DeArrow

YouTube rewards packaging. A title that hides the point and a thumbnail that exaggerates it win the click, and the viewer pays for it in minutes spent on a video that doesn’t deliver what it promised. DeArrow targets that trade directly. The utility, built by Ajay Ramachandran, replaces clickbait titles and thumbnails with community-submitted versions that describe what a video contains.

Ramachandran also makes SponsorBlock, and DeArrow is a separate crowdsourced project: users submit replacements, users vote on them, and the most accurate ones rise to the top. A replacement title can be any text that describes the video better. A replacement thumbnail has a stricter rule: it must be a screenshot from a chosen moment in the video. That rule matters more than it first appears. It stops the community from making its own flattering artwork, and it ties every replacement image to footage the video contains.

When nobody has submitted anything, DeArrow still acts. It can rewrite the original title in a format the user chooses, and it can use a frame from a random timestamp as the thumbnail. Users can switch off either behaviour separately.

On iOS, DeArrow runs as a Safari extension for iPhone and iPad. The project is open source, with a public API, a downloadable database and support for mirror servers.

Reception is positive but thin. The App Store version holds 4.4 out of 5 from 18 ratings. That is a small sample, and a tool that rewrites the feed will always provoke rejection from some viewers. DeArrow is a niche tool for viewers who already resent manipulative thumbnails, and for that audience it makes a reasonable case.

• Community titles and thumbnails: Users submit replacement titles as free text. Replacement thumbnails must be frames from a chosen moment in the video.
• Voting: Submissions compete on accuracy, and the best-rated version takes the slot.
• Automatic title formatting: Videos with no submissions can have their original title rewritten in a format the user picks.
• Random-frame thumbnails: When no community thumbnail exists, a frame from a random timestamp replaces the creator’s artwork.
• Granular settings: Each fallback can be turned off on its own, for example to keep original thumbnails by default.
• Casual mode: A less strict experience for viewers who want a lighter touch.
• Original/replacement toggle: A switch shows the creator’s original title next to the community version.
• Open infrastructure: Open source code, a public API, a downloadable database and mirror server support.
• Safari extension: Works on YouTube inside Safari on iPhone and iPad.

The Voting Loop

DeArrow’s value comes down to one question: how often does a better title already exist when a video shows up in the feed? The community supplies the answer, and the design handles that community well. Submissions compete, votes sort them, and the most accurate version wins. That mechanism fits the problem. Clickbait comes in degrees, and a voting system handles degrees better than a fixed rulebook. The crowd can decide whether a title needs a full rewrite or a light trim.

The free-text title field is a deliberate trade-off. Allowing “any text that describes the video better” lets contributors name the real subject, the real result or the real product. It also invites editorialising, and the only defence against a smug or wrong rewrite is the vote. We think the trade-off is correct. A rigid template would struggle with the range of videos on YouTube, and the voting system already exists to catch bad entries.

The thumbnail rule is the sharper piece of design. Because replacements must be frames from the video, the image works as evidence. A viewer sees what the footage shows at a chosen moment instead of a composite built to provoke. That is the whole use case. DeArrow tells viewers what they are about to watch, and honest packaging can change the decision to watch.

Filling the Gaps

Every crowdsourced tool has a coverage problem. DeArrow’s value depends on enough contributors having submitted a better title or thumbnail before a video appears in the feed. Where submissions are thin, clickbait-style titles stay unchanged. The fallbacks exist to soften this. Automatic formatting rewrites the original title into a user-chosen style, which can remove the shouting even when nobody has written a better title. Random-frame thumbnails replace designed artwork with a still from the video.

Those fallbacks have limits. Reformatting changes how a title looks, not what it claims, so a misleading premise survives in calmer type. A random frame is honest but arbitrary: it may land on something informative or on nothing much. We see the fallbacks as damage control. They are useful defaults that keep the feed from sliding back into full clickbait, but they can’t replace a human-written submission. Letting users disable each fallback separately is the right call, because for some viewers a random frame is worse than the original thumbnail.

Escape Hatches

Replacing a creator’s packaging is an intervention, and DeArrow gives users ways to back off. Casual mode offers a less strict experience for viewers who find full replacement too heavy-handed. The original/replacement toggle lets a viewer see what the creator wrote and judge the difference themselves. That toggle matters a great deal. It keeps the tool open about what it changed, and it gives users a quick check on a bad community submission.

That makes the toggle’s reliability more serious than a secondary setting. On a phone it is a small tap target competing with YouTube’s own controls, and it is the control the whole trust model depends on. If the switch is hard to hit or easy to confuse with the video description, users lose their check on the crowd. The design needs that control to be obvious and dependable.

A Divided Audience

The App Store rating is positive but thin: 4.4 out of 5 from 18 ratings. That is too small a sample to settle much. A tool that edits what people see on a site they use every day will provoke rejection from some viewers; a passive utility rarely does. The positive majority is encouraging, but the sample size means the score should be read as a signal, not a verdict.

Open by Design

DeArrow’s infrastructure may matter more over time than any single feature. The project is open source, has a public API, offers a downloadable database and supports mirror servers. For a tool whose entire value is community data, that is the right architecture. Contributors can see where their work goes, outside developers can build on it, and no single server controls the data. A closed crowdsourcing service asks users to trust it. DeArrow lets them check.

DeArrow has almost no visual identity of its own, and that suits it. Its output lives inside YouTube’s interface: different words in the title slot and a different image in the thumbnail slot. The goal is a feed that looks like YouTube with the volume turned down.

Frame-based thumbnails have an aesthetic cost. A still taken from a video lacks the composition, contrast and text overlays of designed artwork, so a DeArrow feed should look plainer, and sometimes muddier at thumbnail size. For the target audience we consider that a fair price for honesty, and anyone who disagrees can keep original thumbnails.

The controls are minimal: fallback toggles, casual mode and the title switch. That restraint fits a tool meant to fade into the background. It also means the title toggle has to work reliably, because it is the main check on a bad community submission.

On performance, the design asks little of the device. Swapping titles and thumbnails on a page that already loads them is light work for a Safari extension. The bigger dependency is the community database, since every replacement has to come from somewhere. Mirror server support gives the project a structural answer to a single server going down.

• Format on iOS: DeArrow reaches iPhone and iPad as a Safari extension. That format sets its scope: it works on YouTube inside Safari. Viewers who watch mainly in a dedicated app should keep that in mind before buying.
• Pricing: On iOS DeArrow is a paid purchase, $1.99 on the US App Store. At that cost the risk is small.
• Touch input: On a phone, the original/replacement toggle is a small tap target competing with YouTube’s own controls. That collision matters because the toggle is the control users reach for to check a community submission.
• Contributing on mobile: Choosing a precise moment for a thumbnail and typing a careful replacement title are tasks better suited to a keyboard and pointer. Mobile is more naturally a viewing surface than a contributing one, which leaves phone users more exposed to the coverage gap.
• Shared data: Submissions come from one community pool, so iOS users benefit from the wider voting pool.
• Rating confidence: The App Store score of 4.4 rests on 18 ratings, a thin base.

DeArrow solves a real problem with a sensible design. Voting handles the fact that clickbait comes in degrees, and the frame-only thumbnail rule keeps replacement images tied to the video’s actual footage. The open infrastructure means the community’s work isn’t locked inside a single service, and that matters for any crowdsourced tool expected to last.

Its weaknesses come from the same model. Coverage depends on contributors, and the fallbacks soften the gap without closing it: a reformatted title can still mislead, and a random frame can show nothing useful. On iPhone, the original/replacement toggle is a small tap target competing with YouTube’s controls, and it is the control that keeps the system honest. That deserves attention.

At $1.99 for Safari on iPhone and iPad, the cost of finding out is small. For viewers who already distrust YouTube thumbnails, and especially for SponsorBlock users, we can recommend it with caveats. Viewers who want every video in their feed corrected should expect gaps, and most people should treat DeArrow as a narrow fix rather than a default.