Frequently asked questions

Just two things: the Chrome extension (free from the Chrome Web Store) and an Anthropic API key for AI judging. You can sign up for an API key at console.anthropic.com — add $5–$10 of credits to start, which covers 30–100+ trials depending on sketch complexity.

No account with SoloARV is required. Your API key is stored only on your device and is never sent anywhere except directly to Anthropic when you submit a drawing for judging.

After installing SoloARV, right-click the toolbar icon and choose Options (or click the ⚙ gear icon inside the extension). Under Anthropic API key, paste your key (it starts with sk-ant-api03-…) and click Test connection. You should see a green "Connection OK" message within a few seconds.

To get a key: sign up at console.anthropic.com, add billing credits, and create an API key. The key is only used for judging — SoloARV never contacts any other server.

While you're in Settings adding your API key, scroll down to Tasking voice. We recommend an English female voice — across macOS, Windows, and ChromeOS these are consistently the least robotic and the most natural-sounding for tasking, which matters more than it might seem. The prompt is asking you to settle and turn inward, and a flat or mechanical voice works against that.

SoloARV already defaults to one, picking the best English female voice your system offers — Samantha on macOS, Zira on Windows, and so on. If you'd rather choose yourself, look for voices labeled premium, enhanced, or natural: they're noticeably smoother than the older built-in voices. Use Preview to hear any voice before saving, and if you've changed it and want to go back, Use recommended restores the default.

The default speech rate is 0.85×, slightly slower than normal speech. That's deliberate — an unhurried prompt is easier to settle into. Adjust it if you prefer, but slow generally works better than fast.

Each judging request sends two target photographs, your sketch, and your notes to Claude. The typical cost is $0.05–$0.15 per trial, depending on image sizes and note length. A $10 credit top-up comfortably covers 70–200 trials.

You manage your credits directly at console.anthropic.com. SoloARV has no visibility into your billing and never charges you for API usage.

The best questions are ones that resolve clearly to one of exactly two outcomes with no ambiguity. Examples: "Will BTC close above $X at 4 PM?" / "Will Team A win or lose Saturday's match?" / "Will I be offered the job or not?"

Avoid questions with more than two outcomes, vague resolutions ("Will things go well?"), or questions where you have a very strong emotional stake in the answer — emotional attachment can contaminate the session.

Not in the current version. SoloARV uses its built-in curated pool of 800 photographs because they've been selected for visual diversity — the key requirement for good judging. Custom photo support may be added in a future update.

Either works. The canvas is convenient for quick sessions. If you prefer to sketch on paper first (many viewers do), photograph or scan it and upload it using the Upload a sketch button.

If you upload a hand-drawn sketch, use a dark pen, marker, or Sharpie on white paper. Pencil and light lines often don't have enough contrast for Claude to read clearly, which can hurt judging accuracy.

There's no timer. Most viewers spend 5–15 minutes on the tasking session. Go with your first impression — extended analysis tends to introduce conscious reasoning that can override the intuitive signal. Submit when you feel you have a meaningful sketch and some notes, not when the drawing looks "good."

Sensory and emotional impressions work best: colors, shapes, textures, scale (large/small), natural or man-made, indoor or outdoor, feelings evoked, movement or stillness. Example: "rounded shape, blue-green color, reflective surface, feels expansive, outdoors, peaceful."

Don't try to interpret what the target is — describe the raw sensations. Analytical overlay ("I think this is a lake") is less useful to the judge than raw descriptors ("smooth, blue, horizontal, calm").

ARV works best when the session is unhurried and your mind is quiet before you start. A few habits experienced viewers rely on:

• Settle in first. Take a few slow breaths before you begin tasking — a rushed or distracted start tends to produce a rushed, distracted sketch.
• Avoid substances that affect your central nervous system beforehand. Caffeine, alcohol, and other stimulants or depressants can dull or distort subtle impressions — a clear, unaltered state gives the most reliable signal.
• Work somewhere quiet. A private room with minimal noise and no one looking over your shoulder makes it easier to notice subtle impressions.
• Go with your first impression. The earliest shapes, colors, or feelings that arise are usually the most reliable signal. Extended deliberation tends to introduce conscious guessing rather than sharpen accuracy.
• Avoid analytical overlay (AOL). Resist naming the target ("this is a building") and instead capture raw sensory data: shape, texture, color, scale, temperature, motion, mood. Descriptors judge better than guesses.
• Don't research or speculate about the event. Thinking hard about which outcome is "more likely" before or during tasking can bias your sketch toward what you expect rather than what you perceive.
• Keep sketches simple. A few clear shapes with labeled colors and textures are easier for both you and the AI judge to work with than an elaborate, literal drawing.
• Don't judge a session by how "good" the drawing looks. A messy sketch with accurate descriptive notes often scores better than a polished drawing with none.
• Limit yourself to one viewing per day. Running more than one session daily can blur impressions between trials and lead to fatigue. SoloARV's 800-image pool is sized for roughly a year of daily trials before any image repeats.
• Never overlap trials. Always finish a trial — reach the Feedback step and see the winning image — before starting a new one. If a previous target is still unresolved in your mind, your impressions on the next trial can blend with it, making both sessions less reliable.
• Practice consistently. Like any skill, calibration improves with repetition. Review your trial history over time rather than reacting to any single hit or miss.
• Treat every result as a probability, not a certainty. Even a strong correspondence score is a signal to weigh, not a guarantee — use it as one input alongside your own judgment.

SoloARV sends Claude three images — your sketch, Target A, and Target B — plus your written notes. Claude is asked to evaluate which target image better matches the sketch and notes, and to score the match strength from 1 (weak) to 3 (strong).

Claude receives no information about which target corresponds to which outcome. It's a purely visual and descriptive matching task. The AI's reasoning is hidden from the Results screen (so you can't use it to second-guess your wager) and is only revealed on the Feedback screen, after you've recorded the actual outcome.

The Results screen shows the predicted outcome, using the full label you gave it (such as "Up" or "Red Sox"), followed by the AI's match score from 1 to 3: 1 = low, 2 = medium, 3 = high correspondence. So Up – 3 means "Predicted Up with high correspondence" and Down – 1 means "Predicted Down with low correspondence." If neither photo matches clearly, you'll see No clear direction and no prediction is made.

A score of 1 is generally a signal to skip or reduce any wager. A score of 3 means the sketch matched one photo clearly.

Versions before 1.2 showed a one-letter code instead, such as U-3 for Up with a score of 3.

Viewer Strengths, new in version 1.2, learns which kinds of detail you perceive most reliably. After each winning session (a HIT), SoloARV records which of 22 kinds of detail you described, such as colors, shapes, textures, and motion, and whether each one matched the real target.

A kind of detail becomes one of your strengths once it has appeared in at least 10 of your winning sessions, matched at least 75% of the time, and beats your own usual match rate by more than luck would explain. When a new session leans on one of your strengths and it matches the winning image, the Results screen shows a short note saying so.

The note is advice only. It never changes the prediction or the score. Your track record is built from the trial history stored in your browser, and only sessions judged with version 1.2 or later count toward it.

After the real-world event has resolved and you know the actual outcome. SoloARV keeps the trial waiting for you — just re-open it from the Chrome toolbar. On the Feedback screen, select the actual outcome to reveal the winning image and score your trial as HIT or MISS.

Yes — there's a Cancel trial button on every active screen. If you cancel before submitting your sketch for judging, the trial doesn't count toward your free limit. If you cancel after judging has started, the trial counts as used (since the API call was made).

Displacement is when a viewer perceives a real target accurately — just not the one they were tasked with. It's a well-recognized effect in remote viewing, and commonly reported to be more frequent among newer viewers. How often it happens overall isn't established: the prevalence is currently the subject of formal survey research, and estimates vary widely between viewers and protocols.

The bar is higher than a near-miss. Displacement means an unmistakable, detailed match to the wrong target — specific shapes, colors, or unique elements. A vague sketch that loosely resembles something isn't displacement, it's just a weak session.

In an ARV trial it usually shows up in one of three ways: you describe the other photo in the pair rather than the one linked to the actual outcome; you describe a target from a previous trial whose feedback loop was never properly closed; or you describe the target from the trial you're about to run next.

The working explanation among practitioners is that perception isn't constrained by the labels we put on a trial. What anchors a session is the moment of feedback — the instant you actually see the photo. If several feedback moments are open at once, or if one was never resolved, the signal can attach to the wrong one.

Two properties of the images themselves are also thought to pull attention. Entropy is visual complexity — a busy street scene carries far more detail to latch onto than a plain wall. Numinosity is emotional charge — an erupting volcano or a disaster carries more than a quiet landscape. Where one image in a pair is dramatically more complex or more charged than the other, it can out-signal the one you were actually tasked with. SoloARV's pool is curated for visual diversity, which helps, but no pairing is ever perfectly balanced.

What reduces it: finish every trial through the Feedback step before starting another, so no session is left dangling. Keep one trial per day. Use a consistent tasking routine so each session feels distinct. And resist revealing the alternate image out of curiosity — seeing a photo that was never your target creates exactly the kind of loose feedback moment a later session can lock onto.

How to read a miss: a MISS doesn't automatically mean you got no signal. If your sketch is a strong, detailed match for the losing image, that's a very different result from a vague sketch matching nothing — it's a sign the signal was there and landed on the wrong target. Reviewing your trial history for that pattern is far more informative than tracking hit rate alone, and it's one of the more useful things you can do to develop as a viewer.

Your sketch is sent to Anthropic's API — and only Anthropic — when you press Submit my drawing for judging. It goes nowhere else. The SoloARV authors never receive your sketches, notes, or any other data. There is no SoloARV server.

See the full Privacy Policy for details.

Everything — trial history, API key, settings, the encrypted image mapping — is stored in chrome.storage.local on your device. It never leaves your browser except for the judging request. You can export individual trial records as JSON from the Feedback screen, and wipe everything from Settings → Clear all stored trial data.

Every install comes with 10 complete free trials. After your 10th trial is completed, the extension prompts you to subscribe to Pro to continue. Your saved trial history and statistics remain accessible even after the free limit is reached.

Pro is a subscription ($9.99/month or $99/year, renewing automatically) that unlocks unlimited trials. Everything else — the photo pool, AI judging, voice prompts, trial history, JSON export, and all future updates — is identical between Free and Pro.

Open SoloARV, click the ⚙ gear icon, scroll to SoloARV Pro, and choose Pro monthly or Pro yearly. Checkout is handled securely by ExtensionPay and Stripe. When you return to SoloARV, Pro is active immediately — there's no license key to copy or paste. Your subscription renews automatically on whichever billing cycle you chose (monthly or annual) — you can cancel anytime from Settings → Manage subscription.

Yes. On each Chrome profile or computer, open Settings → SoloARV Pro, click Already subscribed? Log in, and enter the email you paid with — you'll get a login link by email. There's no device limit on a single subscription.