I reckon Sony get a few more years of even more profitable rent seeking before the EU regulates them like Apple’s App Store and forces a game purchase to be valid on all platforms it’s playable on.
I think he was pretty brave for standing up against what is generally perceived as an injustice being done by one of the biggest companies in the world, just a few years out of college. I’m not sure how many people in his position would do the same.
I’m sorry for his family. He was clearly a talented engineer. On his LinkedIn he has some competitive programming prizes which are impressive too. He probably had a HN account.
Before others post about the definition of whistleblower or talk about assassination theories just pause to consider whether, if in his position, you would that want that to be written about you or a friend.
I found the same - market is actually getting worse (if HN is representative of the market). This month there were more job seekers on HN than jobs for the first time since 2014.
Thank you! The cloud servers are sufficiently cheap for us that we could afford the extra flexibility we get from them. Hetzner can move around VMs without us noticing but in contrast they are rebooting a number of metal machines for maintenance now and for the last little while, which would have been disruptive especially during the migration. I might have another look next year at metal but I’m happy with the cloud VMs currently.
I didn’t touch on that in the article, but essentially it’s a one line change to add a worker node (or nodes) to the cluster, then it’s automatically enrolled.
We don’t have such bursty requirements fortunately so I have not needed to automate this.
I read about 30 LLM papers a couple months ago dated from 2018-2024. Mostly folks are publishing on the “how do we prompt better” problem, and you can kind of get the gist in about a day by reading a few blogs (RAG, fine tuning, tool use, etc). There is also more progress being made for model capabilities, like multi modality, and each company seems to be pushing in only slightly different directions, but essentially they are still black boxes.
It depends what you are looking for honestly “the latest things happening” is pretty vague. I’d say the place to look is probably just the blogs of OpenAI/Anthropic/Genini, since they are the only teams with inside information and novel findings to report. Everyone else is just using the tools we are given.
I stopped using GitHub copilot. I really didn’t like my train of thought being interrupted.
I have heard more comprehensive “leave the thinking to us” tools like cursor give better results.
Personally I think if your tools or projects are so dull that you require an AI to use them, perhaps you need sharper tools or more interesting projects.
To be fair to them, they have done a significant amount of work to design the network to be open to competing servers, and I think it would be quite tricky to unpick that. In comparison to successful networks like TikTok, Twitter, Facebook, LinkedIn, ATP gives a far fairer playing field and Bluesky hasn't done anything (aside from taking funding) to suggest they're not going to run with it.
You are right that they could change their architecture so that their Relay is more trusted or blocks others in some way, once they capture the market. I am not sure what guarantee they could give with the current ATP arch - perhaps a committee would help, but other social networks have no incentive to support ATP.
Depends why you are doing outbound for a low ARPA product.
What problem is cold email solving, and will it eventually solve it? If the product is good enough for the initial customers, I’d switch to marketing and inbound sales.
One reason to stick with cold email is if the product is not good enough yet, so you need to chase people to talk to you and get them to help you improve it
You can still do cold email don’t get me wrong. But generally for a low ARPA service (<$2k) you want the product to be simple, self-serve, and for customer acquisition to be mostly inbound (ergo from marketing efforts, not outbound emails). Low price points require low CAC, and outbound sales is generally not.
From Apollo, which sells an outbound sales tool: “As a benchmark, try to achieve at least a 20%-30% win rate by closing 2-5 deals per month per rep”. If you have a full time sales rep chasing 5 $100/year deals per month they will close $6k/year. At $1k ARPA they might close $60k/year - after paying salaries you’re not making a profit.
Obviously this doesn’t apply in the early stages, but after you have N customers. Chasing outbound sales for a low price point product is just a recipe for low/slow growth.
Personal network to get the first 20 - literally going to meet people and helping them set up over coffee, then marketing for the rest. The audience were SaaS founders mostly, since it's a marketing tool for SaaS SMEs.
What's important is after you get the first N customers, the way you acquire customers will likely change. You'll exhaust your personal network eventually. Knowing your Ideal Customer Profile (ICP) is important and so is having a good understanding that the value/price of your product dictates what kind of sales/marketing you can do.
Things like cold emailing are fine for your first few customer development calls, but generally if your deal size is less than $1-2,000/year then (with exceptions) you should stop doing this after you get the first customers.
I run a low price point SaaS for SMEs (avg price is <$100/month), so this required switching from just messaging people I knew, to getting SEO, word of mouth, and the viral loop working. Outbound/inbound sales is not economical at that price point.
This would be a tremendous amount of steel for the size of building they are creating.
Crude steel production in 2023: 18.88 billion mt. 20% = 3,776,000,000 metric tons.
If you just laid that on the ground in 1 meter square blocks it would cover a 481km² area (steel mass=7850 kg/m³). If you made a 1m² block tower, it would be enough to reach the moon, wrap around it, and start heading back.
Their original planned area is 26,500 square km, though this has been scaled back to 1.5km long.
I struggle to see how 481,019 m³ of steel will fit in an area 1.5km square without being the largest and tallest building ever constructed. If you just made a block of pure steel in that 1.5km square area it's going to be 320m high (about half the height of the world trade center across an area the size of 40 Manhatten blocks).
Yes I hope they’ll find a solution and get re-elected. In some cases changes to frameworks require a 1 year consultation (planning permission to change planning permission rules!) so I am sure it will take years to see an increase. My point was their changes announced so far seemed bitty and wouldn’t address much of the issue.
The problem is hard! Even if you do build 500k homes a year (2x current) in places people don’t want you to build, it will still take 15 years to catch up, and they might be voted out before then because voters don’t want it.
This is pretty bleak reading for a resident of the UK! It's a good explanation for why there's stagnation (generally not allowed to build here). I'm a fan of their work (see the housing theory of everything [1] which is also good).
I'd be interested to read what they think can be done about the planning issue. The new government hasn't really come through on their promise to address it. They ran out of low hanging fruit pretty quickly. They're focusing more on rental reform rather than on supply. Gov modified the NPPF in odd ways (e.g. reduced targets in London, where need is highest). They set up a panel to look at new towns which will report back in a year.
This bit at the end made me laugh:
> it need simply remove the barriers that stop the private sector from doing what it already wants to do
Unfortunately these supply-side policies causing stagnation are representative of what our ageing population actually wants. The average 50+ voter thinks demand is too high and should be cut until supply catches up (in 33 years) [2][3].
This is great to hear, though I'm curious why photos on your phone / pinterest would be relevant to a recommendation system? Surely the biggest signal would be what Spotify already uses: the features of various relevant factors (your previous listening sessions, your current session, what other similar sessions look like, etc.), that said, their recommendation system is surprisingly terrible given how much easier music recommendations must be versus video, yet YouTube seems to have had this nailed for 15+ years whereas Spotify's "Discover Weekly" is so bad.
> I come up with an idea of generating playlists from images. Images that you shot on your phone yourself, or that you found on Pinterest, or a painting that you really like and feel inspired by.
This is genuinely interesting! Do you send the images to an LLM with a prompt like "generate a list of songs that would go well with this"?