Posts by TM⢠68 publicly visible posts • joined 18 Oct 2021
AI is credible because we have dumbed the whole of human society down to a number.
AI can be a doctor because it can get the pass mark in the tropical diseases exam.
AI can be a programmer because it massively increases our velocity.
AI can be a boss because it temporarily increases our quarterly earnings report.
We're taking everything that makes modern life suck and raising it to the power of 10 with AI.
I've worked at many companies that claim to do agile and all it was, was micromanaged, salami sliced, waterfall. The clue is in the word 'do'.
I've only worked at one company that *is* agile and that's because I was the one who introduced it. We were able to increase our regular (paying) using count by more than two orders of magnitude with virtually no change to the dev team size. Going from tens of thousands to millions. The company is a market leader.
I strongly encourage people to actually read the manifesto and it's 12 principles - it only takes a few minutes. Real agile is nothing like Agile industrial complex agile. It makes programming fun again (you know like the stuff you do in your own time but pays). The only thing I would add is what I call the zeroth law of Agile. It's so obvious the authors never thought to mention it, but it's generally the main thing management don't get about software development:
Software development is primarily a knowledge gaining exercise and is not about managing preplanned work.
Oh I thought I was the only one! I literally wrote the same to a paint company last week. Default search should be AND. How difficult is that? It's hard to know whether the default OR approach is either gross incompetence or some manager's idea of driving trade to other products. Just drives me away. I also use the same Google tick you mention above, although Google is not what it use to be.
Customer: "I'd like a gold chain please"
Retailer: "What about this bike chain sir?"
C: "No Thanks"
R: "What about this dog chain sir?"
C: "No Thanks"
R: "What about this gold pen sir?"
C: "Goodbye"
Tell me how you will measure me, and then I will tell you how I will behave. If you measure me in an illogical way, don’t complain about illogical behaviour. — Eli Goldratt
The heads of these cost departments are almost certainly measured by one thing and one thing only - how much does it cost for their department to *claim* they provide the required services - by which I mean tick a box on a strategic plan. Certainly not on the quality of that service or its long term impact on the company. This only leaves two questions. When outsourcing to AI and India:
1. Which one is cheapest?
2. Which one is most likely to fail the Turing test when asked for help?
I tend to use Ubuntu (usually KUbuntu because I like my UI's to be self documenting) because it just works (TM), but stuff like this adds to my general unease towards the distro with me often considering Debian or Arch (which are less just worky). Obviously the Centos debacle has put me off RHEL/Fedora/Openshift (for both paid and unpaid installations - irony) - I just can't trust IBM anymore. Here's to hoping a good Linux survives not driven by corporate stupidity survives.
P.S. If Canonical's employee assessment process is anything like their hiring process I'm not surprised this has happened. I understand they're heavily into metrics, which works fine for assessing whether a server is overloaded but is less good for assessing something as complex as an employee (see title of this post). It's ironic because any good manager knows where their employees are at, but then they have to been hired by a good manager who isn't beholden to useless metrics (and so on).
The problem is most managers are using a 19th century industrial revolution manufacturing paradigm to manage a 21st century research and development (AKA knowledge gaining AKA programming) project. The manufacturing part of programming is trivial, especially with modern DevOps. You commit the branch, it gets tested and built. Manufacturing. Done. The challenge is designing the solution and that can't be managed like a production line turfing out Model T Fords.
With one tiny rule the US government may have just saved Intel and it's CEO Gelsinger. I bet he's absolutely over the moon at the 'imposition' (not to mention the free money).
Intel went off the rails because of classic American MBA greedy accounting and mangement. Gelsinger was trying to turn it back into an innovator but the bean counters and investors forced him to double down on the old policies that was destroying Intel, not least off loading its manufacturing. With this rule Gelsinger now has an excuse to turn Intel back into an engineering company. Without it AMD and NVidia were going to eat Intel's lunch and pop the bag (yes I know they're fabless, but Intel needs to be innovative not follow).
It's an interesting point. People do often paint Linux as a server success only, but if you include Chromebooks and Phones it's kicking ass. I just wish I had the level of control over my phone and Chromebook that I have over my Linux desktop computer. The former are merely consumer devices but the latter is a productivity device. I wish I could be productive on all of them.
Climate science:
Start with 1,000 versions of a climate model with different parameters.
Check them against previous data and throw away the 750 versions that don't match.
Run for a year. Throw away the 200 versions that don't match the new year's data.
Run for another year and throw away the 40 versions that don't match that data.
Use the remaining 10 versions to make public policy, take money from those who work for a living (but not those that don't) and restrict people's rights and freedoms.
Paint anyone that disagrees with said policies or science as someone equivalent to those who would cover up the state sponsored murder of millions of people within living memory.
N.B. This science can also be applied to the horse racing and the stock market.
I had a ZX81, Spectrum +, +2 and +3. Probably the cheapest professional education any parent could buy (I work in the IT industry despite my degree in Physics).
I use to argue that my specy was better than my friends' commodore 64s. I was so wrong. But, it was cheap.
I went over to the dark side for my Amiga which is still far and away the fastest, most amazing computer I've ever owned. Everything since then has been a bit meh.
Yes, feature factories are usually a disaster, but this reads like another straw man, anti agile argument by people who don't know what being agile is.
Agile is just inspect and adapt. It works because software development is primarily about gaining knowledge rather than defining everything up front (at the point of least understanding). It doesn't stop you from writing reliable, secure code - just the opposite.
And on the subject of 'comprehensive documentation'. That is in the manifesto because software code IS documentation. Unlike stuff written in confluence by people spread all over the world in their second or third language, code is an unambiguous description of what the computer should do. If you employ the right people it's also human readable.
CUDA, that four letter word that means I have to buy an NVidia graphics card for my development work and pretty much pay whatever they are asking. If I was the competition I would be pouring money into coming up with something free, open and good that sets customers free from this particular monopoly.
If it wasn't for CUDA I'd buy an AMD card, even though they're not as good they're prices don't take the micky.
The problem is that replacing these large legacy projects is only successful if you take the smallest possible bit, migrate it, test it, fix it and watch it, then take the next smallest possible bit and so on. This approach simply doesn't fit into the 'give us a fixed, massive upfront cost and fixed scope quote' bucket that manglement wants.
I 'invented' UBI back in the 90s and use to talk to people about it. They politely listened but I think it was generally seen as extremely 'left field'. I let it slide, thinking it would never catch on. I've been amazed to see it now become a mainstream consideration. At the time I thought it was a great idea, but now I have some doubts:
1. If those in charge didn't deliberately create and hide (through deliberately massaging the figures) massive inflation while clamping down on real wage growth we would need UBI far less. I reckon real inflation has generally been roughly 10%pa since the 70s and more like 20%pa recently - that has destroyed real wages while the assets of the rich have just adjusted for the currency devaluation. If Sam Altman wants to do an experiment I challenge him to pay all his employees in the equivalent value of gold (a la pre dropping of the gold standard and the introduction of deliberate wage devaluation). That wouldn't be charity it would just be acting fairly.
2. There is a massive risk that governments will use UBI as leverage over the populace - it would have to be impossible to loose it. I can't imagine any government being that self controlled.
Most of the places I've worked at over the last three decades claimed to be agile - only one of them came close. The one where I was CTO and introduced agile. We increased user count by two orders of magnitude, going from tens of thousands of regular users to millions.
Agile is almost certainly the antithesis of what you have experienced at an 'agile' company. I encourage people to read the manifesto and listen to people who actually know what it is like Alen Holub, MArtin Fowler and Ron Jeffries.
Agile is a mindset, it is not a process, tool or framework you do.
TLDR: Would you rather work at a company that 'does' kindness or is kind?
IIRC a law was passed (locally somewhere in the US?) to prevent local community run cooperatives from installing their own internet infrastructure or getting government funding to do so. Those damn socialist commy bastards.
While nowhere near free, installing fibre is not crazy expensive once you have the initial investment (fuser, micro trencher etc) - which is small once divided by 100+ homes. Multi cable single mode can be laid for approx $1k per km. The problem is bureaucracy, port barrels, and getting people like bridge companies, telegraph pole owners, not to take the piss when charging 'rental'. Ask me how I know.
Like most problems us riff raff have to put up with (e.g. expensive houses) the first step is to just stop those with the power from deliberately make things worse - the whole 'if only we could fix this' narrative is total BS.
What if he was wrong? Here's some differences I can think of:
Software is elastic - through the wonder of digital copies it costs nearly the same to build (as in compile etc) a billion as one.
Software can be delivered immediately, anywhere in the world (with internet) at any time.
Software is easy(ish) to revise (even completely) after it has been delivered - compare that to a bridge or skyscraper.
In software the challenge is designing the first one, in traditional engineering much of the challenge is building the design.
In traditional engineering designs are typically very conservative and the costs of mistakes is high. In software you can experiment for little cost.
And so on.
I've worked at lots of places that claim to be agile, all but one of them (where I was CTO) had nothing to do with agile.
I've skim read the objections in the comments and they don't represent the values and principles in the agile manifesto.
I really do wish people would read it.
https://agilemanifesto.org/
In my experience agile is far and away the best way to create software.
Here is a good talk from Allen Holub on the sort of 'agile' described by the complaints in the comments above.
https://www.youtube.com/watch?v=HZyRQ8Uhhmk
If worked on many Agile projects and none of them were agile - bar one. The one where I was CTO and we were agile. We increased user count by two orders of magnitude and transformed the company into a massive success.
Agile is not a methodology or a process or a practice. It is something you are (based on the dictionary definition of agile). It is based on understanding that software creation is not a repeatable manufacturing process but a R&D process where you are gaining knowledge. The manifesto and its principles merely highlights some of the things that flow out of pragmatically accepting these facts.
Agile is far and away the best approach I've seen to creating software. Please don't confuse your experience of the Agile Industrial Complex, with real agile. Scrum, SAFe etc is not agile.
See https://martinfowler.com/articles/newMethodology.html for a primer (ignore the stuff at the end on the implementations nobody uses any more, it's a very old document).
I'm afraid McKinsey lost what little credibility they had in their infamous measuring developer output report. Them complaining about a situation that they and their ilk are directly responsible for doesn't help either.
We've been in race to the bottom for the last fifty years. We've even avoided the effort of walking down the stairs by jumping out of the window. Here comes the ground.
Seems overly complex to me, but like most of the other posters I agree something needs to happen. I am loathed to do anything to OSS licenses that make them less attractive. The musician analogy/comparison doesn't bode well - musicians really don't so well out of similarish systems. Good to see someone try though. I wish you all the luck. Some thoughts:
1. Someone needs to take IBM to court or close the loophole. To me their restriction is just that and thus illegal, but let's fix it one way or another.
2. Using the code as SaaS just needs to be counted as distribution and customers given the same rights to have the source code as that used on the server. A tweak to the GPL?
3. My guess is hyperscalers et al. would be happyish to throw key OSS developers some crumbs (upto N x $100k pa) for better code if there was a single central organisation that made it really, really easy to do so - i.e. removed any hint of employment liability, and dealt with (self)employed tax and sales tax around the world. The organisation would have to do some due diligence to make sure the right people (and not just the noisy ones) got rewarded. Perhaps throw in some sort of enterprisey like support thing to make the C level suite happy. This would be voluntary, but maybe my opinion of the hyperscalers et al isn't low enough!? They might just do nothing. Perhaps have a public record of who the good and bad guys are?
"Whilst Post Office is responsible for overall programme management, strategy and architecture it has a requirement to engage with third parties to help create the necessary solution," the notice said.
Which is why it will fail, because software creation is in large part: programme management, strategy and architecture. Only someone who mistakes a knowledge gaining exercise like software creation for manufacturing would think of using nineteenth century Taylorism to manage it, i.e. "We've come up with the plan, we just need someone to bolt it all together".
New flash: No software plan survives contact with the enemy. Software programming is not the twenty first century equivalent of brick laying.
Obviously no lessons have been learnt.
Human society has been dumbed down to the point where a mathematical algorithm can now take over. Principally, this has been done by making everything about a single short term number, i.e. cost. For example:
- If I sack my most experienced (AKA expensive) engineers will the share price go up?
- If we keep printing money will our exports get cheaper, our welfare liabilities go down, our civil service costs go down?
- If we focus on delivering on time and on budget (regardless of quality or value) will I get a raise?
- If we internationalise trade will labour costs be driven down?
For the last fifty years we've been in a race to the bottom. Now we've automated it.
Big tech is a monopsony and appear to be working in concert.
The west's race to the bottom on education has produced a massive shortage of skilled people - especially in IT.
With this massive upward pressure on wages how else is big tech going to reduce the cost of employing minions?
Treat em mean keep em keen.
I must say I feel confused.
My understanding is the GPL gives recipients the right to redistribute the source code WITHOUT RESTRICTION. Red Hat threatening to take away a customers' licenses and support is an extremely large restriction. The fact that Red Hat claims they have a right to do that is irrelevant - it doesn't prevent it from being a restriction.
You're free to complain about the mold, leaky roof and rats in the kitchen, but if you do we won't renew your contract.
I for one will not be recommending any company I work with gets into bed with RedHat or IBM - quite the opposite.
The question is: Who can be trusted? I know we're all running to Ubuntu, but I'm left wondering if they're not able to pull the same stunt. Debian? Arch Linux? RedHat's approach had undermined the whole trust model open software is built on. I think someone needs to tweak the GPL to rebuild that trust.
Community to RedHat: "We've revoked your right to use our software on the grounds that we believe (rightly or wrongly) that you add no value to our businesses"
