Hacker News Reader

Sadly the media still likes to portray Houston as a town where everyone wears cowboy hats and that oil comes out of the ground Jed Clampett style.

Literally all of my surrounding neighbors are mechanical engineers for oilfield service companies and the common thing we've talked about recently IS automation. However the reality isn't robots but better engineering/process that lets a part run 50x as long with a greater prediction for need of replacement.

You don't have to keep 100 hands on a drilling site for the peak demand if you can effectively schedule.

> You don't have to keep 100 hands on a drilling site for the peak demand if you can effectively schedule.

Exactly, which is why automation is going to kill so many white-collar jobs. "Effectively schedule" means let a computer do it, not Sally Scheduler who has been doing it in Excel only because they forced her to stop using paper and pen.

This article is basically saying "the planes practically fly themselves", and I'm like "yah but do you still see two pilots in the cockpit?".

There might be 50 extra people on a drill site but there's 5,000 in the back office whose jobs could be replaced with some quick programming and no robots.

Any suggestions on this quick programming? I'm a ChemE turned programmer, it would be a cool opportunity to explore. Feel free to email me, username @ gmail.

"5,000 jobs could be replaced"

"cool opportunity"

what a time to be alive

Why even would someone against that be on HN? It's a place where people interested in startups are. Automating jobs away is always seen as a fantastic opportunity here.

It's a great opportunity for those automating. I'm totally for automation as it creates wealth and I think is good for humanity in the long run, but people suffer from automation (losing jobs, homes, provision for their families) in the short term. People suffering from automation is a major reason why Trump was elected. It's highly insensitive to not be aware of the pain that is caused by automation by technologists like us.

For the most part, automation doesn't create wealth. Sometimes it dramatically improves the functionality of an operation, but mostly it just lets private enterprise replace humans with capital. Thus, it enormously increases the wealth of the owners of that capital, and moderately increases the wealth of the managers of that capital, but overall wealth is not increased, only distributed differently.

Also, to the extent that Trump was elected because of economic factors rather than social factors, a cultural reaction against the rampant libertinism of late-stage liberalism, it was because their jobs were not automated but exported overseas. If their jobs hadn't been exported en masse, they would have been largely protected from automation for the last 30 years and until 10 to 20 years from now when things really get serious.

Back when the labor movements were stronger it could be a good thing for (almost) everyone because in order to introduce new automation, management would have to offer employees something in return (high wages, benefits, shorter working hours, etc).

The end result is that everyone up and down the income distribution benefited from gains in efficiency. And those who were displaced had little problem finding new work because their former coworkers now had more money and/or leisure time to spend on new economic activity.

>wealth is not increased, only distributed differently

When we have the technology to meet the population's needs without its involvement in production, we have something much better than wealth: we have our lives back, our time and energy, our freedom from wage slavery.

We have government for the distribution problem.

The most viscerally disturbing thing I've ever seen a human do, is wish that others be forced to expend useless, unnecessary effort to "earn" what we could just hand out.

I understand the argument about stealing person A's labor output and giving it to person B, but no individual has a legitimate moral claim on the output of a fully automated process.

> The most viscerally disturbing thing I've ever seen a human do, is wish that others be forced to expend useless, unnecessary effort to "earn"

Seriously? That tops eating out of a dumpster (literally, like eating unwrapped, half-eaten discards)? Or defecating in a stairwell?

> what we could just hand out

It didn't work as charity. It isn't working as redistribution through taxation. What's your proposed solution?

>That tops eating out of a dumpster (literally, like eating unwrapped, half-eaten discards)? Or defecating in a stairwell?

Neither of those things are wishing lifetimes of drudgery as a condition of survival onto your fellow human beings.

> It isn't working as redistribution through taxation

Which companies are paying taxes on their automated output comparable to what the payroll would cost?

>>That tops eating out of a dumpster (literally, like eating unwrapped, half-eaten discards)? Or defecating in a stairwell?

> Neither of those things are wishing lifetimes of drudgery as a condition of survival onto your fellow human beings.

You're either not communicating your point effectively, or you don't seem to have any understanding of reality.

Who, exactly, is "wishing lifetimes of drudgery as a condition of survival onto [their] fellow human beings"?

And who, in their right mind, would willingly forego stable access to the necessities of life (food, a home...) because they find such stability intolerable compared to a lifetime of drudgery?

>> It isn't working as redistribution through taxation

> Which companies are paying taxes on their automated output comparable to what the payroll would cost?

We seem to agree that redistribution through taxation categorically does not* solve the problem of providing people with the necessities of life. So what is the point in discussing a specific aspect of taxation?

Instead, I'll ask again - what is your proposed solution, such that people's basic survival needs are accommodated, given that charity doesn't solve the problem either?

* I'm not saying that it can't. I'm saying that it doesn't - primarily because the wealthy have bought 6 decades worth of across the board, increasing tax cuts (among other things) by buying government.

Man is an economic animal. If you want to see what purposeless leisure looks like, look at the underclass: rampant drug problems, animalistically promiscuous women, broken families, bastard, fatherless children, trash culture, and crime.

50 years ago, those people, except for the absolute worst of the worst, were productive, law-abiding citizens with intact families, functional homes, and a place in the world.

When a man is made economically useless, he ceases to be a man. He becomes just another useless mouth to feed, like an animal in a zoo, but unlike the zoo animal in that people do everything in their power to avoid seeing him.

What libertine thing are you referring to? Equal rights for gay people, women making the same wages as men for equivalent work? I'm from a southern state, and I hear a lot of grousing about uppity women and black folks (and people don't use words as nice as that). They want revenge. That's the kind of thing that I see people complaining about liberals. And that kind of behavior is ugly and unneeded. I can think of things that appear as silly to me, like safe spaces, trigger warnings, but I've never encountered such things in real life.

All of them. All social behaviors, sexual and otherwise, which cause and contribute to the breakdown of the family, the community, and the nation.

I don't need to explicate; you know exactly what they are.

P.S. 99% of the guys complaining about liberals are themselves liberals, though they're too dumb to realize it.

We are already there in other sectors. I have worked in Workforce Management and Optimization (WFM/WFO) for the past few years. Scheduling is already common in areas where demand can be easily captured, for example retail banking and contact centers. We use the past data plus other "stuff" to determine the given workload for a position/activity and then schedule employees based on several factors among them are min/max hours, availability, competency, and proficiency. We even have algorithms that determine when you should hire or fire people based on workload.

From my point of view, anything that can measure demand will eventually have automatic scheduling. All those jobs go poof

I think back office is still the current whale in the space. Some of it is harder to determine especially with async work. You are right that thousands of people will continue to lose their jobs here. Back in the day it took my college a long time to process graduation applications. I got a job in that sector that sold software to manage things like students. 6 months of processing buffer down to a button press that takes a few minutes to do everything. So many paper pushes, managers, and other "useless" jobs will go away. Right after that will be those jobs that make sure everything is in order/correct according to business rules and laws. After that will be the huge disruption of lawyers and doctors.

Honestly, I am pessimistic of the future because I see it this way and do not see any political system in my country that is trying to prepare for mass unemployment or people without relevant skills.

Now you have a computer making sure nobody gets full time hours, schedules are different every day and change weekly making it impossible to plan a life around the job. And of course people need more than one job and the other jobs are doing the same thing.

You can see similar patterns with eLearning and performance management technology. The former decreases the cost of turnover for a firm, the latter eliminates the employee's ability to shirk which in turn frees the firm from having to offer Efficiency Wages. The firm is now able to adopt a policy of "burn 'em out and replace 'em" and offer the lowest wages legally permissible. Amazon appears to be at the leading edge of this technology revolution.

>" Amazon appears to be at the leading edge of this technology revolution."

Can you elaborate on this? In which parts of Amazon? Order fulfillment? IT?

>"Honestly, I am pessimistic of the future because I see it this way and do not see any political system in my country that is trying to prepare for mass unemployment or people without relevant skills"

Is that the US? The reasons I ask is because the Trump administration seems to be under the impression that they can "undo" globalization and manufacturing jobs and coal industry jobs are coming back. They seem to be intent on fixing the the former wave of job disruption instead of focusing on the next one. I'm afraid the former ship has sailed and not sure it can be fixed but at least the next one could be prepared for. And there was almost no mention of this on the campaign trail from any candidates.

Interesting Alvin Toffler who died recently and there was a discussion about him on HN talked about many of these things( disruption in post industrial society) that are coming to fruition now in his book Future Shock which was published in 1970.

Yo! I even worked on a software project for automation once upon a time.

Main thing IMHO is that the industry uses a lot of hardware that has payback periods of decades. Nobody's going to spend money on high-end new automated gear unless they have high confidence that there's going to be enough business in the next few years/decades to pay it off. The offshore operators tend to buy in first, both because they have the capital, and because the cost and risk of keeping more people offshore is so much higher. Land operators tend to get this kind of stuff at a trickle.

The headcounts will fall, but not as much or as fast as some would have us think, both for that reason, and because this stuff is hard to get right. The project that I worked on mostly put more people on the rig just to keep all of the stuff running. There were only a handful of rigs in the field with enough automatic hardware to use our system in the first place.

>Drilling rigs don't suck oil out of the ground, they drill a hole...

One of the clever pieces of writing/directing in Deepwater Horizon was at the beginning when the daughter of the character played by Mark Wahlberg did the thing with the coke can to demonstrate how an oil well worked (or didn't).

IMO, it was a brilliant piece of non-exposition exposition, whether or not a lot of the audience really picked up on it. The basic concept was pretty important to the incidents of the movie but to explain it during the normal course of the movie would have been really tough to make seem natural. (I can't come up with a great CS example off the top of my head, but it would be like two programmers making a statement like 'You know, we have to remember that these programs are actually translated into 0s and 1s.')

Yah, just saw the movie last week. Definitely a useful "non-exposition exposition". I thought it was hilarious when everyone became a structural/petroleum engineer overnight when the well head was still spewing weeks after the incident. Everyone had a simple solution. Like when baby Jessica fell down that well in Texas. I'm like "you do realize methane hydrates crystalize at those pressures right?"

This is typical HN clickbait automation porn. I always complain about it and just get censored like this comment will. Some programmers and tech people unfortunately get some kind of sick satisfaction thinking(incorrectly) that blue collar labor will be automated away with physical robots before white collar workers are automated away via software.

I think that's unfair. I think many of us read and submit stories like this because the changes hitting society are fascinating, not because we delight in people losing their livelihood.

For every automation story, there's one on basic income.

"...frankly, how bad these companies are at making software."

Yep. And hardware, and knitting together the boxes, and power, configuration management, and networking and...

If you can't do it with a pipe wrench, why it ain't worth doing :)

The problem with software there isn't the software people. It's the requirements gathering. I don't expect it can actually be resolved. You should see what passes for MRP around Houston.

I believe I first heard about the Iron Roughneck in the '80s, just in idle chatter.

Totally.

Typical company: let's hire a body shop who will fly in a bunch of people who barely know English and pay them crazy bill rates to write reams of completely unreadable "specs"

Programmers cost more than pipe welders but you don't see them getting unintelligable 200 page Word documents instead of certified blueprints from an expert do you?

>LOLOLOL yah, trust me that Fluor, Parsons, Jacobs and all the other oil and gas engineering firms would love this and have been trying to build or integrate things like this for a long time.

This is the same site that, for the most part, thinks Tesla beating out Ford/GM/Toyota/Honda/Nissan/Hyundai/Fiat/Google/Apple is a done deal.

Be careful with your sampling biases. Those of us who are sceptical but don't have a strong opinion are much more likely to say quiet.

> The limit to automation in this industry and others is, quite frankly, how bad these companies are at making software.

Your statement contradicts the rest of your paragraph (that the article is naive and far fetched).

Because if this is really true, then it's only a matter of assembling a good software team, with one or two domain experts (who know the domain and are good s/w engineers), pair it up with the team of engineers and line workers, and a successful automated system can be built, i.e., exactly what the original post is talking about.

Automation as defined by that comment is a really low bar. It probably is possible.

> That means an engineer can design an oil well at his desk. With the press of a button, an automated system would identify the equipment needed from a supplier, create a 3D model and send instructions for building it out into the field, Hashmi says. “That is automation.”

I'd be very surprised if that doesn't already exist: printing a 3D model, submitting an order for the parts, and sending the documentation to the field.

That's assuming the engineer has identified the parts and written the instructions. Automatically doing that might be more of a challenge.

It's not hard to imagine a search engine which matches mechanical descriptions of parts to a proposed design automatically. Just draw roughly what you need, punch in the specs, and let the expert-system provide a least-cost bill-of-materials which comes nearest the ratings + a list of custom parts.

Who is going to do provenance on the parts list? You know that all organizations that buy parts have an entire fleet of people to do this, right? It goes all the way from vendor accounting to physical inventory to finance to shipping to engineering to production to...

OMFG the hubris. I don't know where to start. Oil and Gas exploration and production are a Hard Problem. Visit the Offshore Technology Conference sometime.

> printing a 3D model, submitting an order for the parts, and sending the documentation to the field.

Hahaha. No, let me talk you through it (assuming we are talking new technology, as opposed to a small refinement of existing stuff):

Engineer sits at his desk/wherever and has an idea for some equipment. We're at TRL 0 (following API 17N).

Engineer spends some time demonstrating and documenting how it's likely that this solution will work, possibly involving CFD and/or FEM work. We're at TRL 1.

Engineer gets a simple scale model built, and the scale model is laboratory tested to prove that the concept is likely to work. We're at TRL 2.

Engineer gets a (downscaled and/or simplified) prototype built and tested over a limited range of operating conditions. We're at TRL 3.

Engineer (but really, now we're talking the company) gets a first full scale version built, and this is tested at realistic operating conditions. TRL 4.

The full scale version is integrated with the actual system which it will be running in. TRL 5. This is basically "first deployment", and you've probably spent a few hundred thousand dollars at this point, if not much more.

Your first deployment case has run for an extended period of time under a range of real conditions. TRL 6. The solution is now proven, and can be promoted to other facilities that don't want to be the first adopter.

Your technology has run for a significant amount of time at several installations under a wide range of conditions. The technology is now proven. TRL 7.

According to the article, what I quoted is what Hashmi described when he said “That is automation.”

If that's what he's talking about, I think it's possible if not already in place.

But I agree, replacing all the testing and the skills of an entire company is another matter entirely, and there is certainly a lot more to be done than was described in the comment I was talking about.

It is not a matter of assembling a team. It is a matter of corporate culture.

You can't bolt software on to an organization like that. They simply do not understand, deeply and utterly. It might be willfully. Just identifying defects clearly can be quite a bit beyond some people.

But I heartily recommend that you try it. You will learn a lot about humanity.

This. It's SO HARD to recruit good teams and even when you do projects fail because of the impedance mismatch between software development and bureaucracy.

Bureaucracy has to do Real Work(tm). Gor bless 'em :)

Leadership of tech, especially disruptive tech, is SFAIK mostly half black art and half blackmail.

You do realize that "assembling a good software team, with one or two domain experts ..., pair it up with the team of engineers and line workers)" is like kryptonite for these people right? It's impossible. I've seen them try over and over and fail. The one time I saw a client assemble a great software team with great domain experts we had project managers who wanted to rescue the project from the jaws of success. We fought back, that system went into production and the company laid off scores of high-pay office workers in Houston.

So yes the only thing standing between BACK-OFFICE people and their jobs is how bad these companies are at making software and believe me they are bad. Furthermore, out on the drill/job/construction/plant site robots are not going to be taking anyone's job tomorrow.

> You do realize that "assembling a good software team, with one or two domain experts ..., pair it up with the team of engineers and line workers)" is like kryptonite for these people right?

I don't care if it's a kryptonite for these people. My point is that if that's the problem, then automation is "technically" a solved problem! only that it has "managerial" challenges (and managerial challenges of white-elephant corporations are solved by small startups. So it may be bad news for big corporations, but it's good news for the progress of the rigging industry).

It is theoretically a solved problem, and yes, this is good news for more nimble companies. However the "managerial problem" as you call it is immense. It is almost intractable. Source: They hire me to fix the "managerial problem"

Let me break this down.

You build a company, any company, and it's core business is doing something, anything, OTHER THAN MAKING SOFTWARE. You get good at it, you get big. Internal processes and culture might be totally fucked to the individual observer but as a whole the organism is thriving in its environment. So if the company sucks at making software you can bet (that until this point) being good at software has not been critical for survival in its environment. Hell there are plenty of software companies that are horrible at making software and yet they survive because being good at making software is (for whatever reason) not required for survival in their environment (market).

So you have all these companies that are really good at pouring concrete, or not paying insurance policies, or flying airplanes, drilling for oil. They all suck at software but WHO CARES. They buy technology from companies that make it better than they could and the rest they cobble together SOMEHOW.

Now imagine you've reached peak: materials science, labor cost reduction, managerial science, globalization, etc...

You and your competitors have all outsourced, matrixed, offshored, bought, invested, borrowed, hedged, and most of all deployed the best COTS tech available for your needs.

What do you do now? How do you get an edge over your competitors? What space is left where 10x gains can happen seemingly overnight?

Software.

Companies are slowly arriving at this notion. The problem is they are these very uniquely tailored organisms and everything about their evolution seems almost tailored to suck at software. They throw immense sums of money at software projects. They hire the best firms in the business (they think) and it JUST DOESN'T WORK. Uber and AirBnB quickly gathered more respective capacity than their traditional competitors and it ALL JUST LOOKS SO EASY but these big companies with massive budgets can't make software to save their lives. Even some old software companies can't make software to save their lives.

The problem is systemic:

* Company wide HR practices are toxic to hiring good developers

* Technical ability in leaders isn't respected

* Complexity isn't fundamentally understood

* The first-principles of software are like Klingon

* Bureaucracy has an impedance mismatch with software development

I could go on and on. The only clients I've had that have a chance of overcoming this inertia have leaders starting from th CEO on down drink the Kool-aid and decide that they want to become a tech company. These folks aren't perfect but they begin to reassess everything, compare themselves to companies that are great at making software and start cargo culting everything they can find. That begins to bring about the kind of change that might make them good at software. It's ugly, but its better than the status quo. Still, they are the precious few. The rest hire "agile coaches" like me and waste my time and their money and have a "snowball's chance in hell" of ever getting good at making software.