arios

Software architect. I rebuild old domains and test what actually ranks. Making AI video too. I post real numbers, including the bad ones.

[arios @ariosleveling]

Spent today auditing two sites I rebuilt on aged domains. Same niche, same targets. Both ranking nowhere.

Then found a competitor sitting on page 1 with an objectively worse setup than mine.

Everything I had backwards 🧵

[arios @ariosleveling]

The useful part here is breaking the AI boom into separate, testable questions.

My takeaway for market research: match the event, deadline and resolution source before comparing a forecast with a price. Being bullish on AI isn't a probability estimate.

[Forecasting Research Institute @Research_FRI]

Will the AI boom continue, or is the industry headed toward a bust?

In the latest wave of the Longitudinal Expert AI Panel (LEAP), we asked AI experts and superforecasters to forecast the following:

• The value of semiconductor and software stock indices
• Data center investment in the US
• Investment in IT equipment, and electrical and communication structures
• The combined annualized revenue run rate of OpenAI and Anthropic

🧵 Read on for our key findings:

[arios @ariosleveling]

I'm researching a read-only Polymarket reconciliation report, not a signal service.

Have you recently struggled to explain the difference between your cash balance and displayed P&L? What do you use now, and what remains unclear?

If useful, the proposed pilot is a $5 one-off breakdown of cash flows, trading results, and repeated buys.

No wallet connection, keys, or trades. It's not launched yet; would you be willing to discuss a recent example privately?

[arios @ariosleveling]

Caching is important

[shmidt @shmidtqq]

https://t.co/3NOv6JypnZ

[OpenAI @OpenAI]

We’re ending our partnership with Cursor following its acquisition by SpaceX. Under our proposal, Cursor’s direct access to our models would end on November 12.

We know that the people most affected by this decision are the developers who rely on OpenAI models in Cursor. We care about their experience in this transition and we’re ready to go above and beyond to support them.

https://t.co/OzuCTzUjfX

[arios @ariosleveling]

@X_supporttop1 hello

[arios @ariosleveling]

A noindex tag can be completely correct and Google will still ignore it.

If that same URL is disallowed in robots.txt, Googlebot never fetches the page. So it never sees the tag telling it to leave.

The page stays indexed. Every check says the tag is there. It is. Nobody is reading it.

Watched this play out on someone else's site this week. Tag present, page indexed, answer sitting in a different file.

The 60-second check:

curl -s <your-domain>/robots.txt | grep -i disallow

Anything on that list carrying a noindex is stuck. Allow the crawl so the tag can be read, or send it in the header instead:

X-Robots-Tag: noindex

[arios @ariosleveling]

Grok Imagine wrote zero valid HTML across eight attempts.

One change to the prompt took it to twenty out of twenty.

The failures: inital-scale, </MI>, <title>Documents/title>, and one that just read onsodng,<(<HHTL;

What changed: instead of asking for a code block, I named one quoted string — twitter:player — and said it was the only text allowed in the frame. Large monospace. Everything else described as shapes rather than content: a browser window with a smaller browser window inside it.

It fails on word count, not difficulty. Concept in the picture, code in the post.

[arios @ariosleveling]

cursor shipped a git forge yesterday. 23m views on the announcement.

the post says now live. the page says early beta, paid plans only.

worth knowing before anyone moves a public repo across: your github repo page is an indexed url. readmes pick up long tail queries, and for a lot of small projects that is the only search traffic they get.

that url does not travel with you. github has no way to 301 a repo to another host, so the safe version is leaving the old repo in place pointing at the new home.

whether origin's own pages are crawlable at all, i don't know yet. nothing public to test on.

[arios @ariosleveling]

@PassiveSphere where's the 0.0001% from, your own funnels or a study? short form converted badly for me too but nowhere near that low.

[arios @ariosleveling]

@marclou how many of the individual startup pages are actually indexed? directories that size usually stall at a fraction of them, and those profile pages are the whole long tail.

[arios @ariosleveling]

@iannuttall can you actually see the ai search half in analytics, or is it just ads revenue moving? most of mine turns up with no referrer so i can't separate it from direct.

[arios @ariosleveling]

@fba the part i'd want to know is whether the linking pages get indexed themselves. plenty of DR 80 links i've checked sit on pages google never picked up. do you check that after they land?

[arios @ariosleveling]

@JespernissenSEO the slug on chris's own post is the tell, it's just the hashtags run together. do you know if it regenerates when you edit the post, or is it locked in at publish?

[arios @ariosleveling]

@chrissimpson the quoted one mostly turns up unlinked mentions for me. people copy the phrasing and never link. do you chase those or leave them alone?

[arios @ariosleveling]

a link to your own site can render as a full interactive page inside someone's x timeline.

that is not x doing it. it is one meta tag on your page.

twitter:card=player plus twitter:player is the pair. old embed plugins leave them behind for years, still pointing at a viewer url nobody remembers shipping.

check yours in 60 seconds:

curl -s your-domain | grep -i "twitter:player"

if that returns anything, the page it points at is embeddable by whoever links you.

then check frame-ancestors. x sandboxes its own iframe. nobody else is obliged to.

found one live this morning.

[arios @ariosleveling]

your rank is one sample. i ran the same mobile query five times back to back: 6, 6, 6, 8 and 9 organic results, top three shuffling. that isn't a measurement.

[arios @ariosleveling]

@iannuttall @bot running a few in worktrees too. annoying part is when two of them edit the same file and the merge ends up being the actual work. does conductor handle that or you still do it by hand?

[arios @ariosleveling]

@chrissimpson map embed is the interesting bit here. do the 300 copies actually keep the iframe or do most strip it to plain text?

[arios @ariosleveling]

needed a banner. generated it with two image models to compare.

one cost 1.25 credits per image. the other 0.15. for what i needed - flat colour, text, no faces - i couldn't tell them apart.

try the cheap one first. you can always escalate.