G.E., the 124-Year-Old Software Start-Up
nytimes.com
nytimes.com
GE outsourced the majority of its technical jobs to offshore firms. I remember hearing a statistic that for every 1 GE employee in the technology department there were 4 offshore consultants. The role of GE employees in the technology department shifted from a technical skill-set to a managerial skill-set. Many of my co-workers had degrees in finance and marketing, and had never written a line of code in their life. The lack of technical expertise within the organization created what I would call a culture of fear, where people feared the software they managed because they didn't understand it.
The official term for offshore software engineers was GDC. I have no clue what the acronym stood for, but over time it certainly become a derogatory term. I remember being mocked by my co-workers for writing code and being nicknamed a GDC (as in, why would you do the work we pay people pennies to do). It was almost like a caste system within the company, where these GDC employees were low paid and looked down upon by many.
The offshore model certainly was ineffective and the result of employee discontent, however, the practice was continued and mandated at the executive level. Then the economy crashed. GE decided it was finally time to bring jobs back to the US and hire software engineers. Why? Because they were able to take hundreds of millions in federal tax subsidies from US tax payers. (As an aside, someone mentioned GE moving its HQ to Boston to be closer to engineering talent. Taxes are still the main driver here)
GE soon began hiring US software engineers at a number of designated centers where they were segregated from the rest of the company. The interesting thing is that culturally these employees were largely treated like their GDC predecessors. The perception of software engineers being low-wage assembly line employees was, in my opinion, just too engrained into the company culture.
So when GE describes this transformation to a software company, I believe the most important aspect of this transformation will be changing the company culture and employee attitudes toward software engineers. They will need to undo a decade long effort to turn engineering into an assembly line.
I know lots of examples from the automotive industry that sound exactly like that, and I'm sure it's not that different in other industries.
However most of these companies are slowly getting to the point where they understand that software is crucial for the future and that it's a major (and not a very minor) component. But it might still take some time until they can offer a good working atmosphere for software engineers.
That still seems to be the plan: "The San Ramon complex, home to GE Digital, now employs 1,400 people. The buildings are designed to suit the free-range working ways of software developers: open-plan floors, bench seating, whiteboards, couches for impromptu meetings, balconies overlooking the grounds and kitchen areas with snacks." There are pictures on the GE Digital site. What you don't see are workbenches with electrical products hooked up to computers. They're not doing embedded software.
This has an impact on product design. The GE Digital site says: "One million terabytes per day … that’s how much data Predix will process by 2020." They're thinking "big data", where all this stuff is shipped to the "cloud" for later analysis. The alternative would be digesting it near the source and only sending in the interesting stuff. They're going to collect gigabytes of "Bearing 22 temp 18 C", which is only important if the value is out of normal range or shows a trend.
The alternative would be some smartphone-sized device which listens to the data from your turbine or jet engine and finds the good stuff. But that wouldn't justify expensive cloud services, fees, meetings, etc. Even if it did exactly the same thing, GE couldn't collect huge fees for a smartphone-sized analysis device.
I could see something like this powering Bret Victor's "seeing space" idea, which could be quite cool. (http://worrydream.com/SeeingSpaces/)
Should always keep that in mind when formulating tax policy.
This is GE's business model. When they want to enter a space, they throw money ($B) at it until they have a business. They tend to prefer organic builout, but they will buy companies or partner if they think they can't win organically. If you're willing to do that, you almost always get some traction, just because you have unimaginable resources and are willing to adjust your tactics if the initial gambit fails. They tend to do a lot of advertising for key businesses - this is why you see GE Aviation billboards in airports, and GE Oil & Gas ads in Houston.
The model has worked pretty well in the past - a good example is GE's entry into commercial jet engines in the early 70s. A good example where it failed was GE's computer business in the 60s.
I can't prove it (just a feeling), but I think Predix doesn't really have what it takes yet. We've spent a lot of time studying it, and it doesn't seem to offer real value to customers yet. That doesn't mean it won't get there eventually.
However, if Predix doesn't do well, we expect them to go on an acquisition spree to buy a few of the winners. Bently Nevada is a good example of that play - GE controls something like 70% of the machinery protection market. They didn't build it, they just went and bought Bently.
GE is clearly signaling that they badly want to be in this market, and I think it's pretty likely they'll eventually get there.
I think one month we had a 90% uptime. Our services crash all the time.
Our new boss has said he has around a 500-600 million budget for buying interesting companies.
BUT it might have been how our app was coded we weren't taking advantage of some sort of built in failure hooks or something.
CF is a pile of shit anyway. Its overly bloated and its hard to support and deal with
We track it on PWS, meaning we're upgrading the whole platform live every week or two. So I'm surprised it's unstable for you.
But, correct me if I'm wrong,...isn't predix and cloudfoundry all hosted on top of AWS instances at the moment?
All of our URLs were AWS urls. It was funny.
We were hosting on top of our competition.
PWS is on AWS. But Cloud Foundry itself can be mounted on AWS, OpenStack, Azure, GPC is coming and others I've forgotten.
I'd be surprised if anyone outside Silicon Valley in GE's LOBs wanted to consider AWS for production considering how many have likely invested greatly into their own datacenters as well as the general trends of non-technology companies to fear an external cloud first instinctively regardless of how poorly managed their internal IT is. This puts GE in a poor position potentially in that even if they create a great platform that their customers that are perhaps even more technology phobic will need to invest money as well into areas they historically have failed to generate much ROI from. But perhaps GE's dominance / position can help them here and these concerns are long addressed.
The problem I see with an acquisition growth approach IMO is that I would only expect GE to retain maybe 20% of the engineers after 3 years due to the lack of adequate supporting infrastructure and cultural foundation to help long term engagement. None of the talks I watched from management at GE Digital sounded focused at all either - the vision is all over the place and sounded like children wishing for ponies and candy without explaining how to get there and what to prioritize. That lack of management focus is what start-up folks recognize as a recipe for failure regardless of funding level.
1) Sales/Execs push product. Predix is so beyond fucked its not funny. I don't think anyone can come in and fix all the tech debt. Sales will push new customers and force engineers to meet their deadlines. Execs won't allow engineers time to fix anything. Its just keep pushing out awful hacks and 70% working products.
2) Anything you say your manager might listen but it nothing will happen. In order to get anything done you have 10 meetings about it during the week.
The San Ramon office is like a h1b1 warehouse. I think they know they have these people almost locked in and take advantage of it. I've been here over a year and have yet to see a single person fired.
You could honestly fire 40% of the people in GE Digital and it'll function the same if not better then it did before
So the key thing is: from where I'm sitting, GE are actually serious about this. Really serious. They are not alone.
There's this cliché that big companies are all sitting around, waiting to be disrupted, blissfully unaware that someone in the Valley is going to kill them.
It's getting to be a less and less plausible strawman. Those business leaders are reading the same books, hearing the same ideas and -- crucially -- many of them are consciously seeking out people who can teach them (including, I guess this is a disclosure, Pivotal Labs, our consulting/coaching wing).
Enterprises now want to disrupt themselves, because it 1) hurts their existing competitors and 2) creates moats against startups.
The strawman that giant enterprises are ignorant and ripe for disruption because of their arrogance -- the classical model of hubris -- isn't true.
Hubris cuts both ways. It's the Valley that's becoming lazy and complacent, assuming that they have a monopoly on interesting problems, on big revenues, on agile and lean development.
The barbarians used to be at the gates of Rome. Now the Romans are at the gates of the barbarians, and they're much more heavily armed.
I'll buy that.
I still think they are ripe for disruption because they are giant enterprises.
I can't think of any examples of big bureaucratic organizations with a settled culture transforming into something different and better. Normally they get replaced.
We're engaging with more enterprises at Pivotal than we used to.
They're all different.
Some "get it" really quickly and pow, off they go. Most of our repeated Labs engagements with such customers are because they're trying to transform as quickly as possible by seeding teams all across the org.
Others enterprises need, uh, repeated examples.
Some will presumably never get it. And yeah, they'll find one day that the party is over.
But, to repeat, they're all different. Applying a single rule (enterprise slow, startup fast) to heterogenous situations is a going to pay rich returns ... but not to the venture funds.
People in the middle, whose reckoning includes career prospects and health insurance and worrying about mortgages and where their kids will go to school, tend to lean towards the "never take a risk ever" school of thought.
And a lot of corporate cultures are built on punishment and reward, distributed veto powers and a fondness for finding a neck whenever something goes wrong.
But, again, they are not all the same.
From what reads like a bit of a puff piece about Jack Welch [1], who became Chairman and CEO of GE in 1981:
"Under Welch, GE exited many of the traditional markets it had competed in for years, like consumer appliances and air conditioning, and entered completely new areas like medical technology, finance, television and services."
During the restructurings, Welch laid off 100,000 or so people (depending on how you count), earning him the nickname "Neutron Jack". He also instituted the policy of firing the bottom-performing 10% of managers each year [2].
[1] http://www.investopedia.com/ask/answers/09/neutron-jack-welc... [2] https://en.wikipedia.org/wiki/Jack_Welch
Before that I was working in early-to-middle-stage startups ( but not startups in California ).
What I've seen is that you can actually bring something to "market", and the rest of the company will do everything it can to scuttle it. "The rest of the company" is busy keeping the old fires burning. So you don't get buy-in.
I've seen people leave relatively no-risk millions on the table to scuttle these things. But the main body of the organization is able to overemphasize risk in a way that leads to scuttling.
You'd have to have somebody fighting for the thing full time. Even then, it's not enough. I also won't say it's not that the others simply do not understand what was done. and to be fair, the engineering staff weren't really up to it anyway. When you've adapted to just duct tape fixes, it's hard to embrace a real fix. This, and the people at the coal face of these "products" proved less than capable of deployment.
Bluntly, people like having a buggy product or process to charge hours to. it's less work to jigger or abandon any measurement of defect rates and appeal to nostalgia. Management likes having "engineers" actually just doing clerical work because there's then no deployment cost. And frankly, we probably got it all done too cheaply.
I'm in the process of abandoning that model and going to work directly for the customer ( not a customer of any of the firms I worked for, of course ). This is a "firm" that's going from having contractors do maintenance to taking it "organic".
To quote Walt Kelly - "We has net the enemy, and it is us."
For example: 1. Jet Engines and electric turbines 2. Railroad locomotives 3. Appliances 4. Medical devices and imaging and all other things medical 5. Lighting.
The market values to shareholders would be higher.
The firms could focus on their one product line and would have to work harder against competition since there will not be cross subsidizes.
The smaller firms could also be more responsive to markets and respond more quickly because there would be less bureaucratic inertia.
Was this the main reason for disruption ? as i understand it, most often it came from knowing that disruption most likely will greatly reduce revenues/profits and hurt the stock , at least at the short term, coupled with a short term focus that's just embedded in our economy.
The hubris was mostly just talk.
Also a question since you come from the field: how does predix compare with the thingworx OS(which seem to be extremely capable, but expensive) ?
I work on Cloud Foundry as an opensource contributor for Pivotal, not directly on Predix for GE (or BlueMix for IBM, or Helion for HP etc).
Do they really want that? Or just say that want it? As comparison, everybody says that they want to get in better shape. But how many put in the work?
I think there's an enormous amount of money to be made (and probably, being made right now) in going to existing large companies and selling them "innovation" consulting that gives them the feeling of change without doing anything actually hard. It's the equivalent of buying low-fat cookies and calling it a day.
In practice I expect very, very few of these companies to actually do anything meaningfully different. As an example of that, look at how "agile" has changed over the years. 90% of shops I talk to where people are "doing Agile" are actually doing mini-waterfall. It's the same old bullshit with new labels and a slightly faster release cadence.
That's not to say that startups will do any better on average. So many people chased unicorn status without any thought of building actual businesses. But I still expect the startups with a clue to do a fine job going after large companies in the low-fat cookies camp.
Some. Some not.
My argument was against the stereotype of complacency, which itself inspires a different kind of complacency.
I've got skin in this game, having been one of the handsomely-billed consultants (for Pivotal Labs) until transferring into Cloud R&D.
So I'll tell this joke before someone else tells it. It's adapted, of course, from a joke about therapists.
Q: How many agile consultants does it to change a lightbulb?
A: One, but the lightbulb has to want to change.
That makes total sense to me. The larger a top-down organization is, the more the people on top exist in a reality that is socially mediated. That's especially true given that the theory of the MBA is that it's a universal management degree, meaning that knowledge of the domain is at best secondary. In a social context, confidence is very appealing. Strong belief becomes executives' reality.
They did not have a software culture at the time, everything was outsourced. Project managers did not have any clue of how to manage a software project professionally, they struck me as more financial / admin types. India and China were like wolves and GE threw projects to them expecting the lowest cost to emerge as the winner.
Office was always dead on Friday, there was a culture to often work from home but not actually work.
From my understanding at the time they used to cut 10% of the worst performing staff every year, I believe they were one of the worlds largest employers at the time. My biggest peeve was their culture of having meetings just to have meetings....I think this issue can be found in most large enterprises.
My cousin now works for GE in Australia and says 'they are trying to become a startup'. I'm not convinced, it's pretty cool to be called a startup but true startups will only thrive in GE if they are left alone by management and allowed to utilise large resources. It can work, small teams can make great inroads for these types of organisations.
This explains a lot of recent interest in IoT. Big companies like GE saw what happened to companies like Kodak and want to defend their profits. They could be entirely wrong, computing may not turn out add much value in these industries. But GE probably don't want to leave that to chance.
It's sad and hard to do, but I've told a colleague who's trying to go into this market ( we were both laid off this year ) "no", because the overall culture isn't ready for it.
I don't think good pay and good place to work are synonymous in any way. They happen to be correlated because both are methods to attract talent, but one does not cause the other.
PLC inventories are measured in decades, not years.
It's honorable work, but it's not to be confused with software development.
They were saying, yes "apps" for your phone are topical. The point of their commercial was "we work on big crazy complex shit and you can be part of solving a huge challenge".
The photos in this article of their manufacture engineering plants amaze me. Now couple those projects with sensors that are starting to giving off TBs of information in an hour and you can start to realize the scale of problems you could be working on at GE.
In another 5-10 years, many of the folks grinding it out in silicon valley, or other startupy things will realize they have a desire to have a family and will want to apply their talents towards more meaningful projects than "apps" that do not require being part of silicon valley.
Their commercials and this article painted the perfect picture of what's possible at GE and outside of silicon valley in my mind.
If the ads do attract some developers, that's great too, but that is not enough return to justify the (huge, enormous) cost.
These ads seek to do for GE what the "Think Different" ads did for Apple in the 90s.
Does anyone know of such a list that's a genuine reflection of reality?
- Pay / comp - Perks - Perceived difficult technical problems
They may win on number employeed (they just announced a big software development office in India) or perhaps when comparing these figures to competitors like Honeywell, Caterpillar, Phillips rather than the big, high value brand software companies.
This isn't to say that this means failure. Perhaps they can win on cultural perception (say, work-life balance) or career growth and capitalize upon a backlash against Silicon Valley engineer lifestyles. But everything I've seen about Predix from a customer and engineer perspective, which is what they're betting an awful lot on, is hardly compelling to me.
They just know that "Digital" and "Big Data" are big now and they have to get into it. It reminds me of the internet bubble times when lots of companies did "internet" without knowing what it really means.
You had to be a "C-player" for something like three consecutive review periods before you were let go.
(source: my father was a manager at GE for decades)
I hate to be snarky, but I suspect this is not how GE Digital actually came about.
By my (suspect) arithmetic, that is about 200 cubic metres of Argon at STP in one day. Do they have the stuff on mains around the factory?
Considering how large GE is, I wouldn't be surprised if they had it 'on tap' in a PLT automation network and they take bulk delivery from a supplier every month or two. and some bad process engineer forgot to add a sensor check in loop (or his manager just told him to "screw it, argons cheap, no need to add a flow meter to this process" as a cost cutting factor).
Some things I observed:
#1 I watched a top software development company that was doing a great job get fired because they told GE that a desired deadline was impossible. One of the reasons the non-technical manager who fired them gave for the termination was that he didn't feel like they wrote enough lines of code.
#2 A very corporate culture. Developers could not wear jeans in the office. You needed come to work in business casual or dressier attire. And this was one of the innovation centers.
#3 A "corporate" rollout of TDD where a bunch of enterprise developers who hated TDD were given TDD training and then had no requirement to use it for 6 months....plus none of the other practices that go with it. In the end most devs never adopted the practice. (I am a TDD advocate and trust me I tired to get them to use a different approach but it was like talking to a wall)
#4 Crazy bureaucracy to get anything done. Servers need to be in a private double secure cloud (for non production non sensitive components) and developers worked for over a year with out any actual servers to run code on, and were not allowed to stand up servers under you desk as a stop gap due to corporate security. It would even take weeks to even get basics like office supplies that had to go through a special workflow.
#5 During this period of no servers, I was in charge of getting the CI infrastructure set up. One time I had a six sigma black belt ask me the status of the project. When I told him it was on hold due to lack of infrastructure he said "Lack of servers is no reason for not being able to roll out CI in our organization"
#6 Lots of training for things not needed for your job. Six Sigma (the only useful thing was lean which could have been taught in a quarter of the time). There was a good 20-40 hours of mandatory training per year (at least) for things relating to corporate compliance, safety etc., but not things relating to your job.
#7 For one project the use of Javascript was prohibited in favor of using a server side technology that generates it because the enterprise architect thought it would not be realistic to expect Java developers to learn or be able to also know Javascript.
#8 It was like pulling teeth to get time off to attend a conference even if you paid for it with your own money.
#9 Lots of C-player engineers were promoted because they could play the politics while the good ones who actually could get things done/code left.
All of this was over a 2 year period just a few years ago, after Immelt declared they were a software company at one of their newly established "Innovation/Centers of Excellence". Needless to say I have moved on to more interesting work.
Though this is a little better than what was described during their outsourcing binge I just don't see GE ever really being able to take on software effectively. Their corporate culture is just not set up for it. They would be much better either having completely separate divisions that they own a stake in but do not manage in any way/shape/form or partnering with companies that do software well.
I would love to be wrong, but my guess is that a few decent engineers will join based on the slick ads, and maybe a updated office, things will fall apart and they will move on.
Say what you want about Steve Balmer but he had the right attitude towards that https://www.youtube.com/watch?v=kHI7RTKhlz0&feature=youtu.be
As to the rest, I get the feeling that a lot of players in the market that now need software have the exact same tales being told about them. It is going to be a wonderment to me if I will be about to get through a day in 10 years without a IoT failure.
https://www.youtube.com/watch?v=YDbfudoXjBM&list=PLxRhTjvLly...
The article doesn't mention any startups. Any ideas what startups could be potential disruptors?
It's official now, we're just chickens to the world.
So the meme finally sticks and now we're screwed for the next decade? Who do we blame for the idea that "impromptu meetings" and those baloney is the way of working for programmers?
Well, small glass offices instead of cubicles, with million dollar views of course.
I've worked open plan solo and paired.
Solo: I have ADHD. It's a nightmare.
Paired: I have ADHD. I barely notice the background.
You guys have had quite an influence :)
For Labs, our best outcome is to put ourselves out of a job because the clients are teaching themselves.
On the other hand, it seems like people doing pair programming in open plans would contribute to the overall noise and distraction- a warehouse full of pairs of people talking all the time would also be a bad scene. :(
I can't speak for other companies. We don't do open plan for money, it's just that it's easier to work this way for the way that we work.
That said: for colocated teams of solo developers, or distributed teams, open plan is poison.
While cost IS a factor, the idea seems to be going round that open plan is genuinely superior. (Although rarely so superior that the executives partake...)
Hell why doesn't GE just remodel their offices to look like garages, because obviously that's the reason that Apple was successful...
Have you never had an interesting conversation and then looked up, realising that 4 hours had passed?
In my experience, I find it easier to focus with someone next to me than by myself. To the point that I take half as much Ritalin on pairing days than when I work alone.
This couldn't be farther from the truth. See The Joel Test.
The main difference is that they seem to be dressed in a bit more "business casual" than your typical SF startup in a warehouse.
Like Google, they should open up a large tech office in NYC, where many top level software people want to live.
More importantly, Boston is much heavier into manufacturing technology, hardware, B2B software, etc whereas NYC is more focused on consumer, retail, fashion, etc. Boston is pretty clearly a better choice for GE.
In reality GE's HQ in Boston will represent a tiny fraction of their global employees, though, so it doesn't really matter. https://www.bostonglobe.com/business/2016/03/11/huge-but-its...
And they don't compete with every startup in NYC for those brains.
Far more, he says
Anyway not true
NYC has alot more capital floating around but the subset of it that is VC isn't more than Boston
More importantly, industrial Internet of Shit developers, smart grid companies, and related activities are going on at the Seaport District, right where GE is setting up shop.
They were a customer of mine, they used to incorporate pencils into leases of capital equipment to reduce tax exposure.