With agentic coding people underestimate the agent and over estimate the models value. So it’s important to be specific. What agent are you using? You will see radical performance differences between Claude code and codex compare to copilot for instance. You will also see pretty big differences if you have well groomed, agent specific agents files. Especially if the code base is very large, the agents files need to be able to guide the agent to make connections in the code.
But other than that what I’ve found to be the most important is static tooling. Do you have rules that require tests to be run, do you have linters and code formatters that enforce your standards? Are you using well known tools (build tools, dependency management tools etc) or is that bespoke.
But the less sexy answer is that no, you can’t drop an agent cold into a big codebase and expect it to perform miracles. You need to build out agentic flows as a process that you iterate and improve on. If you prompt an agent and it gets it wrong, evaluate why and build out the tools so next time it won’t get it wrong. You slowly level up the capabilities of the tool by improving it over time.
I can’t emphasize enough the difference in agents though. I’ve been doing a lot of ab tests with copilot against other agents and it’s wild how bad it is, even backed with the same models.
I’m pretty confused by the innuendo in the posters comment to even get what they are saying, but I can report from the south side of chicago that surveillance cameras are largely popular across demographics.
You don't know anything about me or about Oak Park. We killed our Flock contract. I did a good bit of the policy lifting to make that happen, including co-writing our initial pilot ALPR police General Order to lock them down and set up the transparency reporting that enabled us to make the case for killing them. But, keep going, I'm interested in what else you can confidently state about my work.
Speaking as a person of color, the government Oak Park prosecuted me for reasons of my blackness which is racist. The controversey that followed, if you ask me, is why the cameras got shot down. Transparency and not being big brother is just a convenient excuse to put a positive spin on it, as politicians do.
Knowing in extreme and annoying detail exactly the series of events that led us to kill our Flock contract, I do not believe this is at all true. I don't think trolling is a good plan here.
You hypothesis then is that there is not a _single_ public company that has a healthy relationship with its company? Not one, in the entire global public space?
When does this relationship with customers happen? Is it at the IPO? When they file the paperwork? When they contemplate going public for the first time? Or is it that any founder who might one day decide to contemplate going public was doomed to unhealthy customer relations from birth?
The obvious next thing we in society should do is abolish public equity as a concept as a customer protection mechanism?
It is genuinely hard to think of one. I treat all companies as adversarial relationships, where I fully expect them to treat me as disposable at least over any time horizon greater than 1-2y. There are certainly some companies that are more likely to find a mutually beneficial equilibrium. I think of Target, IKEA, sometimes Apple. But I don’t trust any of those companies to take care of me in the future. But I also wouldn’t be the least bit surprised if my next interaction with any of those companies was bad. I just typically expect it to be more mutually beneficial than Comcast, Hertz, or Verizon.
Sure, but very modestly due to scale, not core institutional morals. Go to your average small business with 10 or so people and ask the staff how they are treated and paid, and you'll get an answer not much different than the level of employee satisfaction for Fortune 500 companies. Look a their customer reviews... are small restaurants for instance an order of magnitude different that megacorp chains? In an economy with regulatory capture and highly unequal distribution of wealth, the wealthy set the tone across the board.
From what I can see, it's often when the founder loses control of the company (either voluntarily (e.g. retirement) or not) and it falls to the board (representing the shareholders) to appoint the CEO. At that point it's at best a toss up whether they'll appoint someone who actually intends to create value or someone who intends to extract value.
> The obvious next thing we in society should do is abolish public equity as a concept as a customer protection mechanism?
Abolishing public equity is quite drastic, but there are lots of other things we could (and IMO should) be doing to protect society from the negative externalities it causes. For example:
- Mandating worker representation on company boards. So shareholders still have some power, but less.
- Progressive corporation tax (larger companies pay more tax). This would bias the economy towards smaller companies which generally have less problematic externalities.
It's not instant (well, sometimes it is), more of a slow but inexorable push down a hill. Some public companies are farther along the path than others, but if the company continues to exist and profit it's inevitable. For example, there are no S&P 500 companies with healthy customer relationships.
It's not impossible to run a publicly owned company in the US that isn't insanely hostile towards it's customers or employees... it's just really damn difficult because of bad legal precedent.
Dodge v. Ford is basically the source of all these headaches; the Dodge Brothers owned shares in Ford. Ford refused to pay the dividends he had to pay to the Dodge Brothers, suspecting that they'd use the dividends to start their own car company (he wasn't wrong about that part). The Dodge Brothers sued Ford, upon which Fords defense for not paying out dividends was "I'm investing it in my employees" (an obvious lie, it was very blatantly about not wanting to pay out). The judge sided with the Dodge Brothers and the legal opinion included a remark that the primary purpose of a director is to produce profit to the shareholders.
That's basically become US business doctrine ever since, being twisted into the job of the director being to maximize profits to the shareholders. It's slightly bunk doctrine as far as I know; the actual precedent would mostly translate to "the shareholders can fire the directors if they think they don't do a good job" (since it can be argued that as long as any solid justification exists, producing profit for the shareholders can be assumed[0]; Dodge v. Ford was largely Ford refusing to follow his contracts with money that Dodge knew Ford had in the bank), but nobody in the upper areas of management wants to risk facing lawsuits from shareholders arguing that they made decisions that go against shareholder supremacy[1]. And so, the threats of legal consequences morph into the worst form of corporate ghoulishness that's so pervasive across every publicly traded company in the US. It's why short-term decision making dominates long-term planning for pretty much every public company.
[0]: This is called the "business judgement rule", where courts will broadly defer the judgement on if a business is ran competently or not to the executives of that business.
[1]: Tragically, just because it's bunk legal theory, doesn't change that the potential and disastrous consequences of lawsuits in the US are a very real thing.
It is not broadly believed in corporate governance circles that there is a legal requirement to maximize shareholder value. Nor will you find court judgements that require it.
If anything Milton Friedman is more responsible for this idea that shareholder maximizing is the corporate goal. That is an efficient market argument though not a legal one and he framed it long after the dodge suit. He needed to frame that argument because so many firms were _not_ doing that.
But just because a Chicago school economist says something about governance doesn’t mean it’s broadly applicable in the same way an Austrian economists opinions about inflation aren’t iron rules about monetary policy.
> Most people with private health insurance like it.
Most people don't use it all that much, and in the common case of employer-paid premiums, the actual cost is significantly masked. As your link notes, the more care you need, the less likely you are to enjoy the experience. They dig their heels in more; sometimes egregiously so. https://www.propublica.org/article/unitedhealth-healthcare-i...
Seems like a just-so story given the numbers. Why would heavy users of health services be concentrated in the minority cohort that is dissatisfied with their insurance?
> Why would heavy users of health services be concentrated in the minority cohort that is dissatisfied with their insurance?
"Why would people who drive a lot care the most about gas prices?"
The more you use health insurance, the more chances you have to run into the kafkaesque bits. Someone who sees a GP once a year and thinks their premium is $50/month because that's the bit they have to chip in while their employer covers the rest is largely gonna go "this is fine!"
Right but there's no such selection effect for whether or not people have employer-provided coverage, and the cohort of households that do strongly approve of their current insurance coverage. I don't see how the argument you're making could hold up statistically. There are a lot of chronically ill people with employer-provided coverage; in fact, most non-senior chronically ill pts fall into that bucket.
> Right but there's no such selection effect for whether or not people have employer-provided coverage…
False. Someone with significant medical issues may well need a higher acuity plan than the employer offers. I, for example, was on the exchanges until last year, for this very reason; my employer's coverage would not have made financial sense.
> There are a lot of chronically ill people with employer-provided coverage…
The chronically ill are less likely to be employed.
Most of my South American buyers on Bricklink have me ship to one of the freight forwarders in Miami. Not complicated for me as a seller and it probably makes things simpler for the buyer too.
“ The other was the 2008 bank bailouts. The problem isn’t that the state stepped in to avert a depression. The problem is that they did it by handing the very people who caused the crash a bonus and a promotion and then proceeded to reinflate the housing bubble to lock two generations out of home ownership”
What’s interesting about this telling of it is how it reinterprets history. You are complaining about a lack of trust based on, if not an outright lie, an extremely biased narrative. The most obvious missing piece is you don’t mention the auto makers or uaw workers at all. Or that you say “reinflate the housing bubble” instead of “subsidize mortgages on houses that should have been repossessed”. We forced banks that did have proper risk controls to take tarp funds and the attached compensation limits against their will and made money on many of the assets we bought with tarp funds.
There is a trust gap, but it’s not some one way problem of coastal elites selling fables to enrich themselves and the good proletariat being duped. It’s at least as much a story of the populace not using critical reasoning skills to understand multifaceted and nuanced issues, which I suspect is not new.
> subsidize mortgages on houses that should have been repossessed
But that's also not what happened.
What we did is buy back junk assets from banks to keep the banks from going under. The only way it really "subsidized" mortgages is in that it kept banks afloat which allowed them to keep issuing mortgage loans.
People, particularly people that fell for predatory lending, still lost their homes. The people that were mostly aided by the bailout were investors who bought snake oil mortgage backed securities which had fake credit ratings applied to them.
And the reason people take a dim view on this is because it really was only people with significant assets in the first place that saw a benefit from these government interventions. A direct result of the regulation was it became a lot harder for a few years to get a home loan unless you had significant assets behind you.
That's not to say some percentage of these interventions didn't help everyone. It's always messy. But it is saying that a lot of people would have been in a much better situation had the government, instead of bailing out the banks and investors, taken that same money and given it directly to the citizenry. Even the banks and investors would have ultimately been in a better position as people would have ultimately taken that money and spent it on things like their mortgages which they fell behind on.
There were specific mortgage subsidies as part of tarp. For instance the making homes affordable and hardest hit fund programs.
But beyond that if the mortgages had been sold at market prices many of them would have been snapped up by companies that aggressively went after the secured properties. That’s the _natural_ outcome of letting the market action happen. More people would have been put out of their homes.
I’m fairly ambivalent on tarp. I think letting actors take risky actions and get bailed out creates a moral hazard. But that applies to mortgage holders who were over extended and auto workers who get bailed out ahead of other stake holders too. I can see a strong argument that we should have biased that way, but to say we didn’t help “regular” people is just false narrative.
It's a biased narrative, but perception of truth is equivalent to truth when it comes to trust, and multiple factors make this narrative compelling than more nuanced ones
That’s about 240 GBP at the current spread between what wise is offering and the interbank rate (which is higher than wise is getting on their uk safeguarded accounts).
So you probably blew through that the minute your account triggered elevated kyc.
Courts, law enforcement and contract law. All of which will take a dim view of using a currency which appears designed wholly to make their function harder.
That's the wrong answer. The existence of tokens predates the existence of government. It's the next step after barter. The correct answer is reputation. A vendor who cheats his customers builds up a bad reputation, and the only way he can keep doing it is by changing customer bases, for example by moving to a different town. Think of the traveling snake oil salesman who moves on once people realize his remedies don't work.
The courts etc are there so we don’t have to create a posse and ride out to find the snake oil salesman. You _can_ have commerce without them but it’s much higher friction. So if a crypto wants me to abandon the existing systems it needs to show it creates less friction.
Crypto’s use case isn’t for the layman. It’s for countries that aren’t aligned with America to have a separate currency system. Doubly useful for bypassing sanctions.
There's no need for a mob, government-backed or not. A vendor who scams his customer base is harvesting its good will, and eventually it will run out and he'll no longer able to do business.
A crypto advocate would argue smart contracts can fulfil that role, but also that applies in developed countries but not in the countries where the vast majority of the world population lives.
I think if a true crypto economy does emerge anywhere it's likely to be Nigeria, Lebanon etc - places with a significant population of educated entrepreneurial people but where the state is run abysmally and you can't rely on those institutions anyway
It seems to me that the crypto absolutists have it backwards. You can’t solve the problem of failed states by changing the technology of currency, because the state is there to solve for the counterparty risk at the point of exchange.
The alternative to governments monopoly on violence for enforcement, no matter if you exchange in monero or giant stone discs, is broad use of vigilante violence.
So while crypto seems like an interesting technology for moving money around, it seems like it doesn’t solve for the point of exchange problem and thus crypto that focuses on making that difficult for government mediation are bound to be only useful for illegal activities.
There is no counterparty risk for the seller in the traditional sense with bitcoin or monero, they're bearer assets, once the transaction is confirmed in your wallet there's no risk for you. You don't need to use violence to make sure you get paid?
What you actually have is the opposite problem (in a sense) - the transaction is irreversible, the seller will receive payment and keep it even if they shouldn't (i.e fraud). So there is more risk for the buyer than in a fiat system where transactions can be reversed by legal processes
That’s all counterparty risk. If you deliver the payment before the service/good the buyer takes on the risk. The opposite is true if the payment or good is delivered first.
You can dial the risk in either direction with any payment scheme (20% down balance due on delivery etc) but you can’t eliminate it.
Right yep, I understand what you mean. Yes, ultimately you need some kind of dispute mechanism that probably requires actual human intervention.
A good example is how disputes work on P2P crypto exchanges like bisq - you have a crypto contract of some kind that holds funds in escrow, but ultimately disputes are resolved by a team of actual humans who look at the facts and make a decision, not everything can be "code is law"
Courts and law enforcement certainly provide these things, but they are not required. The inherent design of blockchains makes them trustworthy (an oversimplified statement), which is even better.
Blockchains don’t, and can’t, solve for the risk of the off chain component of an exchange.
The transactions aren’t atomic so someone is taking on counterparty risk. One of governments prime responsibilities is dealing with that risk, no matter the currency in question.
But other than that what I’ve found to be the most important is static tooling. Do you have rules that require tests to be run, do you have linters and code formatters that enforce your standards? Are you using well known tools (build tools, dependency management tools etc) or is that bespoke.
But the less sexy answer is that no, you can’t drop an agent cold into a big codebase and expect it to perform miracles. You need to build out agentic flows as a process that you iterate and improve on. If you prompt an agent and it gets it wrong, evaluate why and build out the tools so next time it won’t get it wrong. You slowly level up the capabilities of the tool by improving it over time.
I can’t emphasize enough the difference in agents though. I’ve been doing a lot of ab tests with copilot against other agents and it’s wild how bad it is, even backed with the same models.
reply