Very curious how we'll look back on Google spending 2 billion dollars to "license IP" and hire a handful of people.
If there's 47m software engineers in the world, at $200/month, and 50% gross profit that's a $56 billion TAM. Not crazy to think it's more if we include the adjacent space of analyst roles that write software (sql, advanced excel, etc).
They'll have to crush it to make a $2 billion acquihire look reasonable, but it's possible.
It’s more about not getting embarrassed ever again. The breakthrough papers for LLMs came from Google, but not the product. This was embarrassing. Even if you have to spend double on everything from now on you do it, because Google effectively got sucker punched by ChatGPT.
that's not the market they're thinking of, they're thinking of the total amount of money spent on developers per year, globally, and capturing a percentage of that
They are doing what they've always done skirting the edges of antitrust laws by buying up competitors before they become competitors so that they can shut them down.
Maybe it's just me, but the words, "Other Immigration Violator" rubbed me the wrong way. I see it's a term from the source data and ICE describes the category as, "Other immigration violators are individuals without any known criminal convictions or pending charges in ICE's system of record at the time of the enforcement action."
ICE alleges these people have violated the civil code so calling them "violators" assumes guilt and comes across as inflammatory. Something like like "No Criminal Status" would be accurate and more neutral.
Personally, I'd call them "Productive Members of Society The Rest of Us Depend On."
Great feedback, will work on improving the language for these categories. I agree that ICE has chosen pretty inflammatory names for these otherwise presumptively innocent detainees
You don't find it ironic at all to call the names ICE has chosen 'inflammatory', while you're here brainstorming with multiple people trying to come up with the perfect terminology, phrases, graphs to include vs not include, even the perfect colors to use in order to best impart your political ideology onto the reader?
I don’t. The data is published with a strong political bias which is morally antithetical to our legal system. The point of publishing in this way is to shed light on the human cost rather than the dehumanized political speech currently embedded within it.
Like it or not, this data is highly political. You can’t correctly interpret it in a vacuum.
Enjoyed the history, but don't get the premise. Has any tech been watched more closely or adopted faster by incumbents?
> The first cars were expensive, unreliable, and slow
We can say the same about the AI features being added to every SaaS product right now. Productization will take a while, but people will figure out where LLMs add value soon enough.
For the most part, winning startups look like new categories rather than those beating an incumbent. Very different than SaaS winners.
"Despite Democrats holding thin majorities in both chambers during a period of intense political polarization, the 117th Congress (2021-2023) oversaw the passage of numerous significant bills, including the Inflation Reduction Act, American Rescue Plan Act, Infrastructure Investment and Jobs Act, Postal Service Reform Act, Bipartisan Safer Communities Act, CHIPS and Science Act, Honoring Our PACT Act, Electoral Count Reform and Presidential Transition Improvement Act, and Respect for Marriage Act."
All of these except the first two were bipartisan and got 60 Senate votes (or more)
It does seem like things are trending toward less public laws passing over the last decade, as well as record low time in session and other congressional activity.
Threads being over is a good thing, isn't it? Truth's been discovered, all parties agree, no more time spent on going in circles, can move on to do other, meaningful things, etc. Unless you are facebook, and you optimize on endless churn, stealing time and showing ads.
I haven't seen the original comment, but the wiki article is moronic. None of the listed example seems even bad to me, claiming that they are the devil is ridiculous. Maybe even a false flag.
The only one that actually has anything to do with "terminating cliche" is "Let's agree to disagree.". But that's just the common phrase you say after you've decided to opt out of an argument. It is not (and can't be) the cause of it, it is the consequence of it.* And it is by no means any bad, or should one avoid it.
* : something something people being able to easily leave an argument makes them do it more. But it would need a lot of stretch to argue that the possibility to go away from arguments is a net negative for humanity
edit: can we agree that the random shit you linked is 100% unrelated to the argument at hand, therefore/and definitely should not be used?
edit2: yeah, it assumes the truthness of some ridiculously nonsensical concepts, and uses them in a meta meta way, that is 2-3 steps away from the topic at hand. Much-much more annoying than anything listed. "This is the hill you want to die on, huh? Naah.. How about.." *points downwards* "..there is this hill there 14000 miles away (actually there is only ocean), how about we move this fight there?" Yeah no thx.
A lot of what happens in Congress is obvious to do and everyone agrees. While the media certainly focuses on the handful of things the two parties are at odds over, most of the lawmaking done by Congress is not controversial between parties, and is simply passed, so we don't hear about it.
What does that matter? We're talking trifectas here, not supermajorities. The filibuster is a cute remnant of "decorum." It's a vestigial rule which will disappear when too inconvenient. (Fun question with not-so-fun answers: why isn't the filibuster gone already?)
I don't think this tells us much. The present distribution of supporters is rather unique in how strongly it correlates with population density, which means that Dems are going to have a major structural handicap in Senate. I don't think there was ever anything similar historically. And it won't change unless and until the coalitions change, which, sure, will happen eventually - but then it'll be a completely different party under the same brand, so why would Republicans today care about that new party's difficulties?
in American politics, 18 months definitely counts as an "anytime soon".
In the 2026 senate election, the Dems could absolutely flip Maine, North Carolina, and two others; maybe Alaska and Ohio.
If Elon Musk makes good on his threat to try to take out sitting GOP senators, that splitting of the vote could mean the Dems pick up a few more as well.
Because I don't think it's vestigial, I think it's serving an important function of governance that never made it into the official
rules but is nonetheless necessary as a stabilizing effect. It doesn't have to be the filibuster but something ought to provide the effect. It should be easier to block legislation than to pass it. It wouldn't be a good thing if you could have huge policy swings when a 51-49 becomes 49-51. Being able to, with effort, demand specific pieces of legislation reach a higher bar biases us toward the status quo.
The answer is to vote out politicians. Getting ranked choice voting on your states ballot would go a long way to fixing this. They would not have Mamdani on the ballot for NY mayor if it wasn't for ranked choice voting. Certain politicans know this and have made RCV illegal in their state. Get RCV on the ballot for your state.
RCV / Ranked Pairs of course. The IRV decision process is still a relic of the two party system, with the possibility for some pretty terrible strategic-voting dynamics as votes diverge from just two major parties.
> Certain politicans know this and have made RCV illegal in their state.
That would be Republicans.
While Democrats have pushed across multiple states for changing voting mechanisms, Republicans in eleven states have pre-emptively banned any and all use of RCV at any level within the state.
If you're doing a new thing anyway then it makes no sense to do something worse instead of something better. Popularity is determined by people; make the better thing the popular one.
It absolutely makes sense. You need buy in from the public. RCV is the most known alternative and it has taken a decade to get it that far. If you want to start the work of informing people about STAR voting then be my guess but RCV is a tremendous improvement from what we have and an acceptable alternative.
Personally I think “approval voting” is almost as good as RCV but orders of magnitude easier to sell to the public.
There’s just a checkbox next to each candidate and you check the box next to any candidate you’re “okay” with. Results in the most “okay-est” candidates getting elected so when the winner is announced everyone goes “…okay.”
Also could make primaries less important, because multiple candidates from a party could theoretically run for the general election without splitting votes.
Communication is easier because in RCV the candidate who gets the most #1 votes doesn’t necessarily win which could lead to a loss of confidence in the system. Its very easy to tell the American public “this guy got the most checkmarks” and no one gets confused.
If I recall the problem with approval voting is that it is much easier to tamper with than RCV. Filling in an empty bubble is a lot easier than changing the order of ranking on a ballot
That’s a good point. Seems like that could be a problem for current ballots too - add a second checkmark to invalidate ballots voting for the “other” guy. Doesn’t seem to be a widespread issue, but detecting it for current ballots would be more obvious.
Maybe that breaks this idea. Maybe ideally you’d maybe want a touchscreen+printer to fill in the bubbles with printer ink and show it to the voter for them to double-check before putting in the stack (or, if wrong bubble filled, put it in rejected stack).
Would love more feedback from people to get a better sense of all pros and cons.
The exact mechanism won’t be standardized, some places fill in ballots with pen today, let the voter feed it into the machine and optically scans to tabulate, and others use a computer that tabulates and (usually?) spools a paper record that’s behind a window so the voter can see that the paper record is accurate. The important part is the actual methodology itself.
Most people don't actually know anything about any of this. If they've heard of RCV at all their understanding of it is at the level of "it's something different than the status quo and supposedly better". You could swap in STAR and they mostly wouldn't even notice that you've changed anything. But you'd notice the difference in the election outcomes, in a good way.
Enough people know about it that it has been put on ballots in several states and has had strong pushes in other states while STAR hasn't at all. If you want to get outside and start informing people about STAR then please do but RCV has a decade long head start and is the path of least resistance.
tl;dr IRV is extremely poor, doesn't actually solve the vote splitting problem, and is radically more complex and cumbersome than superior alternatives like approval voting, which would plausibly scale far faster once gotten off the ground. plus approval voting was adopted by 2/3 majorities in fargo and st louis, so we know it's politically viable.
Does it? Claude Code is the product that works the least well for me, mainly because of its tendency to go off and do tons of stuff. I've found LLMs are at their best when they produce few enough lines of code that I can review and iterate, not when they go off and invent the world.
For that reason, I mainly use Aider and Cursor (the latter mostly in the "give me five lines" comment mode).
I tell Claude Code to make a detailed implementation plan, then come back to me for discussion, then we step through the plan bit by bit together. It's all about guidelines. Don't want it to charge ahead? Tell it not to.
I had a very specific problem I wanted to solve, Opus kept being entirely sure that it knew what the problem was, and that it could fix it. Next iteration, when the previous solution didn't work, Claude was sure it was some other thing. So on and so forth, until I was $35 poorer and had a bunch of code that didn't really seem to do anything, and certainly not fix my problem.
> promises that the contributed code is not the product of an LLM but rather original and understood completely.
> require them to be majority hand written.
We should specify the outcome not the process. Expecting the contributor to understand the patch is a good idea.
> Juniors may be encouraged/required to elide LLM-assisted tooling for a period of time during their onboarding.
This is a terrible idea. Onboarding is a lot of random environment setup hitches that LLMs are often really good at. It's also getting up to speed on code and docs and I've got some great text search/summarizing tools to share.
> Onboarding is a lot of random environment setup hitches
Learning how to navigate these hitches is a really important process
If we streamline every bit of difficulty or complexity out of our lives, it seems trivially obvious that we will soon have no idea what to do when we encounter difficulty or complexity. Is that just me thinking that?
> > Onboarding is a lot of random environment setup hitches
>
> Learning how to navigate these hitches is a really important process
To add to this, a barrier to contribution can reduce low quality/spam contributions. The downside is that a barrier to contribution that's too high reduces all contributions.
There will always be people who know how to handle the complexity we're trying to automate away. If I can't figure out some arcane tax law when filling out my taxes, I ask my accountant, as it's literally their job to know these things.
> There will always be people who know how to handle the complexity we're trying to automate away
This is not a given!
If we automated all accounting, why would anyone still take the time to learn to become an accountant?
Yes, there are sometimes people who are just invested in learning traditional stuff for the sake of it, but is that really what we want to rely on as the fallback when AI fails?
It's highly unlikely that everybody will flock to LLMs, leaving absolutely nobody capable of stringing together a few lines of code on their own. Some devs may enjoy vibe coding, and may even be more productive that way, but there will always be use cases where it is preferable to produced deterministic code via a human dev.
Apple and Microsoft will do the same. The OS war became the browser war becomes the AI assistant war. I suspect there's enough complexity in AI use cases that there will be independent winners for things like coding assist, enterprise agents, etc.
