This is absolutely why LLMs are so disruptive. It used to be that a long, written paper was like a proof-of-work that the author thought about the problem. Now that connection is broken.
One consequence, IMHO, is that we won't value long papers anymore. Instead, we will want very dense, high-bandwidth writing that the author stakes consequences (monetary, reputational, etc.) on its validity.
The Methyl 4-methylpyrrole-2-carboxylate vs ∂²ψ/∂t² = c²∇²ψ distinction. My bet is on Methyl 4-methylpyrrole-2-carboxylate being more actionable. For better or worse.
YouTube and others pay for clicks/views, so obviously you can maximize this by producing lots of mediocre content.
LinkedIn is a place to sell, either a service/product to companies or yourself to a future employer. Again, the incentive is to produce more content for less effort.
Even HN has the incentive of promoting people's startups.
Is it possible to create a social network (or "discussion community", if you prefer) that doesn't have any incentive except human-to-human interaction? I don't mean a place where AI is banned, I mean a place where AI is useless, so people don't bother.
The closest thing would probably be private friend groups, but that's probably already well-served by text messaging and in-person gatherings. Are there any other possibilities?
spot on. The number of times I've came across a poorly made video where half the comments are calling out its inaccuracies. In the end Youtube (or any other platform) and the creator get paid. Any kind of negative interaction with the video either counts as engagement or just means move on to the next whack-a-mole variant.
None of these big tech platforms that involve UGC were ever meant to scale. They are beyond accountable.
Exactly. People spend less time thinking about the underlying structure at play here. Scratch enough at the surface and the problem is always the ads model of internet. Until that is broken or is economically pointless the existing problem will persist.
Elon Musk cops a lot for the degradation of twitter to people who care about that sort of thing, and he definitely plays a part there, but its the monetisation aspect that was the real tilt to all noise in a signal to noise ratio perspective
We've taken a version of the problem in the physical world to the digital world. It runs along the same lines of how high rents (commercial or residential) limit the diversity of people or commercial offering in a place simply because only a certain thing can work or be economically viable. People always want different mixes of things and offering but if the structure (in this case rent) only permits one type of thing then that's all you're going to get
I think incentives is the right way to think about it. Authentic interactions are not monetized. So where are people writing online without expecting payment?
Blogs can have ads, but blogs with RSS feeds are a safer bet as it's hard to monetize an RSS feed. Blogs are a great place to find people who are writing just because they want to write. As I see more AI slop on social media, I spend more time in my feed reader.
I think manifestos are useless without a concrete, real-world example for people to follow and add on to. It's easy to wish for puppies and rainbows, but trying to deliver is hard.
For example, Linear has a useful manifesto (https://linear.app/method) because they have a product that attempts to follows it. I have much more respect for a manifesto that is informed by contact with reality.
I agree, but I always assume manifestos are distillations from experience.
Is it that you want to be able to inspect the experience that informs it side-by-side, in case-studies or product or something?
I take it for granted that you're not sceptical of the authors experience, because lord knows there's some experience behind the contributors and signatories :)
Maggie Appleton
Samuel Arbesman
Daniel Barcay
Rob Hardy
Aishwarya Khanduja
Alex Komoroske
Geoffrey Litt
Michael Masnick
Brendan McCord
Bernhard Seefeld
Ivan Vendrov
Amelia Wattenberger
Zoe Weinberg
Simon Willison
I'm used to manifestos being a call to action or, at least, an explanation of intentions. But if the signers are intent on changing the world, I don't have a concrete view of what world they plan on creating.
"people must serve as primary stewards of their own context": What does this mean? People must be able to change the system prompt whenever they want? Or does "their own context" just mean "their own prompt" but not the system prompt? Does this mean everyone needs to run a local model?
"You must be able to trust there are no hidden agendas or conflicting interests.": Any transaction has conflicting interests. I want to pay as little as possible for software, but I want my salary as a software engineer to be as high as possible. How are they planning on eliminating that conflict?
"No single entity should control the digital spaces we inhabit.": What do they mean by a "digital space"? Do they mean the Internet? If so, then this is already true: no single entity controls the internet. Do they mean any web app that I might use? If I put up a personal blog can I be the only one who controls it? What if my blog gets really popular?
"Software should be open-ended, able to meet the specific, context-dependent needs of each person who uses it." Also, it should take 0 time to manage, solve any problem I might have, and be absolutely free.
"Technology should enable connection and coordination, helping us become better neighbors, collaborators, and stewards of shared spaces, both online and off." This was literally the original dream of the internet. We thought a global network that connected everyone would lead to peace and prosperity. We all agree that it hasn't. No one has any idea how to achieve this other than banning it.
I'm sure we can come up with answers to each one of these questions. But my point is that the manifesto doesn't tell us. We don't know what they have in mind, which makes me suspicious that either (a) they don't know either, or (b) I might not like the answer.
> because they have a product that attempts to follows it.
I had no idea what linear.app was. Then I followed your link.
Then I had no idea what linear.app was. Then I went to their front page.
Then I had no idea what linear.app was.
Manifestos are fine. My personal one is that, if I can't tell immediately from your front page what the fuck it is that you do, then we aren't compatible.
I was working at Groove Networks, a tiny company in Beverly Massachusetts, when I learned that Bill Gates was dropping by. Founded by Ray Ozzie after his success with Lotus Notes, Groove had been negotiating with Microsoft to sell the company, and this was, potentially, the final step.
It was 2003, and Microsoft's next version of Windows, codenamed Longhorn, was not doing well. On top of that, Google and the rest of the internet natives were eating Microsoft's lunch, and many analysts expected the company to fade away, as IBM had done before.
Bill Gates, obviously, didn't want that to happen, and true to his nature, was looking for a technological solution. In many ways, Longhorn was fighting the last war: it invented a new graphical subsystem and a new storage system, at a time when modern apps were happily using HTML and SQL. Groove Networks had developed a peer-to-peer synchronization technology that blended online and offline so an app could have the best of both worlds. It was a true local-first architecture that was also internet native.
Gates arrived early in the morning with one or two assistants. In 2003 he was still the wealthiest person on the planet, but he carried himself like any normal engineer. One of my colleagues, who hadn't been told he was coming, only learned about it because they rode with him in the elevator. He introduced himself and made small talk. You can imagine their shock.
At the meeting, we presented our technology and our ideas for how we could fit into Microsoft's plans. Later I learned that this was a mini-product review, like Sinofsky talks about in the OP. I pitched him (somewhat half-baked) ideas about how Windows files folders could become collaborative, shared folders. He was very engaged, but thankfully polite--he didn't tell me it was "the stupidest thing he had ever heard." I suspect he was on his best behavior because he had already made up his mind to buy the company.
The highlight of the meeting, for me, was watching him and Ray (whom he'd known for a while) riff on everything around Windows, the internet, and technology in general. It was like improv, where every idea someone came up with was followed up with, "Yes, and then you can also...". You could tell Gates was engaged because he kept rocking in his chair, a classic tell we later learned.
Microsoft did ultimately buy Groove Networks, but not for the technology. I realized much later that Gates had bought Groove mostly to hire Ray. In 2006, Gates announced that he was retiring and appointing Ray Ozzie as Chief Software Architect.
At Microsoft, Ray spearheaded a project codenamed Red Dog, which the marketing folks later called "Azure". I'm convinced that Ray doesn't get enough credit for his contributions to turning Microsoft around.
Had Bill already planned all this when he visited us in 2003? Probably not, but seeing his mind work, it wouldn't surprise me if he had a little bit of an inkling. You never know.
Just wanted to say, I really enjoyed using Groove. It was a memory hog and a bit flaky, but I thought it had tremendous potential and was conceptually really interesting. I was sad when it disappeared
1. Sometimes we don't know what is most valuable (from the company's perspective).
2. It is easy to convince ourselves that whatever we want to do is really the most valuable thing (e.g., "Refactoring this massive subsystem will help the company in the long-term" or "Introducing this new technology (that I really like) will make it easier to recruit talent.")
In my view, the meta-advice is to understand the goals and constraints of your boss (and their boss), and work towards those goals (while adhering to the constraints).
With that perspective, we can derive some rules of thumb:
1. Promotions are not a reward for past performance. Instead, they are a bet that you will contribute more towards those goals with a promotion than without one.
2. As the OP says, if you are demonstrating performance at your boss's level, that's evidence/proof that a promotion is warranted. Your boss's goals get implemented (by you), freeing them to work on their boss's goals (and maybe get their own promotion).
3. The more time you spend with your boss, the better you will understand their goals, and symmetrically, the better they will understand your strengths. That means leaving a job after a year or two is not always optimal. It also means following a good boss to another company is often a good move.
4. There will be cases where the goals of your boss (and their boss) diverge from your own goals. They often want to cut costs, but you want a salary increase. There are never easy answers to this dilemma, but seeing their perspective is useful so you can find a win-win scenario. E.g., if you come up with a way to save money in other ways, such as automating an external cost, then your increased salary will be worth it.
5. In some cases, of course, there is no way to reconcile your boss's goals with your own. Realizing that is useful so you can find a different company/boss that is more aligned.
> 1. Promotions are not a reward for past performance. Instead, they are a bet that you will contribute more towards those goals with a promotion than without one.
It's both.
You reasonably can't keep someone in the same position for 5 years when their market value has long gone past that point and they're expecting more. Even if you're not sure they won't be Peter principled out in the better paying position.
The better way if to have an internal pay scale that allows for more specialization without more responsibility, but that's IMHO rare and requires managers that can handle that.
> demonstrating performance at your boss's level
To note, it often results in advices close to "do X job for a while and we'll let you have it", which looks like a no risk move for the company but is not without downsides. I've seen people being half managers for a full year before becoming one, and boy does it kill morale.
It signals to employees they'll be literally working about their pay grade "for free" for an undefined amount of time, and it's an even worse proposition when they're effectively doing two jobs at the same time (they're still expected to excel in their current position while proving they can do the other position as well)
It's a more delicate balance than it might look at first.
And I agree that, taken to an extreme, this is abusive towards employees. But I think most (good) companies handle this pretty well.
I've seen a couple of patterns:
1. Your boss trusts that your instinct are aligned with theirs, and gives you more latitude. Maybe they allow you to design architecture your way rather than requiring detailed review. Maybe they delegate reviewing other people's code to you.
2. You understand enough about your boss's goals/constraints that you can represent them. E.g., they might trust you to represent them at a cross-functional meeting.
Either way, your name will come to their mind when promotions are available.
> if you are demonstrating performance at your boss's level, that's evidence/proof that a promotion is warranted
If you're an engineering IC, and your boss is a manager with 4 other ICs, your boss's goals are twofold: get at least 5 ICs worth of results from the team, and managing people.
So to do what you and TFA suggest literally you can either:
- Do 5 ICs worth of work
- Start managing people at the same level on your team, on your own initiative
I've seen coworkers try to manage their peers, aiming for a promotion. To say the least it harms team unity.
I only managed to do the 2nd once when I was thrown into a project with an absentee manager and doubly-booked half-committed members who were actually happy for someone to organize the work. Those sorts of situations are rare. Or maybe that's the unstated qualification.
And: Do 5x the amount of work, well...
Maybe I'm not thinking outside the box enough here, but I need some examples of how this is generally achievable. Maybe this was specifically _not_ about the IC-manager divide, and more like managers and manager-managers?
What I'd more generally expect is for a manager to explicitly put you in charge of a small, short term project with one or two other people and see how it goes: can everyone contribute, did you achieve results, were you transparent, how did you interact with the other members, etc.
This is a good question. I compressed too much: instead of "performance at your boss's level" I really meant, "helping to achieve your boss's goals".
If you're an engineering IC in a team of 5, what are your boss's goals? It's usually things like: hit your deadlines, avoid production bug catastrophes, and maybe add features that make the sales people happy.
How can your boss achieve those goals? I have a few ideas:
a) Processes: Introduce or refine processes for the team to ensure high-quality code or to gain efficiencies.
b) Mentoring: Help members of the team to function at their highest level.
c) Clearing Obstacles: Coordinate with other teams so they don't slow you down. E.g., make sure teams you depend on are on schedule, and if not, adapt and adjust.
But this is just an example. I think the easiest thing to do is ask your boss what their goals are. What does success look like to them? Once you know that, you might be able to come up with ways of helping that they might not have thought of.
This sounds like advice for how to be promoted to a specific level -- the first point where awareness of things beyond yourself is required (somewhere around the Senior or Staff level for ICs, depending on your company).
Generally everyone in a team should be working towards some shared goal, there's no level at which you can be a chaos agent and not serve some higher purpose. The difference at this level transition is that you realise that for yourself -- someone doesn't need to remind you of the goal and nudge you back on course. That same realisation is not going to cut it at higher levels.
For me the general version of this advice is not something you can just tell the person who's being promoted, it's collective advice, for them, their manager, their tech lead: everyone needs to agree that this person needs to be given more rope, they need to do something useful with that (i.e. not hang themselves with it), the people around them need to watch out for when they start tying a noose and help them untie it (already regretting this analogy), and that's how you get promoted.
The rope takes different forms for different levels. I'll use the level scale I'm familiar with, starting with a newly graduated engineer at L3:
- L3 -> L4. You help decide how to build the feature.
- L4 -> L5. You help decide what features are worth building, and are trusted to maintain them.
- L5 -> L6. You help shape the work and ongoing maintenance of ~10 people's work (what products are worth building and how), over a time horizon of 6 months to a year.
- L6 -> L7. ~50 people's work, 1-2 years.
- L7 -> L8. ~200 people's work, 2-5 years.
- L8 -> L9. Things start to get fuzzy. The pattern suggests that you have a hand in ~1000 people's work, which is possible to do in the moment, but rare. There's two ways I can think of: you're either a world expert in your field, or you have set the technical strategy well for your organisation as it grew to this size.
This is just based on my experience, working largely on infrastructure teams both in big tech and in start ups as both an IC and a manager (currently an IC).
I think those are good examples. I think part of the confusion is that most of those are typical responsibilities of e.g. senior level IC work, so "performance at your boss's level" looks more or less the same as "performance at your current IC level".
I'd say it's about doing things at the next level to show you're ready for that level. So for moving from a Sr to a Staff position might involve doing more mentoring of the team, showing that you are using your knowledge to improve the efficiency of both your team and other teams, etc.
> Start managing people at the same level on your team, on your own initiative
Anecdotically, a coworken in my group started, on his own initiative, to “play manager” in out team, because he wanted to “help us all”. Of course he just wanted to ascend the ladder. That backfired instantly and spectacularly. I would never act with any authority if it was not very clearly delegated by my team, or my superior; and even then I would walk like in thin ice for the first 6 months
> 1. Promotions are not a reward for past performance. Instead, they are a bet that you will contribute more towards those goals with a promotion than without one.
Actually, you operate on the next level for certain amount of the time. You work with your manager to file for your promotion case. That's how the typical big corps work with promotions.
So technically, it is using your past experience to prove that you are operating at the next level
> Actually, you operate on the next level for certain amount of the time. You work with your manager to file for your promotion case. That's how the typical big corps work with promotions.
This has always struck me as a pretty juicy deal going for the corporation. They get N years of "next level" work out of you while still being able to pay those N years in "previous level" salary. Good deal for them.
How ridiculous the opposite sounds: You pay me at the next level for 3 years, and only then I'll know you're serious and will start working at that level. You'd get laughed out of the room. But the company has this exact deal in reverse.
> > Actually, you operate on the next level for certain amount of the time. You work with your manager to file for your promotion case. That's how the typical big corps work with promotions.
> This has always struck me as a pretty juicy deal going for the corporation. They get N years of "next level" work out of you while still being able to pay those N years in "previous level" salary. Good deal for them.
My current company used to work this way, but they moved to a "needs-based" promo process. You can be promoted to L5 if your manager can justify the need for an L5.
Which ends up making promotions significantly harder to come by. It's near impossible to justify the need for an L5 role when you already have L4s doing the work. No matter how far outside their level competencies a person works, that work becomes L4 work... because an L4 is successfully performing it.
I'm in this exact situation described in the two comments above.
I explained to my manager that the project I have been working on has developed a lot since the last two years and if he would hire a replacement he would be looking at a senior person, not a junior. He agrees but he gets rejected when he made the case to his boss. My performance reviews have been above expectations.
His boss claimed that it would not be fair to other people that stayed in the position for a similar amount of time before getting a promotion, essentially ignoring my exceptional performance.
My company, for e.g. is fairly flat, and my boss is more or less aware of everyone’s contributions in my team, he often works with them directly.
I also work with my report’s reports directly and am fairly aware of their work.
Despite this, some engineers, to my surprise, act as we have a strict hierarchy and try to reach to me through their managers.
From the sounds of your description, there are a few possibilities:
1. Your boss’s boss is aware of your work. She is also aware of others’ and she does not think that yours particularly stand out and she is willing to risk your departure. In this case, you would need to really look at this objectively. Are you really exceptional? Why does not she think so if that’s the case? Is there someone else who are also great (or giving that impression) that you are not aware?
2. She does not know you very well. If so, why is this the case? Does she not know anyone, or are you keeping your work to yourself? I’ve definitely been in this situation, despite architecting our whole core systems, years later I found nobody other than my fellow engineers knew. Was a hard-earned lesson for me, you need to start speaking about your work outside of your 1-1s, but not in a promotional way. By frequently offering your hard-earned wisdom where it is helpful.
3. She is not interested in knowing anyone. She will manage her team at a high level and she either won’t promote anyone until she is forced to (e.g. you are leaving otherwise), or when she is given a budget and asked for it, which she will then ask for recommendations, your chances than unlikely to be proportional to your work but be circumstantial. If this is the case, you should start interviewing.
One thing that I've seen implemented to prevent that is to have the pay bands for level N and N+1 overlap. So in the time that you're doing "next level" work, you're expecting to be at the top of your current pay band, and then the promotion doesn't automatically give you a big pay raise, but it unlocks a pay band that you can go up in.
This works if performing at the top of your current level equates to performing at the bottom of the next level. That said, there's a problem where sometimes a "promotion" is really a new role, meaning to perform at the next level, you have to kind of not perform well at the current level.
It's all about risk/reward tradeoffs. Once you get past the junior->senior level, each promotion is hiring you for a completely different job. As an individual, there are only a few ways to get that job:
1. Trial run at your current company (could be wasting your time, but also you have domain knowledge and relationships to help)
2. Join a smaller company and hope it grows (could rapidly accelerate growth due to needs, but could also go very poorly if the company stagnates)
3. Try to lateral to another company with a promotion (pretty difficult in general)
It's not really that juicy for the corp. If they hire (promote) you without experience, they are hiring someone without experience for a position and then have to go and hire again to replace someone else. Vs. just hiring someone with experience
> This has always struck me as a pretty juicy deal going for the corporation.
It's a good deal if you deserve the promo. Giving someone the opportunity to take on projects at the next level and having them not deliver can be enormously expensive. The higher the level, the more expensive it is.
Possibly. It's the only way it actually works though, because of the Peter Priciple.
Imagine the other way - you have peopel dong a role, and the people who do the best job at that role get promoted to the next one. Some of them will be good and the new role, some of them won't. The ones who are good will carry on getting promoted. The ones who aren't will get stuck in that role. The problem is that everyone rises to a point at which they can't do the job, and every role is filled by someone who has been promoted one step too far.
In a healthy structure, it should be a halfway house - you shouldn't have to be doing the whole job that you're trying to get promoted to, you should be doing enough bits and pieces of it that you demonstrate that you CAN do it. That way the company has information that they're not promoting you to a position of incompetence.
I suppose it balances in the end, though. If you could make more money elsewhere you'd go elsewhere, so the whole reason you are willing to accept being underpaid through the transitionary phase is because you realize that you will be overpaid afterwards.
How exactly do you suggest it should work, then? A timer starts and when it runs out you get promoted and everyone just hopes you didn't just get moved up above your level of competence?
> if you are demonstrating performance at your boss's level, that's evidence/proof that a promotion is warranted.
It can not be farther from the truth.
The best way to stay in the bottom is to work hard, to focus on work so that others have time to focus on advertising themselves, take credit of your good work and backstab you for everything else, befriend and lick the shoes strategically -even develop bed skills, for some- while you isolate yourself by sweating and believing everyone will understand or care about how you optimized that for loop.
Cynicism is a seductive drug. It makes you feel good because you don't have to do anything--the game is rigged, so why bother trying? But like all drugs it is ultimately self-sabotaging.
Careers are like love: you have to risk heartbreak or you'll never experience joy.
I don't think there's a one size fits all here. If you don't go out of your comfort zone and "do more" you may never get a promotion because you're seen as average. But it's also true that if you work hard and constantly deliver you may still never get the promotion because you're seen as critical where you are.
You might be disappointed either way. Like any recipe, there are many ingredients needed to pull it off. Delivering results, solving your boss' or boss' boss problems, doing it visibly, having support from above, doing it at the right time, etc. all contribute.
This is premised on promotions and other work rewards having any kind of rational basis or connection to the work.
It could simply be that spending time with your boss makes them know and like you more, and people tend to reward people they know and like, making up some post hoc rationalization about performance or whatever to justify it.
No one wants to think of themselves like this, though, so they would never admit, even to themselves, that this is what's going on, but I suspect for most people it's the actual reality.
> 1. Promotions are not a reward for past performance. Instead, they are a bet that you will contribute more towards those goals with a promotion than without one.
> 2. As the OP says, if you are demonstrating performance at your boss's level, that's evidence/proof that a promotion is warranted.
That's not evidence for 1. At least you haven't explained a reason why it would be.
I agree with the OP that "whoever owns the weights, owns the values". But by that criteria, Grok is an example to follow. Musk is very clear on his values, and we know what we're getting when we use Grok. Obviously, not everyone agrees with its values, but so what? We will never be able to create a useful AI that everyone agrees with.
In contrast, we don't know what values are programmed into ChatGPT, Claude, etc. What are they optimizing for? Alignment to some cabal of experts? Maximum usage? Minimum controversy? We don't entirely know.
Isn't it better to have multiple AIs with obvious values so that we can choose the most appropriate one?
Musk isn't clear at all. He trumpets "free speech" then literally censors objective fact-based criticism which annoys him.
The problem isn't Grok-on-X, it's that Grok is supposed to be a commercial product used by individuals and businesses.
Machines do not usually have values. Now we're being asked to pay for a service that not only has values which affect the quality of its output, but which is constantly being tweaked according to the capricious whims of its owner.
Today it's white supremacy, tomorrow it might be programmed criticism of competing EVs and AI projects, or promotion of narratives that support traditional corporations over threatening startups.
Do you really want to pay for a service that is trying to manipulate your values while you use it, and could potentially be used to undermine you and your work without you being consciously aware of it?
> It used to be that a well-written document was a proof-of-work that the author thought things through (or at least spent some time thinking about it).
I think you hit the nail on the head here. The problem isn't so much that people can do bad work faster than ever now, its that we can no longer rely on the same heuristics for quickly assessing a given piece of work. I dont have a great answer. But I do still think it has something to do with trust and how we build relationships with each other.
Before AI, if someone submitted a well-formatted, well-structured document, we could assume they spent a lot of time on it and probably got the substance right. It's like the document is a proof-of-work that means I can probably trust the results.
Maybe we need a different document structure--something that has verification/justification built in.
I'd like to see a conclusion up front ("We should invest $x billion on a new factory in Malaysia") followed by an interrogation dialogue with all the obvious questions answered: "Why Malaysia and not Indonesia?", "Why $x and not $y billion?", etc.
At that point, maybe I don't care if the whole thing was produced by AI. As long as I have the justification in front of me, I'm happy. And this format makes it easy to see what's missing. If there's a question I would have asked that's not in the document, then it's not ready.
One consequence, IMHO, is that we won't value long papers anymore. Instead, we will want very dense, high-bandwidth writing that the author stakes consequences (monetary, reputational, etc.) on its validity.
reply