Prediction markets are betting exchanges. Useful ones, but betting exchanges, and we should call them that. Before the die-hard prediction market believers come swinging, hear me out.
I spent years in market data roles. So when I see @Kalshi and @Polymarket prices quoted on the news like election forecasts, something bothers me.
A price is not a forecast. It’s where buyers and sellers agreed to trade.
I know this from the inside. On an exchange, a price reflects who showed up, how much money they had, how confident they felt, and what they’d already read. That’s useful information. But it isn’t a model, it isn’t a sample, and it doesn’t come with a methodology you can audit.
Compare that with serious polling and forecasting:
→ Samples designed to look like the actual electorate
→ Weighting to correct for who’s over- or under-represented
→ Likely-voter models to estimate turnout
→ Published methods and margins of error so others can check the work
When a poll misses, you can find out why. When a market misses, all you get is a price that moved.
And the misses add up. Last month, both platforms gave Flávio Bolsonaro roughly a 4-in-5 chance in Brazil. He lost. In Israel, traders made Eisenkot the clear favorite even though polls showed neither side had a path to a majority. After the count, his odds collapsed.
Even this week’s midterms, being celebrated as a win for prediction markets, tell a more complicated story:
• Markets got the direction right. Democrats took both chambers.
• But a week out, Kalshi traders projected 52 Senate seats and rated Kansas and Texas “Lean Republican.” Democrats flipped both and reached 54.
• In South Carolina, markets gave the Republican about a 70% chance on election afternoon. She lost.
• Nate Silver’s model was actually more confident in a Democratic sweep (82%) than Kalshi was (74%).
In other words, the markets mostly read the same consensus the polls and models were already showing. They didn’t beat them.
To be fair, there are areas where these markets do well. Research shows Kalshi matches Wall Street on Fed rate decisions, which are liquid, short-horizon questions with lots of informed traders. Thin foreign-election markets are a different animal.
So my view is simple. These are betting exchanges, and useful ones. Let’s just call them that. Treating them as forecasting authorities, and letting newsrooms present odds as if they were polls, gives them a credibility their track record hasn’t earned.
Also remember: these platforms earn fees on volume, not accuracy. Their accuracy studies are often their own.
To the data scientists and quants here: how would you actually measure whether these markets are any good? Count how often the favorite wins? Check whether things priced at 70% really happen about 70% of the time? Focus only on the close races? And do they add anything that a good polling average doesn’t already tell us?
#PredictionMarkets #DataScience #Elections #Quants #FinancialMarkets #Markets #Kalshi #Polymarket
Credit to @Polymarket for publishing a Brier score and their method. Genuinely more than most do.
I’m curious about one thing, though. Since the numbers cover every market, 4 hours before resolution, and 78% of them resolve “No,” I wonder how it looks for elections alone, a month out, next to a polling average. Has anyone run that comparison? I’d love to see it.
@RichardHanania Agree they should stay legal, and election night is where they’re genuinely great. But 97% with a quarter of the votes counted and a 500K lead isn’t a prediction, it’s reading the scoreboard really fast. Useful betting exchange? Absolutely. Crystal ball? Not quite.
Full nerd version pinned on my profile if you’re into this stuff 🤓
@Kalshi Nice edge over the poll averages, genuinely. But the tweet says you beat “the models” and I don’t see a single model on this chart. Add them and I’ll be the first to clap.
Full nerd version pinned on my profile for those interested in data, polling and predictions markets.
Congrats, genuinely. You called it. But 98.4% is like a weather app bragging it predicted “hot” in Phoenix in July. Most of these were safe seats. The fun part is the close races, and half of those are still being counted. Let’s check back then.
Full nerd version pinned on my profile if you’re into this stuff.
Apple would love that, but retailers would rather print fax receipts than hand Apple their customer data 😂 The app is where they keep your cart, your habits, and your loyalty. Giving that to Apple is like Sam’s letting Costco run the exit door.
My bet: Wallet holds the membership card, the app keeps the brains.
@SamsClub just shipped a Live Activity for Scan & Go, and I genuinely love it. You pay in the app, and a “Thank you! Skip the registers and head to the exit” banner with your exit QR code stays on your lock screen. No digging through the app while juggling a 48-pack of paper towels and a rotisserie chicken. That’s thoughtful, user-first design, and I want to give credit where it’s due. 👏
But.
Look at image 2. That’s the same banner when my phone is unlocked. iOS politely switches the background to white, and the text… stays white. The message is still there, technically. It’s just in witness protection.
The best part? The QR code adapts perfectly, flipping from white-on-black to black-on-white like a pro. So the QR code read the Human Interface Guidelines and the text did not. One element is fluent in light mode; the other is holding out for dark mode forever.
For the non-iOS folks: this is the classic hardcoded .white versus semantic color problem. Use .primary (or set an explicit background tint so it can’t change under you) and the system handles contrast for you. It’s a one-line fix, and exactly the kind of thing that slips through when everyone tests on a locked phone in a dim office.
So I have to ask, with love: was this vibe coded? 😄 Did someone prompt “make a cool lock screen banner,” get a gorgeous demo screenshot, and ship it?
Either way, great idea, 90% great execution, and I’ll keep using it. Sam’s Club, if your iOS team sees this, the fix is a coffee break away. ☕
#iOS #SwiftUI #UXDesign #Accessibility #LiveActivities
@MMVFirm@SamsClub Give it a year. Once shoppers get used to walking out with a lock screen pass, every store without one starts to feel like a fax machine. The real question is who copies it first, and whether they remember dark mode exists 😄
@MatthewBerman Impressive. I have to ask though, what's the backstory here? 😂 was it Acrobat rage or more of a "let's see what AI can ship" thing? and how long did it actually take you
The "6-7 hours of back and forth" part hits way too close to home. The iteration cost is the real killer right now…it's not even the final render, it's the endless re-rolling and fighting with models that keep making mistakes.
You said you need a framework to make it 10x faster... not to shamelessly plug my own thing, but we built Nircle Spark Studio for exactly this workflow problem. It’s prompt → full series, with the image/video models abstracted away so you’re not burning time/model credits figuring out the pipeline. It won’t magically make Higgsfield free, but it’s designed to keep you focused on story instead of re-rolling tool settings.
Full disclosure: I’m the founder, so I’m biased 🙂
https://t.co/5WTrRawS9P use code: oRqr9WHYIdHDCbyz for 20% off signup.
This is incredibly generous of you to share…..the level of detail here for locking in 2D facial consistency is honestly a work of art. Saving this immediately!
It does make me wonder, though: has anyone tried adapting a hyper-detailed reference sheet like this for 3D model generation (or feeding it into an LLM for 3D pipelines)? Or would it just create absolute noise?
Because I'd imagine a 2D AI can cheat a grid, but a text-to-3D engine wouldn't know what to do with a flat multi-angle sheet as it might try to extrude the grid and give you a Cronenberg blob with three heads in a T-pose instead of a clean wireframe. Plus, asking a 3D engine to "not have a CG feel" feels like asking a fish not to look wet.
I'd love to know if you (or anyone in the replies) has tried translating this kind of 2D reference logic into a 3D workflow, or if it's better to just hand this to a human 3D artist with a cup of coffee and a prayer?
Nothing says “vibe-coded startup” like discovering GTA San Andreas has no laptop model, so CJ types on a full desktop PC_1 (2190) sitting on officedesk1 (2008) like the CRT hum is his Series A soundtrack. The fact that someone read the .ifp to learn the sit animation drops the body 0.403m onto a 0.49m kb_swivelchair1 seat is more due diligence than most VCs do. And freezing him before he lands makes him play the falling pose instead of sitting….which is just accurate founder energy. Model ID 1714 has never worked harder.
Here is the prompt to make your own GTA San Andreas mission on MacOS, just paste it in your clanker, you can change the mission of course to your own story:
You are helping me build a custom story mission for GTA San Andreas (classic PC version) running on macOS through CrossOver. Work step by step, check each step works before moving on, and back up everything before changing files.
GOAL
A playable mission called "Vibe Coded". CJ walks into a marker in San Fierro, drives to a diner, sits at a desk with his co-founder and "vibe codes" a startup called Vibes AI in a minigame. Then he drives to an investor office and pitches. The investors invest $10,000,000 and the mission passes. All dialogue is voiced and subtitled.
PART 1: PREPARE THE GAME
1. Back up the full game folder and the save folder ("Documents/GTA San Andreas User Files") before touching anything.
2. Downgrade the game to v1.0 US. The modding tool CLEO only works on v1.0, and the Steam and Rockstar Launcher versions are newer. Use Jetpack Downgrader (open source, on GitHub), which supports the Rockstar Launcher version. You don't need its Windows GUI. Its package contains the full v1.0 exe (game.jpp is a zip holding it) plus JojoDiff binary patches for 15 data files. Apply each patch with its bundled patcher.exe, run through CrossOver's wine command line. Syntax: patcher.exe original patch output. The files are anim.img, gta_int.img, gta3.img, TrakLkup.dat, nine audio streams, script.img and main.scm. Verify: gta3.img should be 940,064,768 bytes with CRC32 c8ccf06f, and main.scm should be 3,079,599 bytes.
3. Install CLEO 5 (the version bundled with Silent's ASI Loader) and SilentPatch. SilentPatch is required, because without it v1.0 refuses to start on a Mac with "Cannot find 800x600x32 video mode".
4. Make the save compatible. v1.0 uses v1 scripts, so a save from the newer versions won't load. Use https://t.co/R4yTH4wx5g to download a save from the right point in the story (we used one right after "Are You Going To San Fierro?"), converted to v1. Put it in a free slot without overwriting existing saves.
5. Launch gta_sa.exe directly, not through the Rockstar Launcher, which may undo the downgrade. Create a CrossOver shortcut for it inside the bottle. Don't put shortcut files on the Mac desktop.
PART 2: MAC/CROSSOVER GOTCHAS (IMPORTANT)
- The game crashes on launch if a settings file (gta_sa.set) saved by a previous session exists. Delete it before every launch. We made the shortcut run a small .bat that deletes it and then starts the game. The downside is that in-game settings reset each launch.
- The game stops loading whenever its window isn't the active window. After launching, keep the window focused (Cmd+Tab to it) even if it looks black or invisible.
- Force-quitting leaves orphaned Windows processes in the bottle, which silently block future launches. If the game won't start, kill every process in the bottle (or use CrossOver's Force Quit / Restart Bottle), then launch again.
- If the camera spins with mouse and keyboard, set the Wine registry value HKCU\Software\Wine\DirectInput MouseWarpOverride = force. Don't enable Retina mode, because it broke launching.
- Windowed mode through dxwrapper produced an invisible window on the Mac, so we removed it. Film the full-screen game with Cmd+Shift+5 instead.
PART 3: WRITE THE MISSION
1. Write the mission in Sanny Builder 4 syntax, in GTA SA v1.0 SBL mode, as a CLEO script (.cs). Compile it with Sanny Builder's command-line option (sanny.exe --no-splash --mode sa_sbl --compile input.txt output.cs), run through wine. Compile in a different CrossOver bottle from the game's, so compiling never disturbs a running game. Put the .cs file in the game's cleo folder. It loads whenever a save is loaded.
2. Use real data from the game files, never guesses:
- Get interior coordinates and interior IDs from the "enex" sections of the game's .ipl map files. We used the SF diner (DINER2, interior 5) and the Planning Department office (PAPER, interior 3).
- Get object model IDs from the .ide files: office chair kb_swivelchair1 (1714), office desk officedesk1 (2008, has a computer on it, top at 0.8 m), computer PC_1 (2190). There is no laptop model in the game.
- Measure object sizes by reading vertex bounds from the .dff models inside gta3.img and gta_int.img.
- Verify animation names by parsing the .ifp files. We used OFF_Sit_Type_Loop from INT_OFFICE, the prtial_gngtlk talking animations and the business handshake from GANGS, and shout_01 from ON_LOOKERS.
3. Placement rules that took several tries to get right:
- The enex coordinates are not floor height. Find the real floor with get_ground_z_for_3d_coord once collision has loaded (after load_scene and about a 1 second wait).
- set_char_coordinates already adds the ped's standing height (about 1 m). Pass the floor height itself, not floor + 1, or characters float in mid-air.
- A frozen character in mid-air plays a falling pose instead of its animation. Freeze characters only once they're on the floor.
- The sitting animation lowers the body by 0.403 m (read from the .ifp file), which lands the hips on a 0.49 m chair seat.
4. Mission flow:
- Trigger: a $ blip and marker outside the Doherty garage. Start only when not already on a mission, then set $ONMISSION = 1.
- Diner: fade out, teleport inside, place the desk, chairs, computer and co-founder, play the typing animation, set a fixed camera, fade in.
- Minigame: draw a fake laptop screen with draw_rect and display_text_formatted. Tap SPACE to prompt the AI and raise a progress bar. A scrolling agent log appears with jokes (spawning subagents, context compaction, auto-approved rm -rf, deleting the tests). Random red bugs appear, and you have 4 seconds to press Y to accept the AI fix. Each bug has its own funny fix reply. Failing costs 12% progress.
- At 100%: "SHIPPED!", teleport outside, and a blip marks the investor office downtown.
- Office pitch: teleport in, spawn three investors, then three questions with Y or N answers. After each choice, show what CJ said ("CJ: ...") before the investor reacts. Then the $10M offer. Answering N triggers a counter-offer gag where the investors panic and cave.
- End: handshake, fade out, teleport outside, add $10,000,000, play the mission-passed tune, show MISSION PASSED.
- Fail: if the player dies or gets busted, clean up all spawned objects and characters and show MISSION FAILED.
5. Sanny Builder pitfalls: some variable names clash with built-in names (for example "blip" and "fx"), so rename them. There is a limit on local variables, so reuse them. Check the compiler log after every build.
PART 4: VOICE ACTING
1. Write a list of every spoken line, one file per line, with clear filenames (for example vc_02_q1, cj_01_q1_yes, cofounder_01). Spell out numbers for text-to-speech.
2. Generate the voices with text-to-speech, using original or designed voices
3. Process the clips with ffmpeg: trim leading and trailing silence, cut any extra takes, normalize loudness so all characters play at the same volume, and convert to mono mp3.
4. Transcribe the final clips (we used Whisper locally) and adjust the subtitles to match exactly what was said.
5. Play the clips with CLEO 5's audio plugin (https://t.co/5S72Zv8zaS.cleo): load_audio_stream with "cleo:\vibes\filename.mp3", play it, show the subtitle for exactly the clip's length, then remove the stream. If a clip fails to load, show the subtitle without sound instead of crashing.
6. Optional: real character lines can be pulled from the game's own audio. The file audio/SFX/SCRIPT holds 218 banks of raw 16-bit PCM speech (find them via PakFiles.dat and BankLkup.dat), about 6,800 clips. Extract them to WAV and transcribe them to search for specific lines.
PART 5: TESTING
After every change, reload the save and play through the changed part. Ask for screenshots when something looks wrong, especially positions and sitting. Change one thing at a time.
No, you’re not the only one left, you’re just in the Trello Preservation Society. We meet on the third Tuesday of every month. The agenda is a card in the “Doing” list that has not moved since 2019. 😄
Jokes apart though, @trello is still the undisputed champion of “I need a visual board in 30 seconds with zero training.” It’s not dead it’s the vinyl record of project management: slightly nostalgic, deeply satisfying, and still better than it has any right to be.
Stay. The board misses you.
Yeah, $6 for 4 attempts at 15s is brutal — the iteration cost is the real killer, not the final render.
Not to shamelessly plug my own thing, but we built Nircle Spark Studio for exactly this workflow problem: prompt → full series, with the image/video models abstracted away so you’re not burning time/model credits figuring out the pipeline. It won’t magically make Seedance free, but it’s designed to keep you focused on story instead of re-rolling tool settings.
Full disclosure: I’m the founder, so I’m biased 🙂
https://t.co/XAWNv4SPMT — use code: oRqr9WHYIdHDCbyz for 20% off signup.
Your circle. Your light. A device crafted for stories, communication, and commerce, powered by Nircle OS with 49+ native apps, beautifully connected.
https://t.co/oGWO3TGA5j
Powered by the Nircle Operating System
Nircle Spark AI is a sounding board: it lets you model behaviors and test how stories will land before they're ever produced.
That applies just as much to a marketing video ad as it does to a film, because an ad is a story too. You can factor in prediction at that stage: see how the narrative lands, where attention holds, what moves people, before production even begins.
Nircle's Spark does this behind the scenes. It takes your idea, models it against your target audience using our red teams, and picks out what resonates before you commit to production.
Every game is designed for a phone held in one hand first, and is fully playable on a desktop with a keyboard and mouse. Creators can capture screenshots from inside their own game, add a cover, and submit to the Nircle app store when they're happy. Start creating at https://t.co/SgQU9SP6yf
Nircle today opened Spark Game Studio, a new way to make playable 3D games without writing a line of code.
Creators describe the game they want, and the Studio builds it, then stays out of the way while they change it.
Spark Game Studio joins Nircle's existing Spark Studio tools for films and web comics, extending the same idea to games: the creator makes the decisions, and the tooling does the work nobody enjoys.
The Studio ships with a library of thousand characters and creatures modern people, workers and officials, fantasy figures, robots and animals, all free to use and ready to drop into a game. Every one can be previewed in 3D and watched moving before it's chosen.
When the library isn't enough, creators can make a character that has never existed. They describe the person, review the artwork, and the Studio turns that artwork into a 3D character that walks and runs in their game. Creators can also bring a model they made or found elsewhere and use it directly.
Games made in the Studio are not take-it-or-leave-it. A creator can ask for a specific change, slower traffic, brighter pickups, a gentler first level and the Studio edits the game that exists rather than replacing it. Casting, sound, level layout and difficulty are all editable, in plain language, at any point.