HNHacker News
TopNewBestAskShowJobs

dkarapetyan

1 karma · joined December 4, 2019

submissionscomments
dkarapetyan··on Cancer Isn’t a Logic Problem
If you look at how cells work on the micro-scale it looks a lot like cancer. They constantly divide and try to proliferate by allocating and using resources. The only thing keeping them in check is basically the "cellular honor code". You can certainly try to make it less likely any given set of cells breaks the honor code but you can't "cure" cancer. It's an inherent property of multi-cellular life. Not unlike bee and ant colonies. There is always one queen (or several) but sometimes there will be a rogue queen born that bifurcates the colony. Bees and ants go to great lengths to avoid these situations but the way colonies are set up and structured rogue queens are an unavoidable part of the colony given a long enough timeline.

I characterize scurvy as a sub-optimal stable operating state where optimality can be restored. Cancer does not have similar characterization. It's more like an unstable runaway loop.

But I'm neither a biologist nor a doctor. My understanding is basically a bunch of analogies between game theory and general systems in the context of biology and maybe that's not the proper set of analogies for the problem at hand. In that context cancer is the breakdown of feedback loops that keep the system stable.

dkarapetyan··on Cancer Isn’t a Logic Problem
> Cancer relates to our fundamental constitution as multicellular organisms, our limited lifespan, the epidemiology of the aging population, socioeconomics and the future of society.

This is the most important part. Life is cancerous. When people claim to cure cancer it's sort of an oxymoron. There are indeed no such things as cancer genes but more like a breakdown in the cooperative biochemical networks that make multi-cellular life possible. It's actually a great wonder how all those cells in your body manage to cooperate at all without stepping on each others' toes all the time.

In a way the delicate balance is a constant fight against entropy and entropy always wins. Enough breakdowns and misunderstandings in the biochemical networks and the multi-cellular ecology and you have cancer.

dkarapetyan··on Fuchsia: a new operating system
Oh they're using Dart. That's really cool. Dart is surprisingly pleasant to work with.
dkarapetyan··on Ask HN: What do you want to see in Ubuntu 17.10?
FLAVOR: Server, Core HEADLINE: Snaps are too hard to make ROLE: DevOps, release, infrastructure engineer DESCRIPTION: In fact they are so hard and convoluted to make that I always fall back on just comiling and packaging everything inside a Vagrant VM or a Docker container and then just generating a tar or deb with FPM. You guys really need to simplify the process if you want software to be delivered through snaps.
dkarapetyan··on Ask HN: Google Cloud Certification vs. AWS Certification
Pretty sure certification is a racket. Why would you play into it?
dkarapetyan··on AI vs. MD: what happens when the diagnosis is automated
Doctors are terrible. The sooner they're out of the diagnostic loop the better. Every time I go to see one I get a different diagnosis or something generic. It's basically a crapshoot based on the experiences of the doctor.

I'd much rather have a machine do all the pattern matching without any human bias.

dkarapetyan··on Employee burnout is becoming a huge problem in the American workforce
I don't think it is a matter of getting away. I can not be productive (truly productive) for more than 4 hours a day. The rest is just overhead my employer pays to keep me on hand.

I guess in theory they could just pay me for those 4 hours as much as they'd pay me for the 8 hours but that's way too much sense for most folks.

dkarapetyan··on You could have invented Parser Combinators
Yup, parser combinators are a lot like regexes and if you can write a basic regex engine then with a few tweaks you get parser combinators.
dkarapetyan··on Ask HN: What technical skill should I learn to prepare for the next 10 Years?
Study the fundamentals. Learn some timeless science like physics or math.
dkarapetyan··on Uber is facing the biggest crisis in its history
Yes. I'm assuming the ridesharing component is indeed not profitable.
dkarapetyan··on Uber is facing the biggest crisis in its history
But they actually can't survive. You're assuming the current demand can offset all the money they're burning.
dkarapetyan··on How to Be Someone People Love to Talk To (2015)
Are you for real? Develop substance my friend. The rest is all cosmetic and won't take you far.
dkarapetyan··on How to Be Someone People Love to Talk To (2015)
There was that one time I observed a peculiar quality about a certain CEO. No matter what he was talking about it somehow would always circle back to talking about whatever company he was currently at and the conversation would always end with a joke and hearty laugh for all involved. This happened consistently enough that I thought it was a pre-determined act on his part.

Once I realized he was always practicing I kinda stopped talking to the guy because there was never any genuine interaction. He was always on the job and he was always practicing selling. Every conversation was just another opportunity for him to practice his messaging. I dubbed this mode of interacting and talking ceoesque.

dkarapetyan··on Leverage Points: Places to Intervene in a System
I have that book. It is were I learned the term "socio-technical system". It is a great book and I recommend it to people whenever I get the chance. It is also much more than just safety engineering and is one of the more clearer expositions of cybernetic systems that I've read.
dkarapetyan··on Photos Reveal More Than 200 Arctic Lakes Have Started Bubbling with Methane Gas
I don't think water levels are the issue but runaway temperature changes. The thing with complex dynamical systems is that they have phase changes. Most models seem to predict phase changes right around a few degrees of change.
dkarapetyan··on Ask HN: Need advice for a senior lead
Treat the junior devs as peers and you'll be fine.
dkarapetyan··on Amazon's Tepid Response to Counterfeiters Frustrates Sellers
Isn't the obvious solution to have some kind of registration process? Or am I missing something?

If I'm publishing a book then anyone else that wants to sell that book in large quantities must register with me and get my sign-off. Large quantities being somewhat vaguely defined.

dkarapetyan··on Photos Reveal More Than 200 Arctic Lakes Have Started Bubbling with Methane Gas
How do you supposed you are going to survive without all the affordances of some kind of civilization. Do you imagine some kind of postman or waterworld type of thing?

I'm pretty sure either we all make it or none of us do. There is no escaping the inevitability of mammalian biology. You need to keep your body at a certain temperature and it's looking more and more like that is going to be impossible in the new climate regime.

dkarapetyan··on Photos Reveal More Than 200 Arctic Lakes Have Started Bubbling with Methane Gas
I think it's going to be a much slower burn this time around. More like boiling a crab very, very slowly. We had a good run but in the end the dinosaurs had the last laugh.
dkarapetyan··on Thanks for Submitting Your Résumé to This Black Hole
Sure but what do you trust more, a properly designed control system with various safeguards to allow for the right amount of river pollution, or politicians arguing about which of their buddies should get the rights to polluting that river the most?

In the future I imagine your question about the river pollution is almost nonsensical. That is if we manage to actually get there. Going by current news about the future of earth's ecology we might not actually make it much further. Gotta admit though heroic human optimizations seemed like they got us pretty far but the lag time in reacting to things is a bit too long when humans are in the loop. So the sooner they are out of that critical loop the better.

Speaking of which: http://www.sciencealert.com/photos-reveal-more-than-200-brig...

dkarapetyan··on Thanks for Submitting Your Résumé to This Black Hole
I'm more and more convinced that being a cog in some corporate machine is coming to an end. We've collectively refined corporate processes to the point that it no longer requires human ingenuity or creativity. This has all happened under the guise of making things more efficient so that a stock ticker will be less volatile and will consistently move up and to the right. The candidate tracking software that this article parodies is one example of such an automated and "efficient" system. One gets the sense that the recruiters are almost secondary. Soon a "dog" will really be able to screen candidates for a job.

I don't think I'm saying anything new here. The mechanization of such work has been happening for a while now and the smart move is to start planning for that inevitability. Anything that requires basic pattern matching and procedures is pretty much gone. My retirement account I think is currently managed by a "robo advisor". Hiring humans to do such mechanical tasks will start getting more and more expensive relative to tweaking some parameters in some neural network coupled maybe with some domain specific policy/optimization framework.

Personally I don't think this is such a bad state of affairs. Why should societal optimization tasks be handled with heroic human effort when we can just do it with math?

dkarapetyan··on Opening a new chapter of my work in AI
Was about to say the same thing. The incumbents are probably even at a disadvantage because as a young upstart you can always get in on the action by building a more efficient and simpler model and then querying the incumbents to bootstrap yourself.
dkarapetyan··on Ask HN: I got duped into working on legacy code, should I leave?
Really? Which agency is that?
dkarapetyan··on Ask HN: I got duped into working on legacy code, should I leave?
Always several sides to a story and during interviews everyone is on their best behavior. I have yet to be in an interview where I get the full and complete picture without any embellishments. It's only after you have been on the ground floor for a few months that you see things for what they are.

Also, OP's story is currently colored by how upset he is so I'm sure the interviewers made the job sound more appealing than necessary but I'm sure there is also some revisionist history happening.

dkarapetyan··on Ask HN: I got duped into working on legacy code, should I leave?
Was with you until you got a little too aggressive. There is a better way to say what you're saying.
dkarapetyan··on Ask HN: I got duped into working on legacy code, should I leave?
No, no what? Do you have counterexamples where established businesses don't have to deal or interface with legacy code? If so I want to work at this magical enterprise.

According to OP this company has been around for 10+ years. It'd be a miracle if they didn't have legacy code.

dkarapetyan··on Ask HN: I got duped into working on legacy code, should I leave?
All code is legacy code. Use it as a proving ground for learning how to deal with legacy code. Otherwise you will always be bitter and upset no matter where you go. The only way to work on greenfield projects is to start something from scratch on your own.
dkarapetyan··on Uber president Jeff Jones is quitting
Hypergrowth is a lot like cancer. Either they take swift action to excise all the toxic elements or it is going to be a slow and steady decline.
dkarapetyan··on Terraform 0.9
This isn't much of an argument. That's like saying all programming languages suck and it's a crapshoot. It's not. There are design principles that work and scale to large scale systems and then there are things that don't.

As far as I'm concerned Chef is the right way to do configuration management. Everyone else with their YAML DSLs is doing it wrong. Expressing dependencies between resources does not require specialized external DSLs and sometimes you need to do imperative things to get to the end which is either impossible or just horribly convoluted with any tool that tries to hide complexity behind a markup/serialization language.

dkarapetyan··on JavaScript once had a JSX-like syntax called E4X
Anyone remember Yahoo Pipes? I wrote some flows at some point and used E4X. It was fun.
← PreviousPage 7 of 34Next →