Starting With the Search Term “A Tour of Go 中文版”: I Decided to Maintain a Go Tour Chinese Version

Recently, I have been thinking about a question:

Where else can the website find new traffic growth next?

In the past, I was actually quite optimistic about the English site.

My plan was to gradually translate some existing historical Chinese articles into English. By expanding the volume of English content, the website could gain more overseas search traffic. Initially, I even hoped this would bring about a 30% increase in traffic.

But after running it for a while, the growth I currently see is only about 10%.

Ad revenue has not shown a particularly noticeable improvement either.

However, it is obviously still too early to conclude that the English site direction lacks value.

On one hand, the English site has not been online for very long, and it takes time for search engines to discover, index, and re-establish rankings.

On the other hand, the translation quality of the historical articles on the English site is currently not entirely consistent.

Early on, I successively tried different translation solutions such as AutoPoly, Chrome’s built-in AI, and Yandex. Later, after multiple rounds of actual testing, I finally settled on using GLM-5.2 as the primary full-text translation model.

Therefore, a significant portion of the historical English articles were still not translated by GLM-5.2.

I am currently gradually re-translating these historical articles using GLM-5.2.

Thus, the current traffic and ad revenue performance of the English site is itself still in a continuous adjustment phase.

However, this situation also led me to start considering another direction:

If I merely keep translating existing Chinese content into English, achieving a significant short-term increase in website traffic might not be easy.

In contrast, if I can find a direction that already has clear search demand but where the existing content supply happens to be problematic, it might be more worth investing time in.

Recently, I happened to discover exactly such an opportunity.

1. Starting from a real search keyword

While recently reviewing the website’s Google Search natural query data, I found a very interesting keyword:

A Tour of Go 中文版

It currently only generated 1 impression and 1 click, so the data volume is very small.

But for me, what really matters is not this single click, but rather that this keyword made me start paying attention to:

Are people still searching for a Chinese version of the Go Tour?

I usually continue to review and strengthen my Go-related knowledge anyway, so after seeing this search term, I casually searched for the current Chinese version of the Go Tour.

I discovered something very interesting.

2. The official Go site still retains the Simplified Chinese entry

The official Go site currently still provides A Tour of Go:

https://go.dev/tour

On the Tour’s language selection page, you can still see:

Simplified Chinese — 中文（简体）

This means that, at least from the official page itself, the Simplified Chinese version has not disappeared from the language list. (Go)

This naturally led me to click on this link.

It points to the former Chinese Go Tour:

tour.go-zh.org

But it is no longer accessible now.

3. The official Chinese entry remains, but the target site is defunct

When actually visiting tour.go-zh.org, it currently fails to open properly.

When I access it, Firefox ultimately displays:

Secure Connection Failed

The error code is:

PR_END_OF_FILE_ERROR

And judging from the server request results, this domain currently cannot serve the Tour page properly either.

This creates a rather interesting situation:

The official Go site still displays the Simplified Chinese entry, but the corresponding former Chinese site is no longer usable.

At the same time, my website actually saw the search query:

A Tour of Go 中文版

Although a single data point is far from proving the existence of a large search market, it at least shows:

This demand has not completely disappeared.

Moreover, this is fundamentally different from simply translating existing blog content into another language.

The former is more like:

Existing Chinese articles → translated into English → waiting for new English search traffic.

Whereas the direction of a Chinese Go Tour is:

Users already have relevant searches → existing Chinese entry is defunct → attempting to fill this gap.

This is the main reason I decided to investigate this project further.

4. Initially, I just wanted to redeploy the old Chinese version

When I first discovered this issue, my idea was actually very simple:

Since someone already made a Chinese version before, can I just redeploy the old repository?

I did indeed find the relevant historical open-source project.

If the goal is just to make the website accessible again, this is obviously the fastest solution:

Find the old source code, deploy it to a server, configure the domain, and then connect a CDN.

Almost no new development is needed.

But after further consideration, I gradually felt that this was not a good long-term solution.

The problem is not:

Whether I can get this website running again today.

The real problem is:

What happens when the official A Tour of Go updates in the future?

If I directly use a Chinese project that stopped being maintained years ago, then it is very likely that:

The official Tour continues to update, while the Chinese version remains stuck in the past version.

A few years later, the problem we see today will be encountered all over again.

Therefore, I gradually shifted the goal from:

Restoring the old Go Tour Chinese version

to:

Rebuilding a Simplified Chinese Go Tour that can be continuously maintained alongside the official version.

5. I care more about how to maintain it later, not the first translation

Translating the entire Tour into Chinese all at once is actually not the most difficult part.

The real trouble is:

What happens when the official version updates later?

For example, if the official site modifies a chapter.

The Chinese version needs to know:

That this English text has changed;

That the current Chinese translation corresponds to the old content;

That it therefore needs to be retranslated;

And that it needs to be reviewed again after translation.

So I hope what gets built in the future is not a one-time translation process, but a continuous maintenance mechanism.

The basic idea is:

Official Go Tour updates → detect English content changes → identify parts needing retranslation → call GLM-5.2 → automatic verification → manual review → update Chinese version.

Here, GLM-5.2 is responsible for only one core task:

Translation.

While the program is responsible for deciding:

Which content has changed;

Which content should be translated;

Which content must not be modified;

Whether the translation is already outdated;

Whether the code has changed;

Whether the page structure is still intact;

Whether the translation result meets requirements.

This is actually quite close to the approach I have been continuously refining for the WordPress AI translation process recently.

Except this time, the target shifts from:

WordPress Chinese articles → English

to:

Official Go Tour English → Simplified Chinese.

6. Why choose GLM-5.2

This choice was not made after the project started.

Because over the past period, I have already tested multiple solutions for translating historical WordPress articles.

From early AutoPoly, to Chrome’s built-in AI, to Yandex and other solutions, after multiple rounds of actual testing, full-text translation has gradually unified onto GLM-5.2.

At the same time, I have also dealt with many problems that easily occur in AI translation, such as:

Code blocks must not be mistranslated;

HTML and Gutenberg structures must not be broken;

Technical terms need to remain consistent;

Special structures need protection;

Completeness verification must be done after translation;

Failed tasks must be re-executable.

This experience can be directly transferred to the Go Tour Chinese version.

Moreover, compared to complex WordPress Gutenberg content, the content structure of the Go Tour is actually more regular.

Therefore, there is no need to compare translation models from scratch again this time.

The first version will directly use:

GLM-5.2.

7. This project itself can also be implemented in Go

Since the target is the Go Tour, I also want the project itself to use Go as much as possible.

For example:

Official content synchronization;

Version diff detection;

File parsing;

GLM-5.2 API requests;

Translation status management;

Structure verification;

Terminology validation;

CLI;

Web services;

These can all eventually be implemented step by step using Go.

This endeavor has an additional value.

I normally need to continuously review and strengthen my Go skills anyway, and compared to writing standalone demos that exist just to demonstrate syntax, a project that actually needs to run long-term is clearly better suited for systematically reviewing various Go capabilities.

For example, it will naturally involve:

HTTP Client, JSON, file handling, error handling, testing, context, goroutines, channels, concurrency control, Web Server, etc.

These capabilities are not forced into the project just to “practice a certain syntax”.

Rather, they will be needed by the project itself once it develops to a certain stage.

So this project may ultimately serve two purposes simultaneously:

Maintaining the Go Tour Chinese version.

And:

Serving as a long-term practical Go project for continuously reviewing and strengthening Go skills.

8. Gin is tentatively chosen, but no rush to rewrite the Tour

Since I decided to use Go as the primary development language, Gin is naturally an easy choice to think of.

In the future, when building a translation management API, such as viewing translation status, triggering retranslations, or viewing upstream diffs, Gin is well-suited to handle these Web/API functions.

However, I currently do not plan to use Gin to rewrite the entire Go Tour right from the start.

The reason is also very simple:

The official A Tour of Go is itself a Go project.

Before carefully reading the official source code, deciding to rewrite it in Gin directly does not make much sense.

So the current order is:

First, pull the latest official source code;

Run it first;

Then read the code;

Figure out its existing Web Server, routing, templates, course data, and code execution mechanisms;

Finally, decide which layer Gin should fit into.

Gin will likely end up primarily used for:

Translation management;

Internal APIs;

Status queries;

Synchronization management.

While the Tour itself will retain the official implementation as much as possible.

This way, when following upstream updates in the future, unnecessary modifications can also be minimized.

9. The Chinese version domain is tentatively set to zh.go-tour.shuijingwanwq.com

The domain also went through several changes.

Initially, I considered:

tour.shuijingwanwq.com

Later, I also considered:

go-tour.shuijingwanwq.com

And:

go-tour-zh.shuijingwanwq.com

Currently, I lean more towards:

zh.go-tour.shuijingwanwq.com

Here, it can be understood as:

zh represents the language;

go-tour represents the project;

shuijingwanwq.com belongs to the existing website system.

This way, if other languages do exist in the future, it can also be naturally expanded.

For example:

ja.go-tour.shuijingwanwq.com

ko.go-tour.shuijingwanwq.com

Of course, currently this just leaves that possibility open from a domain and architecture perspective.

The first phase will only do:

Simplified Chinese.

There is no need to maintain an English version ourselves, because the official A Tour of Go is the original English version itself.

10. The Chinese site plans to continue using EdgeOne

This project is primarily aimed at Chinese Go users.

Therefore, unlike the English site, zh.go-tour.shuijingwanwq.com plans to use EdgeOne instead of Cloudflare for the first phase.

The expected production chain is:

User → EdgeOne → Nginx → Go service

The reason is mainly the access experience in mainland China.

The search demand this site hopes to capture in the future is itself:

A Tour of Go 中文版

Go Tour 中文

Golang Tour 中文

and other Chinese keywords.

Therefore, the access experience in mainland China should be a primary consideration for the Chinese site from day one.

And the CDN and the application itself just need to remain independent.

The Go service does not need to know whether EdgeOne, Cloudflare, or another CDN is used in front of it.

11. The goal of the Chinese version is not to reinvent the Go Tour

Regarding the content of the Chinese version, I actually hope to exercise restraint for now.

I do not plan to add a large amount of:

Chinese learning tips;

Extra tutorials;

Custom knowledge points;

Large amounts of original expansions.

just to make the Chinese site look “richer”.

At least, these are not the core goals of this project.

I even more hope that:

The Chinese version stays as consistent with the official English version as possible.

However many chapters the official version has, the Chinese version will have the corresponding number.

Whatever the official version modifies, the Chinese version follows suit.

Content not in the official version will, in principle, not be added arbitrarily.

What truly needs focused investment is:

Whether the translation is accurate.

Whether technical terminology is consistent.

Whether the code remains exactly as is.

Whether the content corresponds to the current official version.

Whether updates can be promptly detected and synchronized after the official version updates.

Even if a small amount of Chinese-specific content needs to be added in the future, it should be limited to necessary translation notes, copyright information, version information, and unofficial disclaimers, rather than changing the Tour’s teaching content itself.

I ultimately hope it becomes:

A faithful, accurate, and continuously maintained Simplified Chinese version of A Tour of Go.

12. Translation status cannot just be “has Chinese” and “no Chinese”

If the goal is long-term synchronization with the official version, then another very important problem needs to be solved:

How do we know which English version the existing translation corresponds to?

For example, a certain English paragraph is translated today.

Six months later, the official site modifies this paragraph.

Even if the Chinese file still exists, it can no longer simply be considered:

“Translation completed”.

It should actually become:

Translation outdated.

So later on, it will very likely be necessary to record:

Source content version;

Source content Hash;

Source version corresponding to the translation;

Translation status;

Review status.

The most basic statuses can be similar to:

new

translated

reviewed

outdated

Once the official English changes, the corresponding Chinese translation is automatically marked as outdated.

This way, the project can truly follow upstream long-term, rather than relying on human memory to judge what has changed.

13. Terminology consistency should also be ensured with programmatic help

There are many fixed technical terms in Go tutorials.

For example:

slice

map

interface

method

receiver

goroutine

channel

If the model is left completely free to translate, different pages will easily end up with different expressions.

So, in the first phase of the project, a Simplified Chinese glossary can be established.

On one hand, it can be added to GLM-5.2’s translation rules;

On the other hand, it can also be provided to the Go program for automated checking.

This helps ensure as much as possible that:

The same Go concept maintains a unified expression throughout the Tour.

A principle is still followed here:

Be as faithful to the official content as possible, rather than reinterpreting it.

14. The first phase temporarily does not make ads a launch condition

The starting point of this project is indeed related to traffic growth and ad revenue.

If A Tour of Go 中文版 can continuously gain search traffic in the future, commercialization can naturally be evaluated.

But in the first phase, I do not plan to make:

Integrating ads

a condition for launching the project.

What needs to be verified first is:

Whether the Chinese site can run normally;

Whether the translation process is reliable;

Whether Google indexes it properly;

Whether relevant Chinese search terms continue to get impressions;

Whether users will actually use the Chinese version of the Tour.

Only after these basic issues are verified is it more reasonable to consider ads.

That is to say, the focus of the first phase is still:

First build the product and verify the real demand.

15. This time I plan to develop primarily in VS Code

This project also has one difference from many recent automation projects:

I plan to complete the development primarily in VS Code.

Over the past period, I have heavily used Codex and other AI programming tools.

For tasks like server troubleshooting, WordPress automation, and batch processing tools, this approach clearly improves efficiency.

But for the Go Tour Chinese version, in addition to the final result, I also hope to use the entire development process to systematically review Go again.

So this time, I want to do more of the following myself:

Read the source code;

Check types;

Trace call relationships;

Use Go to Definition;

Use Find References;

Review tests;

Debug;

Analyze Git diffs;

Understand why the project is designed this way.

AI and Codex will still continue to be used.

But their roles will change somewhat.

Compared to letting AI complete an entire module at once, I prefer:

Letting AI assist with analysis, explanation, reviews, and handling specific issues, while keeping the development process itself as readable and understandable as possible.

The project advancement speed might be a bit slower.

But in the long run, this process itself is part of the project’s value.

16. The first phase only does a very small MVP

There is no need to directly translate the entire A Tour of Go right now.

The first phase only needs to verify whether this path works.

Tentative goals include:

1. Fetch the current official A Tour of Go source code;
2. Run it locally;
3. Use VS Code to read the core code structure;
4. Determine the organization between upstream and the Chinese project;
5. Establish my own Go project;
6. Run the GLM-5.2 API using Go;
7. Complete the Chinese translation for a small number of pages;
8. Establish basic automated verification;
9. Deploy zh.go-tour.shuijingwanwq.com;
10. Connect EdgeOne;
11. Wait for search engine indexing;
12. Observe real search performance.

There is no need to translate dozens of pages even in the first phase.

First complete a few pages, and get the entire:

Official source code → translation → verification → publishing

chain running.

Then decide whether to proceed with full-scale translation.

17. The entire process can also form a new blog series

This project has an additional value.

From initially discovering the search keyword, to source code reading, translation system development, deployment, CDN, SEO, and subsequent traffic verification, the entire process itself can be continuously documented.

Subsequent entries can include:

First run of the latest version of A Tour of Go;

Analysis of the official Go Tour source code structure;

How to design the Chinese localization architecture;

Calling the GLM-5.2 API from Go;

Implementing translation status management in Go;

How to protect Go source code from being translated by AI;

How to detect upstream updates;

How to determine if a translation is outdated;

What role Gin ultimately plays in the project;

Production deployment of the Go service;

EdgeOne integration;

Search indexing after the Chinese version goes live;

Whether A Tour of Go 中文版 actually brought in new traffic.

In this way, the development itself will continuously generate new blog content.

The project drives content updates, and the content can continue to bring new search entry points to the website.

18. Conclusion

The starting point of this project was actually very simple.

I just found in the website’s Google Search natural queries that:

A Tour of Go 中文版

generated one impression and one click.

Following this keyword further, I also discovered:

The official Go site still retains the Simplified Chinese entry, but the corresponding former Chinese site is no longer usable.

Thus, the initial thought of:

Should I redeploy the old Chinese version?

gradually evolved into the current plan:

Based on the current official A Tour of Go, rebuild a Simplified Chinese version that can continuously synchronize, uses GLM-5.2 for assisted translation, is automatically verified by a Go program, and undergoes manual review.

It hopes to solve a very clear problem:

Letting A Tour of Go have an accurate, usable Simplified Chinese version again that can continuously follow official updates.

At the same time, I also hope to use this project to systematically review and strengthen my Go skills again, re-connecting HTTP, Web, concurrency, testing, CLI, automation, and more within a genuinely long-term running project.

As for whether it can ultimately achieve the expected search traffic, I do not know yet. After all, what I see right now is just a very small search signal. But unlike simply guessing a direction, this time there is at least a real starting point: someone has already searched for it. Next, I plan to first pull the current latest version of the A Tour of Go source code locally, open the project in VS Code, and see how this official Go project, which has been running for years, is actually implemented.