SCP‑211: Creation of the International UX Workstream

Proposal Discussion
1

[Ideation] SCP‑211: Creation of the International UX Workstream, March 2026 to August 2026

Workstream Name: International UX Workstream

Workstream Leader: Giantkin

  • Empowered per SCP‑92 and SCP‑151 (granted budget allocation, administrative authority, and leadership responsibilities).

  • Oversees contributor payments and bounty distributions, and manages the Workstream budget.

  • Requests no personal compensation during this term.

Executive Contributor: Fireb0mb1
International UX and Translations Lead (Full Job Description)

  • Responsible for operational execution of translations

    • Translation coordination, translation tooling setup/maintenance

    • French translations

    • QA reviews, bounty translations budget tracking, and DAO documentation updates concerning translation processes.

  • Contributes to Product/#polish efforts through frequent testing, bug reporting, and UI refinement suggestions. Both to enhance the quality of the English and localized user experiences (UX).

  • Maintains the listings for ShapeShift products that require any amount of development and GitHub Pull requests (e.g. Zapper, Defilama, ethereum.org, etc.).

  • Ensures community engagement on Discord (sometimes referred to as the “greeter” role) to guide users seeking help/support or just joining our community.

Summary

This proposal seeks the DAO’s approval to establish and fund the new International UX Workstream under the leadership of Giantkin for a six-month term with the purpose of continuing professional multilingual support and a variety of important operational activities overseen by Fireb0mb1 (previously within the MKDB Workstream). The Workstream will ensure high-quality translations of ShapeShift’s app and website in order to preserve brand trust and global user engagement. The term will run from March 1, 2026, through August 31, 2026 (retroactive start reflecting ongoing operational continuity).

Abstract

Following the MKBD Workstream’s sunset, ShapeShift faces a gap in maintaining professional localization for its global user base. This new Workstream aims to preserve continuity, improve quality assurance, and maintain the professional standards required to inspire confidence among ShapeShift’s international users.

The translation services and other operational activities described here have already been provided to the DAO by the same team of qualified contributors under multiple Workstreams over the years. During this time they have been refined and for the localization already make use of Machine Translations/AI and automation, but always rely on human translations and reviews to ensure the level of quality required for financial services.

Motivation

Translations are essential for a global DeFi platform like ShapeShift, providing clarity and trust for international users making financial decisions. Over the last year, approximately 20% of users who connected a wallet used a language other than English.

To that effect, this Workstream will:

  • Provide translations for:

    • The app: app.shapeshift.com in French, German, Japanese, Portuguese, Russian, Spanish, Turkish, Ukrainian, and Simplified Chinese.

    • The official Website: shapeshift.com in French (manually reviewed), German, Russian, Spanish, and Simplified Chinese.

  • Preserve the current tasks and role executed by Fireb0mb1 (see Job Description above)

  • Support the automation efforts initiated recently by the Engineering Workstream by providing translations/reviews to create better rules/glossaries for the new AI Agent-based translations system within the means provided by the allocated budget for bounty translations.
    The target for the viability of this new system remains set at 1-2% error rate which is what our translators provide currently. If it cannot be reached without human reviews, they will remain in place to maintain our product’s quality.

  • Continue the frequent tests, bug reporting, UX improvements, and suggestions in both English and translated versions of the app.

From a financial standpoint, quality translations have shown to be a strategically valuable investment for the DAO. Over the last 12 months, users whose app locale was set to a non-English language generated approximately $10.3 million in trading volume, resulting in about $58,000 in protocol revenues (assuming a constant 56 bps fee, which are now getting increased to 60 bps).

Chart: https://i.imgur.com/jGd8wCc.png

At the same time, the total annual cost for translation-related operations, including tools, bounties, and coordination was around $48,000, meaning localization efforts covered and exceeded their cost. This assumes translations drove part/most of the non‑English trading volume, which is likely but very difficult to prove. This also doesn’t include the revenues from private.shapeshift.com where such tracking is not possible, but it suggests the overall impact of localization is likely even higher.

Translation expenses are relatively fixed and independent of trading volume or user count but promote a wider global access to the DAO’s services, therefore, any growth in these areas enabled by these translations directly increases the DAO’s net revenue.

Specification

Translation Bounties

  • 7 active translators are maintaining the following languages: German, Japanese, Portuguese, Russian, Spanish, Turkish, Ukrainian, and Chinese. All native speakers, experienced in crypto, and with knowledge of the ShapeShift products acquired over the years.

  • Current bounty rates:

    • Translations $0.16 per word (equivalent value in FOX).

    • Reviews: $0.07 per word (equivalent value in FOX).

  • Fireb0mb1 maintains a transparent budgeting spreadsheet. Giantkin is in charge of the monthly distribution of the bounty compensations.

Budget Summary

Full Budget Spreadsheet

Recurrent monthly:

  • Fireb0mb1 (Part Time): $3,500 paid in FOX

  • Bounty translators (On-demand): ~$600 paid in FOX.
    This amount varies monthly based on the type (translation/reviews rates) and the volume of work. The latter depends on the frequency/size of the new features and changes in the app. This budget is relying on the historical average (~$500 with some operational margin). Any surplus from this budget item will be returned to the DAO or transferred to the next term if the Workstream were to be renewed.

Tooling/Misc.:

  • Weglot “Pro” Plan: $500 (Translation SaaS platform for the website translations)
    Billed currently at a yearly discounted rate of 416.50€ (normal rate: 790€), which was paid by the Foundation last year. The next billing is scheduled for July 20, 2026 and will be covered by this Workstream. The budget assumes a stable rate with a small margin for potential foreign exchange fluctuations and transaction fees. Any surplus from this budget item will be returned to the DAO or transferred to the next term if the Workstream were to be renewed. This will be paid in USDC to Giantkin because the payment can only happen by credit card.

  • Gitlocalize: $0 (Translation SaaS platform for the app translations, currently free).

Six-Month Total: $25,100 (paid in FOX, except for $500 in USDC for tooling)

Compensation Policy:

  • All salaries and compensations for contributors and bounties are paid in FOX.

  • The DAO contributors may choose to receive locked FOX incentives (per SCP-81), this does not apply to bounties.

Benefits

  • Ensures seamless continuity with proven translators familiar with ShapeShift’s brand and DeFi-specific terminology.

  • Delivers professional translations to maintain trust and credibility for international users of our financial app.

  • Supports global accessibility (~20% non-English users) to widen the potential audience of our products.

  • Generates positive ROI by supporting non-English user segments that already drive significant trade volumes.

  • Retains Fireb0mb1’s proven coordination role for translations, UX testing, bug reporting, and community support.

Drawbacks

  • The Workstream introduces a recurring monthly expense which may reduce treasury flexibility for other initiatives.

  • The ROI on these services depends on growing volume, a major decline in international usage could reduce its direct profitability.

  • The Workstream generally requires coordination with other Workstreams (currently Product and Engineering) to keep content aligned and translatable across languages which occasionally involves small time commitments from those teams.

Goals and KPIs

  • Maintain translation coverage for all supported languages with an overall target of 95% of content translated at all times.

  • Deliver and maintain accurate language glossaries and rules to support Engineering’s AI-assisted translation pipeline, enabling measurable efficiency improvements.

  • Support improving UX through proactive testing, bug reporting, and UI/UX feedback in both English and localized versions.

Vote

Yes - Approve the creation and funding of a 6-month International UX Workstream (March 1 2026 to August 31, 2026).

No - Reject this proposal. Evaluate other alternatives

Yes, with changes - Support this proposal but request modifications. Please specify changes in comments/forum.

4 Likes
2

Ideation vote here: Snapshot

Some context about:

Recently, @Jibles from the Engineering Workstream has been working on automating translations for the app using AI agents and implementing a dedicated skill for this purpose. His announced goal is to reduce the costs for the DAO by potentially removing the need for human translations with “pretty good translations”. Our translators have graciously provided a cursory and unpaid review (since it happened after the MKBD Workstream sunset and after the transition to Product was discontinued) of the output of this new Machine Translation system in its current early-stage state.

For reference, currently, our main app has roughly ~2300 strings (~13,500 words in English). Here’s a summary of the results of this review of the AI suggestions for a 200 strings sample in four languages:

  • French: 79% of the AI suggestions required to be corrected.

  • Chinese: 77% of the AI suggestions required to be corrected.

  • Russian: 55% of the AI suggestions required to be corrected.

  • Turkish: 40% of the AI suggestions required to be corrected.

These tests were conducted as a “first-pass” with very basic rules/glossaries which explains the high rates of errors. According to the developer these rates can be improved significantly by building better rules/glossaries for each language. This is required because each language has its own idiosyncrasies and nuances. To progressively refine the system and improve its reliability we will require a specific type of advanced human review in which each translator will have to describe exactly why each suggestion needs corrections (grammatical rules, tone, bad context, variable placeholders, etc.). Our translators are willing to participate in this process to build these rules/glossaries at the current rate ($0.16/word) and we can use some of the budget allocated for the current process to bootstrap this and see if it’s a worthwhile endeavor.

For reference, the current process involves full translations with the ability to use Machine Translation with a rate of error around ~25% (based on internal review experience, with the model provided for free by Gitlocalize) brought back down to at most 1-2% errors by human reviews. This is done at the rate of $0.16/word and implies an average cost of $500/month for all bounty-based languages (except French which is included in my own compensation).

I’m not sure what is the Engineering target error rate, but I expect that his model would reach a better rate than Gitlocalize’s model of course (it has more context and with a good set of rules it will get more things right) and this would imply shorter reviews. If so we could apply the Reviewer rate for these. Although I still cannot fathom blindly trusting the output at this point, even a 5% error rate (without reviews) would currently lead to over a 100 strings containing errors without any automated way to make sure it’s not a string that will mislead a user during their financial decisions. This is way too high and justifies still relying on human reviewers both to preserve the current quality and to be able to maintain/update the rules/glossaries.

Lastly, I don’t have the financial data on the Engineering cost to create/implement/maintain this new system, nor do I know the cost of running it frequently (subscriptions/AI inference costs), maybe Engineering can elaborate on these. But it might be interesting for them to evaluate if, without removing human reviews completely, this is really worth it compared to the current established process/budget.
Unless there is clear evidence that automated translations can meet or exceed current human-quality standards (1-2% error rate), complete reliance on machine output would be premature and could pose reputational and UX risks. Should DAO resources not support continued human review (i.e. this proposal is refused and we can’t fund translations), my advice would remain the same as in a previous message, an English-only interface would be a safer temporary option than deploying inadequately reviewed translations that might mislead users of financial services.

3 Likes
3

Snapshot Snapshot vote Posted

Please vote

1 Like
4

I didnt vote on ideation, but i seen no dissenting votes, so I assume all was good.

So i voted this time.

(silence doesnt work in Governance) meme here.

1 Like
5

The proposal has passed! Thanks for the unanimous support from voters and to @Giantkin for taking on the leadership of this Workstream :blue_heart:

We’ve got some catching up to do for the recent period without bounty translations, but we should get back on track very soon!