Great, so now OpenAI has opened the door to pricing people out of AI access.
The o1-pro model in their charts is only ever so slightly better than the one I can get for $20 a month. To blur the lines of this they add in other features for $200 a month, but make no mistake, their best model is now 10x more expensive for 1% or so better results based on their charts.
What's next? The best models will soon cost $500 a month and only be available to enterprises? Seems they are opening the door to taking away public access to powerful models.
Why not, if people are willing to pay? You can think of them as subsidies for the weaker models. They're determining the price elasticity. And the better models will eventually get cheaper, as competition encroaches.
> What's next? The best models will soon cost $500 a month and only be available to enterprises? Seems they are opening the door to taking away public access to powerful models.
I can truly relate to poor management. I'm unsure though if the author is referring at times to poor engineering/HR managers, or poor product managers. At my company they call the latter Product Owners but it's the same thing I believe. I deal so often with both being terrible. I've had terrible Engineering Managers, who simply don't understand the tech, and in the worst case they think they do but don't. Add in some office politics and they can make life truly unbearable. Same for Product Managers but it's a whole other array of chaos. I've had a tough time with poor managers, and so I can really relate to that aspect of this.
What I wonder though reading between the lines is if Gitlab has reached a point where they suffer from poor Product Managers/Product Owners. I see that in later stage startups and especially large orgs. One would hope Gitlab, by nature of its product space, would never suffer from that, so if they are that's disheartening. I'm unsure what at this point would really differentiate Gitlab from Github unless it's devex, and to have good devex they'd need engineering centric product vision, I hope Gitlab is not losing that by hiring an army of "Business Analysts" or MBA's. On the positive, in some respects I like hearing that HN comments drive the vision at Gitlab :)
I really relate to aspects of both of these comments above. 1. There is a last 1% problem, heck it's probably more like 10% problem with LLM generated code similar to the last mile problem with driverless cars.
2. The bigger thing that will impact the job market is, everyone, at global scale, is being told to learn to code. Every middle schooler in the US is seemingly assumed a failure by their school if they don't "learn to code". There will be so much exponential growth in junior devs over the coming years. Looking outside the US, software engineer salaries aren't that great. Can the US economy create an exponential number of high paying engineer jobs to offset the global exponential rise in qualified engineers trying to get those jobs? Companies will offshore via third parties like Accenture who will skim off the top the livable wage away from those people outside the US who Accenture offshores to, so even those people outside the US won't be winning. Any way you look at it, I believe there will be massive world wide pressure on US engineer jobs.
I can't agree more, totally opaque data will almost certainly be sent to an ever growing number of locations. If I so much as ask about it I'll be labelled a nutter.
Think, anyone young enough to be a trend-seeking kid when smartphones were around. So yes, this includes some millennials. iOS well and truly won the “culture war”.
It’s just kids being kids.
But I’m sure I won’t have to hold my breath long to see this shovelware ragebait article catch HN readers hook, line, and sinker. I can already see the paragraph-long comment about how Apple’s continued dominance represents the downfall of society. When in reality we may as well be talking about Team Edward vs Team Jacob, or Team Rue vs Team Jules, or whatever.
I used this software to design my tiny house that I built with great success. It really helped me visualize the space. I didn't create an engineering ready floor plan with this so to speak. In other words, I intentionally left the measurements slightly relative which wouldn't work for most construction projects. In other words, if my bedroom ended up being 14ft by 10ft and 2 and 1/2 inches instead of 14x10' I was fine with that, it just meant other rooms slightly smaller. But give or take the width of dry wall and other factors it was worked fine. I found getting super precise measurements with this app difficult and so I didn't even try. I just let the guy building walls have a tad bit of freedom. Like I said, for most projects that would not work. Having said that, this app was a huge life savor and allowed me to really visualize the space and make key design decisions.
We believe this is the next generation of ads after the online advertising bubble pops. They can reach many people repetitively, although are unobtrusive on the horizon unlike billboards.
We want to facilitate the placement of logos on the surface of city streets, and give 90% of the advertising revenue back to the cities. This is an innovative new funding source that will generate substantial revenue for social issues and infrastructure needs, all without increasing taxes, fees, or debt.
Something I have wanted to ask the HN community, that this article brings up: what exactly does a scrum master do?
From what I have observed at one business, there is one scrum master on every team. Many of them know nothing about the business that business does, and have no technical background. I have seen developers spend sometimes an hour or more a day explaining to scrum masters what they just built, so that they can then go "communicate that up" to upper management or project management in opaque meetings developers aren't allowed to attend.
But what I can't figure out, is what they are actually supposed to do. At best, I see two points:
1. They do the traditional work of a project manager but on a team level, and communicate status up.
2. They are supposed to be some kind of "thought leader" on agile methodologies.
Can HN help shed some light on what this role should do? Perhaps my background has just been a bad experience in this regard. I suppose where I get stuck is, this seems to be the work that a dev lead used to do. Why make a full time job for a non-technical person to lead a technical team, below the management level?
In my company, each team has a scrum master (a SM being part of 2/3 teams). They have near-zero technical knowledge, and this is what they do (no particular order):
- Give an organisation to the team, and make sure that this organisation does not decay with time
- Help the devs work faster by helping them with the methodology
- Help identifying & solving problems (a problem goes from "a dev spent one hour more than expected on this ticket" to "the client has no vision on what the product should be"). This is probably the most important task
- Help the client to lead his project. This is a huge part too, the client does not know how to lead a project (they just came to us with a business problem they want to solve): there's no way the project will succeed if the client doesn't manage to have a clear vision for his product, prioritise tasks, build indicators, get user feedback...
Minor tasks:
- Organise meetings/demos
- Lead the meetings, make sure they're productive
We're a service company without management, "communicating status up" is a concept that doesn't exist.
> This is probably the most important task - Help the client to lead his project
For someone to do this, they would have to have client communication skills, at least somewhat thorough understanding of software development or the process, and an understanding of the client's business. I see this as a good thing. However, your example sounds to be a company that has external clients that you write software for. Do you not have account managers/principle types who interact with the clients, or is that more or less what a scrum master does there? Just curious, if you were an internal only dev shop that just develops software for stakeholders at your own company, would you see the role of a scrum master as significantly different?
I've never had a scrum master across 7-8 companies. Usually it was a rotating role, or a manager/tech lead. But certainly not one person dedicated to it.
spotify seems to have a few professional scrum masters and it seems to work for them. I'm less interested in typical cynicism and more interested in how this gets done -well-. Any spotifier's care to comment on having fulltime scrum masters?
edit: It seems they are called "Agile Coach"es.. is that any different?
>I have seen developers spend sometimes an hour or more a day explaining to scrum masters what they just built, so that they can then go "communicate that up" to upper management or project management in opaque meetings developers aren't allowed to attend.
Your scrum masters really aren't. The SM role is supposed to be a few minutes a day, and focus more on how people work rather than the technical details.
Examples are things like telling management that the team has often missed deliverables because a key resource (some piece of hardware, etc) is often unavailable. Or even telling managers that their employees think one of the manager's policies sucks.
To an extent, it's about holding the team accountable as well: Why are some people working on a lower priority item when no one is working on a higher priority item?
Don't get me wrong - I'm no fan of scrum, and am not saying a SM is needed. But, as with management, a good SM is a valuable contribution to the team.
The o1-pro model in their charts is only ever so slightly better than the one I can get for $20 a month. To blur the lines of this they add in other features for $200 a month, but make no mistake, their best model is now 10x more expensive for 1% or so better results based on their charts.
What's next? The best models will soon cost $500 a month and only be available to enterprises? Seems they are opening the door to taking away public access to powerful models.