Steven Hutton

2.4K posts

Steven Hutton banner
Steven Hutton

Steven Hutton

@StevenBHutton

Lead Tools Programmer, talk mainly about programming, games design and software development. Tweets my own, naturally.

Newcastle Upon Tyne, England Katılım Mayıs 2010
62 Takip Edilen338 Takipçiler
Sabitlenmiş Tweet
Steven Hutton
Steven Hutton@StevenBHutton·
I wanna talk a bit more about @cmuratori's excellent terminal rendering demo. youtube.com/watch?v=hxM8Qm… There isn't too much too it, Muratori shows that terminal rendering can be 10-100x faster than windows terminal for outputting large files with a simple solution.
YouTube video
YouTube
English
5
49
224
0
Steven Hutton retweetledi
Jeremy Corbyn
Jeremy Corbyn@jeremycorbyn·
78 years ago, 800,000 Palestinians were expelled from their homes. Britain made it possible. Today, Israel is committing genocide in Gaza. Britain made it possible. 78 years of the Nakba - and 78 years of British complicity in crimes against the Palestinian people.
English
4.2K
17.5K
52.1K
953.8K
Steven Hutton
Steven Hutton@StevenBHutton·
Fox McCloud in : The Raid on The Uncanny Valley.
English
0
0
0
55
Steven Hutton retweetledi
Cairo Smith
Cairo Smith@cairoasmith·
There's a common misconception that Brutalist buildings were unpainted, but thanks to microscopic analysis of the exteriors we can now recreate what they looked like in their prime.
Cairo Smith tweet media
English
443
3.4K
39.5K
7.4M
Steven Hutton
Steven Hutton@StevenBHutton·
@rfleury We crossed some kind of Complexity Threshold in the 2010s where this stuff just became much harder to manage. I don't think having a business-model based on nostalgia is a good long-term plan but I can't blame them for doing this, it's pretty sensible considering their options.
English
0
0
0
42
Steven Hutton
Steven Hutton@StevenBHutton·
@rfleury The typical game project takes like 7 years now and costs like $50 -$100 million dollars (before marketing). A big ass-content rich game like an elder scrolls probably more. New games are very high risk, beloved old games are low risk and help offset the risk of new stuff.
English
1
0
2
174
Steven Hutton retweetledi
Ryan Fleury
Ryan Fleury@rfleury·
Best anti-AI case I’ve possibly seen so far
JNS@_devJNS

English
84
1.3K
42.9K
1.3M
Steven Hutton
Steven Hutton@StevenBHutton·
@TheOGZaxthan @aabbereos40875 @RiotAugust She gets resistences for killing minions and monsters but uh.... doesn't everyone get that? From XP? So her passive is "Some of your normal level up progression is moved to your passive."... cool, what an exciting ability.
English
1
0
1
93
Zax
Zax@TheOGZaxthan·
@aabbereos40875 @RiotAugust The passive is so good actually idk why we complain. Her whole kit has dmg and barely any sustain. It is fine for her to get resistances
English
4
0
0
705
August
August@RiotAugust·
Shyvana’s rework came out strong (57% winrate). We expect her to only get stronger as players optimize so we put together a micro patch nerf. It should be live now.
August tweet media
English
245
74
1.6K
615.9K
Steven Hutton
Steven Hutton@StevenBHutton·
@cmuratori @DemetriSpanos It seems like the previous limiting factor was model size (compute) and amount of data. When that was broken we saw a massive, rapid improvement (as you'd expect). What factor, if removed, would lead to a similar massive 10-100x improvement in the quality of model outputs?
English
1
0
1
62
Steven Hutton
Steven Hutton@StevenBHutton·
@cmuratori @DemetriSpanos Is it possible / acceptable to submit questions for upcoming episodes? I'd like to ask Demetri : "What is the current limiting factor on the improvement of model performance? And how difficult is that factor to address?"
English
2
0
1
125
Casey Muratori
Casey Muratori@cmuratori·
Wading Through AI episode 2 is now up! In it, @DemetriSpanos and I discuss the recent Anthropic announcement about building a C compiler with "parallel Claudes", and how to properly assess AI announcements in general.
Casey Muratori tweet media
English
10
18
242
52.8K
Steven Hutton
Steven Hutton@StevenBHutton·
If I'm right that means no AGI anytime soon. It means models will continue to improve but at some rate less than moore's law until we identify and break the new constraint.
English
0
0
1
37
Steven Hutton
Steven Hutton@StevenBHutton·
So my totally uninformed theory is that model performance improvement has slowed and the limiting factor is something that's not easy to remove. Either it's WAY more hardware, or WAY more data or a new breakthrough in understanding of AI or some other factor.
English
1
0
0
39
Steven Hutton
Steven Hutton@StevenBHutton·
I have an AI theory. It's totally uninformed. LLMs were bottlenecked on scale. (amount of data, amount of compute). We broke that bottleneck a few years ago and we got what you get when you break a constraint. HUGE IMPROVEMENT REALLY FAST but...
English
1
0
0
88
Steven Hutton
Steven Hutton@StevenBHutton·
My only career advice that I will stand on. If a job advertisement lists "free food" as a perk, that means you won't be going home for dinner.
English
0
0
1
54
Steven Hutton retweetledi
Dexerto
Dexerto@Dexerto·
Game dev Cakez77 reacts to how much money his game made 30 hours after release following four years of development
English
274
1.3K
33.4K
1.9M
Steven Hutton
Steven Hutton@StevenBHutton·
Doesn't make me think Claude is impressive. Does make me think "marketing" is a scam.
Ole Lehmann@itsolelehmann

i can't believe nobody caught this. Anthropic's entire growth marketing team was just ONE PERSON (for 10 months, confirmed) a single non-technical person ran paid search, paid social, app stores, email marketing, and SEO for the $380B company behind claude here's exactly how one human is doing the job of a full marketing team: it starts with a CSV. 1. he exports all his existing ads from his ad platforms along with their performance metrics (click-through rates, conversions, spend, etc) 2. feeds the whole file into claude code 3. and tells it to find what's underperforming. claude analyzes the data, flags the weak ads, and generates new copy variations on the spot this is where he gets clever: he then splits the work into 2 specialized sub-agents: 1. one that only writes headlines (capped at 30 characters) 2. and one that only writes descriptions (capped at 90 characters). each agent is tuned to its specific constraint so the quality is way higher than cramming both into a single prompt so now he's got hundreds of fresh headlines and descriptions. but that's just the text. he still needs the actual visual ad creative, the images and banners that go on facebook, google, etc. so he built a figma plugin that: 1. takes all those new headlines and descriptions 2. finds the ad templates in his figma files 3. and automatically swaps the copy into each one. up to 100 ready-to-publish ad variations generated at half a second per batch. what used to take hours of duplicating frames and copy-pasting text by hand so now the ads are live. the next question is which ones are actually working. for that he built an MCP server (basically a custom integration that lets claude talk directly to external tools) connected to the meta ads API. so he can ask claude things like: • "which ads had the best conversion rate this week" • or "where am i wasting spend" and get real answers from live campaign data without ever opening the meta ads dashboard and the part that ties it all together and closes the loop: he set up a memory system that logs every hypothesis and experiment result across ad iterations. so when he goes back to step one and generates the next batch of variations... claude automatically pulls in what worked and what didn't from all previous rounds. the system literally gets smarter every cycle. that kind of systematic experimentation across hundreds of ads would normally need a dedicated analytics person just to track the numbers from the doc: ad creation went from 2 hours to 15 minutes. 10x more creative output. and he's now testing more variations across more channels than most full marketing teams a $380 billion company. and their entire growth marketing operation (not GTM) = just one person and claude code lol truly unbelievable

English
0
0
0
63
Steven Hutton
Steven Hutton@StevenBHutton·
@ThePrimeagen This is for people who have a meeting that's important enough to prepare for ahead of time but not so important that a few hallucinations will matter?
English
0
0
0
22