parameters A that each can vary this amount, and given model B that takes these parameters and generates a design, measure set of result-based KPIs C and iterate over A to optimize C."

kleinbl00  ·  12 days ago  ·  link  ·  

In mechanics, you're talking about parametric design; Solidworks and Fusion will do it all day long. "The distance from A to B is 1/2 the distance from B to C." Move C, A will move. Those interdependencies allow you to change things up pretty easily if you set your model up right.

Generative design in mechanics runs exactly as you describe, but the goals are different:

"Given torsion forces A, B and C, compressive forces D, E and F, shear G H and I and clamping forces J and K, generate a continuous cross-section between pins Alpha-Beta and holes Gamma-Theta that minimizes the weight of the assembly. Presume a Young's Modulus of x, a shear modulus of y, a density of z and a factor of safety of five now GO."

Things had to get computationally intensive to do this. you're goal-seeking through finite element analysis which, back when I had to do it, took ten minutes of server time on a DEC Alpha just to model a bike frame made of constant-diameter tube.

bfv  ·  15 days ago  ·  link  ·  

So is generative design using a genetic algorithm to design parts like Rechenberg in the 60s just more convenient to manufacture now, or am I missing something?

kleinbl00  ·  15 days ago  ·  link  ·  

"just" if you're willing to invoke 3D printing in the manufacturing process somewhere, which as listed above, puts you in a pretty spectacularly weird regime at the moment.

If you look at that second swing arm, it only exists if you can cast it off an algorithm. Yeah you could probably painstakingly shape it by hand but you're not going to because you don't have any good way to verify that you put stuff in the right place and the weight savings are probably not entirely worth bothering with (I mean, a tire is going to weigh more than what you're saving there). However, as soon as your construction method becomes "make it thicker in one direction by fractions of a mm" it starts to make sense.

veen  ·  17 days ago  ·  link  ·    ·  parent  ·  post: What This Seat Bracket Says About the Future of Automotive Manufacturing

I've become very interested in generative design lately. We had Autodesk over at our company the other day to talk about it, and after playing with the tools for a bit I've become convinced that it could revolutionize urban developments and urban design. Urban design is about figuring out what to do with a place, designing for stakeholders' desires while taking (policy) limitations into account. That has always sounded like an optimization problem to me, with constraints (policy) and goals to maximize or minimize (in the GM case lightness/cost/etc, in our case project cost/sustainability/...). I found the Alkmaar example here absolutely mesmerizing, and I'm arranging meetings with the company that did the pilot. If it still looks interesting after hearing what happened when the rubber met the road, I'm gonna be the one to helm our expedition into generative design.

kleinbl00  ·  16 days ago  ·  link  ·  

Interesting. In a way, it's very similar to permaculture design principles, which claim heritage from African agriculture. Curious to hear how it works.

It hasn't escaped me that generatively designed watch plates might look pretty damn cool... and are the sort of thing you can have only if you're 3d-printing wax for bronze castings, as opposed to machining the things on a VMC. I suspect you'd still need to do boring on a VMC but you could get some really cool lookin' movements that way. Stay tuned, I guess?

veen  ·  16 days ago  ·  link  ·  

Generative design rests firmly on parametrically designed products - e.g. if a watch is 42mm, the hour hand length should be less than half that, so

    hourHandLength  = watchDiam * 0.4

The demo example that I've been playing with is that of three towers. Each tower has an x and y coordinate for its centerpoint and a height parameter. The three towers together have a width and length parameter. Based on those simple inputs, you can have the program design three towers, adding for example the constraint that the middle tower must be 20% higher than the other two or that they have to comply to the NYC setback.

The goal in this example is to maximize inner volume while also minimizing the outer surface area. I think the idea was to limit the amount of glass needed on the outside. So a good strategy is to have the towers overlap, but how much and in what formation? Basically, with the variable inputs (x,y,height,...), the parametrically defined shapes and constraints, and the desired outputs to optimize for(MaxVolume, MinSurfaceArea), the software will try many variations on the input variables and score them on the output variables.

Of course, many different goals / KPIs can be considered, as long as they can be calculated based on the resulting design shape. So with this bracket example I am pretty sure that they also ran the computer-generated designs through some strucural simulation models in Fusion to see how strong it would be against certain loads.

I think the interesting thing is that it is a very flexible framework for creating wayyyy more alternative solutions to a problem. It also makes the consequences of your design choices much more visible - you can actually see what it means to choose sustainability over profitability, or volume over surface area, or any set of evaluation KPIs over another set of KPIs. Currently, those design choices and consequences are hidden from view, or neglected, or made on gut feeling / expertise. If I can sit down with relevant stakeholders and say "here's 200 designs that are all up to code but each score differently on these factors you value, now what do you guys really want?", that's gonna be incredibly valuable I think.

kleinbl00  ·  15 days ago  ·  link  ·  

Yeah but the watch hand length isn't parametric. It's preferential. Parametric is "there's this much torque from the main barrel, reduce material until only the necessary structure withstands the force on the jewel holes:

Autodesk's archetypal example - the one that's at all the trade shows - is the motorcycle swing arm:

veen  ·  12 days ago  ·  link  ·  

I might very well have my terms mixed up - I'm by no means well-versed in this type of engineering. (Yet?) My understanding of parametric is that it is a relational method of design; thing A's size depends on thing B. It can be made much more complex than that, and it can be types of "inputs A should lead to shapes B constrained by strenght model C and solution space D", but fundamentally it's about the relationships between design elements. Could've made that clearer, I suppose, or I'm missing something.

Generative design as I have dabbled with is "given input parameters A that each can vary this amount, and given model B that takes these parameters and generates a design, measure set of result-based KPIs C and iterate over A to optimize C."

kleinbl00  ·  12 days ago  ·  link  ·  

In mechanics, you're talking about parametric design; Solidworks and Fusion will do it all day long. "The distance from A to B is 1/2 the distance from B to C." Move C, A will move. Those interdependencies allow you to change things up pretty easily if you set your model up right.

Generative design in mechanics runs exactly as you describe, but the goals are different:

"Given torsion forces A, B and C, compressive forces D, E and F, shear G H and I and clamping forces J and K, generate a continuous cross-section between pins Alpha-Beta and holes Gamma-Theta that minimizes the weight of the assembly. Presume a Young's Modulus of x, a shear modulus of y, a density of z and a factor of safety of five now GO."

Things had to get computationally intensive to do this. you're goal-seeking through finite element analysis which, back when I had to do it, took ten minutes of server time on a DEC Alpha just to model a bike frame made of constant-diameter tube.

bfv  ·  15 days ago  ·  link  ·  

So is generative design using a genetic algorithm to design parts like Rechenberg in the 60s just more convenient to manufacture now, or am I missing something?

kleinbl00  ·  15 days ago  ·  link  ·  

"just" if you're willing to invoke 3D printing in the manufacturing process somewhere, which as listed above, puts you in a pretty spectacularly weird regime at the moment.

If you look at that second swing arm, it only exists if you can cast it off an algorithm. Yeah you could probably painstakingly shape it by hand but you're not going to because you don't have any good way to verify that you put stuff in the right place and the weight savings are probably not entirely worth bothering with (I mean, a tire is going to weigh more than what you're saving there). However, as soon as your construction method becomes "make it thicker in one direction by fractions of a mm" it starts to make sense.

veen  ·  17 days ago  ·  link  ·    ·  parent  ·  post: Cal Newport on Why We'll Look Back at Our Smartphones Like Cigarettes

    Is it bad I'm reading this article on gasp a smartphone?

No, see also what I said to binder below. The comparison to cigarettes is not about whether it literally causes cancer, it's about cultural norms and what is looked down upon.

Have you read this article? Sadly, it seems to have gone, but Wayback archive still has it. The gist of it is that games like Farmville were intentionally maliciously designed to fuck with your brain's dopamine response, that their creators are more than aware of it, and that they did it anyway.

That article was written in good ol' 2011, but the practice has metastasized all over our smartphone apps. You can say a lot about books or television, but it does not deliver the completely personalized addiction-inducing behaviour that Instagram does. Instagram, for example, will track your usage patterns to make sure that every time you check the app, you see new likes and interactions. It'll hold back likes and comment notifications just to get you to check the app faster. Ain't no book pulling that shit on me.

veen  ·  17 days ago  ·  link  ·    ·  parent  ·  post: Cal Newport on Why We'll Look Back at Our Smartphones Like Cigarettes

I think I'm coming down to thinking of tech more and more like a utility. I mean - my homescreen is full of apps that give me instant access to a lot of knowledge. That's not a bad thing.

What is bad is that so many people resort to their smartphones at the slightest hint of boredom, that we are now assumed to be always connected and available, and that apps try to mimick social cues to get people to spend more time liking, hearting and swiping. Remember the original iPhone presentation? It was pitched like a phone, an iPod and a web browser. A decade of feature creep later and it can now do pretty much whatever you want. You can be fully connected to the world, to your colleagues, to your friends, at any moment. Which makes them distracting sonsabitches and leads to the aforementioned problems.

I don't believe in abstaining or shunning the tech, because it still has too much utility for that. It kinda feels like throwing your Swiss army knife in the garbage because you keep cutting yourself with one of its blades, even though the other tools are still useful.

The question that is now still out in the open for me is this: what role do I want this tech to have in my life? "Everything"(i.e. unconstrained techno-maximalism) clearly hasn't worked out. "Nothing" is dumb too. So I'm looking for a healthy, productive middle ground.

katakowsj  ·  15 days ago  ·  link  ·  

Agreed. A person can place the blame all they like on the sources of temptation, but however we slice it, the burden always falls on us, the consumer, to manage the use of technology, alcohol, food, or whatever could become our personal time suck.

I expect our Paleolithic ancestors had to deal with the same stuff in a different context.

Maybe it went like this, “Me, Grog have leg injury from last mastodon hunt. Me need rest to recover. Jackass tribe mates telling me of upcoming mastodon hunt. They tell me of glory of rest of tribe if I go on upcoming hunt of my life. They post cave painting of past success and what future mastodon hunt will look like without me.”

FOMO is invented. Grog must decide to stay home responsibly and survive, or go chasing glory.

veen  ·  19 days ago  ·  link  ·    ·  parent  ·  post: Pubski: January 30, 2019

Yeah, we went for a week of skiing it wouldn't be an issue, but it's only 3 days of skiing. Last year was my first skiing holiday, and it was more perilous than I hoped - for some stupid reason a blue ski run can mean anything between 'kids playground' and morituri te salutant almost-black steepness in Austria. (Green doesn't exist.)

It's so bad I have been considering writing code that can pre-analyze and rank the steepness of ski runs based on DEMs for me.

I had ski lessons, but once confidence is lost it's really hard to get it back when you're on a steep as shit slope and dreading your next move. Only on the third day did I figure out which ski runs are good to practice on, and by that time it was already nearly too late. All I want this year is to whizz down a mountain at a slow and comfortable pace without breaking anything.

veen  ·  19 days ago  ·  link  ·    ·  parent  ·  post: Pubski: January 30, 2019

Had one of my first ski lessons this week, on one of those carpet conveyor belt like things. Reminded me of how the best learning devices are as difficult as they are rewarding. If all goes well I will feel confident enough to ski with fun the yearly company ski trip this March. And I wanna surprise the gf with an indoor ski day for Valentines, cause I know she really likes skiing but hasn't done it in ages.

elizabeth  ·  19 days ago  ·  link  ·  

I feel it only takes a couple days of skiiing before you can go down the easy slopes pretty confidently. It's pretty intuitive once you get into it - the trick is just to zigzag to slow down instead of trying to go straight down and brake pizza-style.

Winter sports are the only thing making winter bearable right now, enjoy!

veen  ·  19 days ago  ·  link  ·  

Yeah, we went for a week of skiing it wouldn't be an issue, but it's only 3 days of skiing. Last year was my first skiing holiday, and it was more perilous than I hoped - for some stupid reason a blue ski run can mean anything between 'kids playground' and morituri te salutant almost-black steepness in Austria. (Green doesn't exist.)

It's so bad I have been considering writing code that can pre-analyze and rank the steepness of ski runs based on DEMs for me.

I had ski lessons, but once confidence is lost it's really hard to get it back when you're on a steep as shit slope and dreading your next move. Only on the third day did I figure out which ski runs are good to practice on, and by that time it was already nearly too late. All I want this year is to whizz down a mountain at a slow and comfortable pace without breaking anything.

veen  ·  20 days ago  ·  link  ·    ·  parent  ·  post: Why Are Young People Pretending to Love Work?

    unlimited sick

Pardon my ignorance but is that not the norm? Do people just show up if they're terribly ill but ran out of sick days?

I have to admit I can't complain either. 30 free days excluding federal holidays, free pension, free PT, yearly ski trip, holiday bonus and yearly bonus which adds up to about a month and a half of $$.

Devac  ·  20 days ago  ·  link  ·  

I think it's a shorthand for "unlimited paid sick days".

_refugee_  ·  18 days ago  ·  link  ·  

You're correct.