HN user

HuwFulcher

276 karma

AI Engineering Lead

You can find me at: https://huwfulcher.com

[ my public key: https://keybase.io/huwfulcher; my proof: https://keybase.io/huwfulcher/sigs/oxU3GC_S7uKggXi5WhVmGcTi9JpDaUoJmw8zC-APRkM ]

Posts17
Comments88
View on HN
The 512KB Club 9 months ago

While a fun idea, arbitrary limits like this just aren’t necessary. Yes it’s all well and good in the name of reducing trackers, etc but what if I want to have an image heavy site? Why does that get perceived as a bad thing?

Nostr 10 months ago

There is the concept of Data Vending Machines (DVMs) and curated follow lists now.

There’s also work on a Web of Trust in some clients that filters notes from people that don’t meet the WOT score. It’s essentially a weighted score based on who you follow and who they follow

This is a challenge which I don't think AI tools like Cursor have cracked yet. They're great for laying "fresh pavement" but it's akin to being a project manager contracting the work out.

Even if I use Cursor (or some other equivalent) and review the code I find my mental model of the system is much more lacking. It actually had a net negative on my productivity as it gave me anxiety at going back to the codebase.

If an AI tool could help a user interactively learn the mental model I think that would be a great step in the right direction.

I like the idea of there being interest specific mastodon servers. It makes content curation easier but the locking of accounts to a particular server is problematic.

Protocols like Nostr offer a good alternative to this. At the moment relays are just used as nodes in the decentralised network but there is work on how to view feeds from specific relays which could make the idea of content specific relays more likely

As the other response said. It’s DevOps for ML. They have Amazon SageMaker which is the managed ML/MLOps offering that we use extensively because we’re a small team. The documentation is awful

AWS specifically have really dropped the ball on this.

I interact regularly with AWS to support our needs in MLOps and to some extent GenAI. 3 of the experts we talked to have all left for competitors in the last year.

re:Invent London this year presented nothing new of note on the GenAI front. The year before was full of promise on Bedrock.

Outside of AWS, I still can’t fathom how they haven’t integrated an AI assistant into Alexa yet either

I've been reflecting recently on our patterns of content consumption online. Curious to hear people's perspectives on the matter.

Yes a regular check-up wouldn't be the silver bullet but it would help expose larger problems under the surface which could be prevented with prescribed exercise, diet, etc.

Ultimately it's down to the patient to put in the work but if the NHS focused more on creating community practices where these things could be more widely available then maybe we should see more meaningful change.

This is a big problem in the UK I think. The idea of having a regular checkup with a doctor is unheard of and will get you some weird looks if you ask for it.

The NHS doesn't seem (as an outsider looking in) to do well with preventative care. I firmly believe (without necessarily a lot of evidence) that if we focused more on healthy lifestyles and made them more cost effective for people the amount of money we plough into the NHS every year could be reduced drastically.

They make $300/mo from donations from those that self-host. The point they are making is that enterprise level users are self-hosting and not putting back into the running of Plausible.

I still think the way they've done this stinks though.

Being cynical it looks like a way of holding features hostage for the purpose of extracting revenue.

I sympathise with the problem of re-sellers abusing their hard work but they are also creating two different classes of product with this change.

I’m not concerned about myself being banned on Masotodon as I keep away from political subjects. I don’t find the platform useful as the conversations that I’ve been able to find are highly politicised. I guess I need to work better at tuning out the noise on there.

On X I mainly occupy the indie hacker space which is focused on that topic. There are definitely more political conversations that come up but it’s much easier to filter those out.

I'm not looking for value through political debate and heterodoxy. In fact I'm trying to look for the opposite, I'm glad you seem to have found it! Algorithmic content is a double edged sword. I've found it much easier to find people on X to discuss the non-political content I want. On the other hand it's much harder on Mastodon.

I find that Mastodon has become the mirror-verse version of Twitter/X. I've tried spending time on there but unless if you subscribe to certain political/ideological values then you wont find much to hang around for.

On the flip side I find the discourse on X a lot more varied. There are definitely big problems with the types of conversations that happen on X but I have to go out of my way to find them. On Mastodon it's not that hard, mainly down to it being a smaller pool of content.

ActivityPub/Mastodon is flawed at its core with its concept of federation. They paint a pretty utopian picture of decentralised communities but there really isn't much difference between them. They are mostly part of the same political zeitgeist and if you don't toe the line you get banned. Once you're banned that's it, you've lost your profile, sure you can recreate elsewhere but have to start from scratch.

For the problems alternatives have like Nostr at least they are truly censorship resistant.

It's what you've stated but also there's an ethical side to it. Many founders want to give the option for competent users to self-host, only way to do that is open up the source code. They could go completely down the open source route and rely on donations but that doesn't usually cut it. The business model then is to provide a managed service where the founders deal with the hosting, updating, security, etc.

Based on my experience of consulting with public sector, most projects “fail” in one way or another.

Gov departments are notorious for not giving correct requirements, changing requirements or swamping the project in bureaucracy.

If we said firms couldn’t rebid after a failed project (without the cool off) I suspect there wouldn’t be many firms left.

It shouldn’t be this way but unfortunately it is.

Thanks will take a look! At the risk of sounding ignorant, I can’t understand why such seemingly simple peripherals are so expensive