Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

Goodbye, data science

ryxcommar.com

Post reply on HN
Print view View original

351–360 of 415 posts

Re: Goodbye, data science

#351

Post by baazaa »

People who make important decisions get paid a lot and have a lot of power. Why on earth would they then delegate that decision-making to data-peeps and make themselves entirely redundant? Data scientists are always going to be funnelled into make-work projects just so companies can claim they're doing well on the data-side of things.

The same is true of automation incidentally, there's lots of big companies doing a lot of easily automatable work, but the guy who manages all those people doing easily automatable-work is hardly going to be scrap his own area by calling in some SWEs.

On a completely unrelated note, the author's data eng sounds like nothing I've seen. Hell veto power over code? I'm not even sure all of them can code. They're just glorified sys-admins who now can provision some cloud infra. Somehow data engs are probably even more incompetent on average than data scientists, and the reason you move upstream is because it's an easier job and you don't need to spend your weekends grinding through maths-problems or learning new languages while probably still having higher value-add.

Re: Goodbye, data science

#352

Post by ZephyrBlu »

PragmaticPulp wrote: post #196
scottLobster wrote: post #156

Earlier quoted context omitted.

Not to get too off topic, but as a 35 year old engineer it seems the world in general has far fewer consequences than I was raised to expect. Everything from businesses with bullshit ideas flourishing at a loss, to January 6 even being possible (politics aside I expected the Capitol Police to crack a lot more skulls than they did once people started smashing windows), to the whole FTX situation and the tepid response…

Consequences often catch up slowly. It took years for Elizabeth Holmes to be sentenced because it takes time to collect evidence, build an airtight case, and give people their due process. As I get older, I'm actually noticing more and more consequences catching up with people, albeit slowly. The people I knew who drank heavily through their 20s and 30s are in much worse shape than basically anyone who made an effort…

The problem is that time value is extremely relevant. If it takes 10-20 years for consequences to catch up, the person is likely to have already built up an unassailable lead that the consequence barely dents.

> He literally had to leave the state and go somewhere new to escape his past network and get new jobs after 10 years of this.

That's not even that bad of a consequence. It sounds like his strategy was worth it tbh.

Personally I hate this kind of behaviour, but from a maximization POV (Especially in regards to career) it seems like the best move. There is likely some risk of ruin, but the upside appears to be much greater.

Re: Goodbye, data science

#353

Post by ZephyrBlu »

alexpetralia wrote: post #205
chasely wrote: post #42

Earlier quoted context omitted.

I've been pitched by many "data-driven" vendors offering predictions. They often have very impressive accuracy metrics (RMSE, R2, etc). When I dive into the details these metrics are often reported using in-sample predictions. I see this pointing to any of the following: a) DS teams overpromising the accuracy of their approaches b) marketing driving the narrative and DS getting pulled along c) incompetence from the D…

Are these inferential statistics not designed to be in-sample? I would imagine predictive statistics use more out-of-sample metrics like precision and recall.

I feel like you might be conflating a couple of things, though I'm not a DS so could be off base here.

My reading of the OP's description is that the vendors were offering interpolative predictions, but did not use a test/train split of data. This is in contrast to extrapolative predictions which I would call out-of-sample.

Thus due to not using a test/train split, they achieved extremely good accuracy because they were testing on the same data they trained on. Even though this is "in-sample", you can't use the same data for testing and training.

Re: Goodbye, data science

#354

Post by caddemon »

Jensson wrote: post #334
cjglo wrote: post #104

Earlier quoted context omitted.

Could be wrong here, but in physics and most natural sciences, you don’t throw away your model if you have one experiment against it. Usually isn’t it looking for an experiment that proves it and is repeatable? If I discover a new element in one experiment, the results are published. After publication, many labs will try to repeat and its not taken away if one can’t do it. Only if all can’t and it casts doubt on whet…

Scientific method? Models are disproved, not proved. Do data scientists not know about science? People usually understand how science works here on HN, but not in this thread. Example of a test that invalidated our old theory of gravity and validated Einsteins claims: https://en.wikipedia.org/wiki/Eddington_experiment This is how science is done. But apparently not data science.

Models can also be knowingly/intentionally incomplete, which necessarily introduces noise that you are not controlling for. Meaning you have to use statistics, and there isn't really a concept of prove or disprove in the true sense of those words.

Maybe very far in the future there will be models of human biology that are as robust as classical physics, but right now there is such a large amount that is not understood, it's simply not feasible. A drug could work for one person and not another for reasons beyond the realistic scope of the original development hypothesis. It requires a probabilistic view to make any sort of statement about the efficacy then.

I suppose you could argue these models are just wrong and thus trivially disproven, but I don't think that's a productive framing. I doubt any biologist or doctor would claim they have anywhere near a complete model of how their specialty works. That doesn't mean a particular model isn't useful or isn't the best we currently have to work with.

Plus maybe the third best model will actually turn out to explain a separate puzzle piece in an eventual better model. Mechanistic models in biology aren't always well done in practice, but it's certainly not binary either.

Re: Goodbye, data science

#355

Post by ZephyrBlu »

wheels wrote: post #342
fnands wrote: post #135

Earlier quoted context omitted.

Honestly, I can't tell you how many jobs ads I saw where I was wondering: "What would they expect me to bring to the table here?" Some companies just don't have the data, or heck even the need, for data scientist yet try and hire them anyway. Give smart people a fundamentally ill-posed problem and they won't get anywhere anyway.

My ex worked at a startup where she was hired as a the second or third data scientist. Their entire Posgress database dump was 20 MB. And they had three people working full time on analyzing ... that 20 MB.

I call it artisanal data science *chefs kiss*.

Re: Goodbye, data science

#356

Post by xiaodai »

> Rather the main bottlenecks I’ve faced were always crappy infrastructure and lacking (quality) data, so it has always felt natural to focus my efforts toward learning that stuff to unblock myself.

Right. What can you learn to over come crappy infrastructure?

Re: Goodbye, data science

#357

Post by ZephyrBlu »

euix wrote: post #189

I have to agree with a lot of this - I started my career as a data scientist right out of a STEM PhD back when the term just started coming into existence. At the time, anyone who wanted to get hired as a Data Scientist needed to be trained as a professional scientist, i.e. have a PhD - at first my expectation that the purpose of my job was to apply the scientific method to solve business problems by leveraging the c…

Props for saying this, but you didn't realize the "marginal business impact" bit until you were a Director..? Seems awfully convenient.

Re: Goodbye, data science

#358

Post by caddemon »

clircle wrote: post #260
r-zip wrote: post #235

Earlier quoted context omitted.

Can you recommend a few such places? :)

Look up the technical presenters at your favorite math or stats conference and look at the orgs they work for. That's how you can tell which orgs are serious about solving research problems and spreading solutions.

You could also look into the successful quant trading shops, if you're alright with not sharing results outside the firm. The nice thing about them is success or failure of your work is apparent much more quickly than the timescale of science, and it's harder to spin than other businesses. So the data science that gets done is generally pretty technically legit.

Re: Goodbye, data science

#359

Post by itake »

wheels wrote: post #342
fnands wrote: post #135

Earlier quoted context omitted.

Honestly, I can't tell you how many jobs ads I saw where I was wondering: "What would they expect me to bring to the table here?" Some companies just don't have the data, or heck even the need, for data scientist yet try and hire them anyway. Give smart people a fundamentally ill-posed problem and they won't get anywhere anyway.

My ex worked at a startup where she was hired as a the second or third data scientist. Their entire Posgress database dump was 20 MB. And they had three people working full time on analyzing ... that 20 MB.

this was me in 2017 lol

Re: Goodbye, data science

#360

Post by layman51 »

scottLobster wrote: post #156
_jx7j wrote: post #95

Earlier quoted context omitted.

>Meaning you could absolutely suck at your job or be incredible at it and you’d get nearly the same regards in either case. One of the things I don't like about statements like this said in a Data Science context, is that they are true outside of Data Science as well. Executives make big decisions, managers make smaller decisions, nobody can evaluate how good/bad they really were for months or years. Engineers build…

Not to get too off topic, but as a 35 year old engineer it seems the world in general has far fewer consequences than I was raised to expect. Everything from businesses with bullshit ideas flourishing at a loss, to January 6 even being possible (politics aside I expected the Capitol Police to crack a lot more skulls than they did once people started smashing windows), to the whole FTX situation and the tepid response…

After reading your comment, I think you have captured some of my own thoughts about consequences and deserts (i.e., worthiness or entitlement to reward or punishment). I agree with the other comment that replied to you that says that thinking like this is a product of being raised in a bubble.

I am not sure if apathy/nihilism is growing in the larger society. I think that things have always been like this because people have always struggled to find meaning in life. After taking an intro psychology class, I was exposed to the idea that society wants an individual to police him/herself. The "super-ego" that makes one feel guilty for breaking rules and want to aim for perfection.

Post reply on HN