本文聚焦於 個人數位資產管理（Personal Digital Assets Management），說明 early-career 專業人士如何在職涯早期建立可持續、可遷移的數位資產結構。

繁體中文版請見後半段

Opening

When people get their first personal computer, there is almost no formal guidance on how to manage digital assets over the long term. Files tend to accumulate organically, backups are often improvised or ignored, and migration between devices is treated as an afterthought—until a failure occurs. Despite its importance, personal digital asset management is rarely taught explicitly, even though the topic is substantial enough to warrant full-length treatments such as The Science of Managing Our Digital Stuff  (Ofer Bergman & Steve Whittaker, MIT Press)  and Personal Digital Asset Management (Priya Prabhakar) .

This post does not attempt to reproduce a complete theory of personal information management. Instead, it focuses on a practical, early-career–oriented subset: how to structure files, backups, and personal media so that data loss, retrieval, and device transitions remain manageable as workloads and responsibilities grow. It is intended for people getting their first computer  (such as high school students) , as well as for professionals who want to improve long-term productivity and strategic control over their digital assets. In this post, “personal computer  (PC) ” refers to desktop or laptop computers.

A Goal-Driven View of Personal Digital Assets

Rather than presenting a collection of best practices, this post treats personal digital asset management as a goal-driven system. Each design choice—backup strategy, directory structure, and media workflow—is evaluated by whether it satisfies concrete recovery, retrieval, and transition guarantees. The idea is simple: if a system cannot meet its guarantees under realistic failure scenarios, it is not a good system, no matter how tidy it looks.

Twelve Practical Goals

A well-managed personal digital asset framework should satisfy the following twelve goals:

1. If a PC suffers irreversible damage or is lost, recovery to the latest save point should take less than 6 hours. This includes retrieve backup and rework.
2. If the home suffers irreversible damage (e.g., flooding, fire, or earthquake) , and local external backups are not recoverable, recovery should take less than one week.
3. If a phone is lost or irreversibly damaged, no more than six months of photos and videos should be lost.
4. Files intended for sharing or collaboration can be located and packaged in under 5 minutes.
5. When outside and needing to retrieve a file from the PC, it should take less than 3 minutes.
6. Transitioning from an old PC to a new PC should require no more than the actual file copy time plus one additional hour.
7. To access a photo taken five years ago: (a) if the exact date is known, retrieval should take less than 3 minutes; (b) if the date is unknown, retrieval should take less than 10 minutes.
8. If a project folder is revisited after several years, its purpose, origin, and version state should be understandable without external explanation.
9. Folder names and paths remain stable and compatible with automation, scripts, and version control systems across platforms.
10. Deciding where to save a new file should take less than 10 seconds, without hesitation or reorganization.
11. At any moment, it should be immediately clear which files are backed up, how often, and to which locations.
12. Managing photos and videos captured on a phone should be a bounded, periodic task  (for example, half a day per season) , rather than an ever-growing source of stress that is continually postponed.

These goals intentionally mix technical constraints, human behavior, and emotional load. Personal digital assets fail not only because of hardware loss, but also because systems quietly become too heavy to maintain.

To achieve these guarantees—and to protect against productivity loss during study, research, and early professional life—we will introduce the 3-2-1 backup rule, a personal computer directory naming hierarchy, and a personal media management strategy.

Section	What problem it primarily solves	Goals addressed
3-2-1 Backup Rule	Fast recovery from device loss, home disaster, or accidental deletion	1, 2, 3, 5, 6, 11
Directory Naming Hierarchy	Knowing where things go and why, even years later	4, 8, 9, 10, 11
Personal Media Management Strategy	Preventing photos/videos from becoming an unmanageable burden. Automation reduce long-term maintenance cost and human error.	3, 7, 11, 12 With automation: 9, 10

How to Read This Post  (and How Not To)

This post stays within the main theme of Personal Digital Assets Management for Early Careers, but it inevitably touches on external concepts that influence decisions. For example:

• Some people do not value long-term memory preservation and choose to keep minimal media.
• Some people care deeply about partners, family members, or even pets, and intentionally create folders for them at higher levels.
• Some people prioritize autonomy and control; others prioritize convenience.

There is no single “correct” value system assumed here.

To reflect this, recommendations in this post implicitly fall into different levels:

• Must have: basic, invariant constraints; breaking them usually leads to failure.
• Strong recommendation: choices that significantly reduce future pain.
• Acceptable tradeoff: non-ideal but reasonable under constraints.
• Personal preference: shaped primarily by values and lifestyle.

Readers are encouraged to judge urgency and applicability based on what matters to them.

Also, some suggestions inevitably think like a senior engineer rather than a high school or early undergraduate student. That is intentional. This post is meant to show the map, not require you to walk every path immediately. There is no need to implement everything at once. Starting early simply gives you more options later.

3-2-1 Backup Rule

The Real Cost of Data Loss Goes Beyond Files

Data loss has a huge impact on both productivity and mental state. When I was doing my undergraduate research in chemistry at National Tsing Hua University, there was a PhD student in our lab whose laptop was sent for repair. Without clearly informing him, the shop reinstalled the operating system, and all of his data was wiped. After professional disk recovery, only PDF files were recovered; all experimental data were permanently lost. He eventually took a leave from his PhD program. That experience made it very clear to me how critical proper data backup is for academic and professional work.

What is even more frightening is to imagine losing all digital assets from before college, and if family archives were not preserved, and only a few photos remained scattered across social media. It would not feel like merely losing some files, but rather a deep sense of unreality, as if an entire chapter of one’s life, including growth, relationships, efforts, and failures, had been erased.

Implementation of 3-2-1 Backup Rule

The 3-2-1 rule is a widely adopted backup principle built around redundancy and geographic separation:

• Three copies of data: the original  (production data)  plus two backups.
• Two different media: for example, a local drive and a cloud storage service.
• One off-site copy: stored in a physically separate location to protect against natural or physical disasters.

In practice, this can be implemented with a combination of a cloud storage service and an external hard drive.

In practice, this can be achieved by a cloud storage service and a hard drive.

1. A small to medium-sized main working directory  (e.g., less than 200 GB)  is synchronized to a cloud service after each work session  (such as morning, afternoon, or evening) . This provides an off-site copy and enables rollback to recent versions if local mistakes occur.
2. A medium to large external backup  (typically under 1 TB for many early-career users)  is updated on a monthly or seasonal basis using an external hard drive.

The total storage required for personal digital assets can vary significantly—from under 1 TB to several terabytes—depending on profession and media-handling habits, which will be discussed later. However, for most students and early-career professionals, this two-tier setup  (cloud backup for active work and periodic external backups for the full dataset)  is sufficient to satisfy the recovery guarantees defined earlier.

Beyond redundancy and disaster recovery, synchronizing the main working directory to a cloud service provides two additional practical benefits.

First, it enables remote access. When outside without a laptop, files stored in the main working directory can still be retrieved directly from a phone or another computer, eliminating the need for ad-hoc file transfers or emergency requests.

Second, it improves search performance. Local file search—especially on systems that scan directories sequentially—can be slow when dealing with deep hierarchies and large file counts. Cloud storage services typically index file names and metadata in advance, making name-based searches substantially faster and more reliable.

As of 2026, for reference, a 200 GB Google Drive plan costs approximately NT$900 per year, and a 1 TB external hard drive typically costs between NT$2,000 and NT$3,000. For repetitive file synchronization tasks, free tools such as Synkron can be used to automate mirroring between local folders and external drives. Together, these options make a robust 3-2-1–compliant backup setup achievable with a relatively modest financial investment.

With this implementation of the 3-2-1 backup rule, the system directly satisfies Goals 1, 2, 3, 5, 6, and 11, and partially supports Goals 9 and 12 defined earlier.

Personal Computer Directory Naming Hierarchy

A well-designed computer folder-naming structure greatly improves the speed and accuracy of content retrieval. Below is an example of a directory organization that works well from high school through university and into the early stages of a professional career.

High-level File management: Drive

At the highest level of file management, the core principle is to strictly separate system infrastructure from personal data.

• C drive: Operating system, installed programs, and system-related files
• D drive: All personal documents, data, and creative work (organized, easy to back up, and portable)

In short: programs live on the C drive; personal data lives entirely on the D drive.

This separation provides several concrete advantages:

• It significantly reduces the risk that system failures or software issues will damage personal data.
• When reinstalling the operating system, only the C drive needs to be touched; data on the D drive remains largely unaffected.
• Backup, restoration, and data migration workflows become cleaner, more systematic, and easier to reason about.

On macOS or Linux systems, there is no literal concept of C and D drives, but the same logic applies using different mount points. Common examples include:

• /Users/your-username/ on macOS (default location for personal data)
• /home/your-username/ on most Linux distributions
• A separately mounted partition such as /data/

The key idea is logical separation, not the specific path name or operating system convention.

Under the data drive (or equivalent mount point), personal files are further organized into a small number of top-level functional categories, each with a distinct operational role.

D:\
├── 01_Primary\
│   ├── 01_ (username) \
│   └── 02_Memory\
├── 02_Storage\
├── 03_Temporary\
├── 05_Dataset\
└── 07_Git_Repo\

At this level, the goal is data classification by usage pattern, not detailed semantics.

• 01_Primary contains frequently accessed, high-priority personal data.
• 02_Storage holds large files that are infrequently accessed.
• 03_Temporary contains disposable or short-lived files.
• 05_Dataset stores large, typically immutable datasets.
• 07_Git_Repo contains version-controlled repositories whose history is managed externally.

Among these folders, 01_Primary/01_ (username)  is the most important, as it serves as the main working directory where active learning, career development, and knowledge accumulation occur.

The remainder of this section focuses exclusively on how to organize inside this main working folder.

Mid-level File management: Main working folder

The main working folder  (01_Primary/01_ (username) )  is where semantic organization matters most. This is where files must remain understandable, reusable, and maintainable across multiple life stages.

Ideally, each major stage of life—such as high school, undergraduate study, graduate study, and early professional work—will have a dedicated folder within this space. At the same time, reusable knowledge, career logistics, and personal administration must remain accessible beyond any single stage.

An example mid-level structure is shown below:

01_ (username) \
├── 00_Personal\
│   ├── 01_ (username) \
│   └── 02_Memory\
├── 01_Organize\
├── 02_Career\
│   ├── 01_CV-SOP\
│   ├── 02_Applications\
│   ├── 03_Internship\
│   └── 04_Interview\
├── 04_Finance \
├── 11_Science\
│   ├── 01_Math\
│   ├── 02_Physics\
│   ├── 03_Chemistry\
│   ├── 04_Biology\
│   └── 05_EAPS\
├── 12_EECS\
│   ├── 01_Python\
│   ├── 02_SQL\
│   ├── 03_Excel VBA\
│   ├── 04_SSH\
│   └── 05_Website\
├── 21_HSNU\
├── 22_NTU\
├── 23_MIT\
├── 31_Music\
└── 32_Maker\

This structure balances long-term reuse and time-bounded archiving, which is formalized in the following subsections.

Four orthogonal categories

We group folders in the main working directory along four orthogonal categories, each serving a distinct purpose:

1. Function |  (e.g., 00_Personal, 02_Career, 04_Finance)
   These folders contain cross-cutting, non-domain-specific materials related to life administration, career logistics, and personal records.
2. Domain |  (e.g., 11_Science, 12_EECS)
   Domain folders store reusable professional knowledge and skills—materials intended to persist across life stages, institutions, and jobs.
3. Life stage |  (e.g., 21_HSNU, 22_NTU, 23_MIT)
   Life-stage folders capture time-bound artifacts associated with a specific period, such as coursework, projects, and institutional records.
4. Hobby |  (e.g., 31_Music, 32_Maker)
   These folders contain long-term, non-career interests that evolve independently of formal education or employment.

Numbering Strategy:

Numbers are used for stable sequencing, not for chronological ordering. Folders are assigned numeric ranges with intentional gaps to allow future expansion:

• 00–09: personal and administrative functions
• 11–19: major professional domains
• 21–29: life stages
• 31–39: hobbies and personal interests

This scheme avoids renaming folders later when new domains, careers, or interests emerge.

Rule for Folder Placement

When a file could plausibly belong to both a domain and a life stage, use the following rule:

• Life-stage folders store time-bound artifacts.
• Domain folders store reusable knowledge.

For example, a chemistry course taken at NTU belongs under the corresponding life-stage folder:

22_NTU\
└── c-2019-F\
    └── Advanced Inorganic Chemistry II

Here, c-2019-F denotes the semester  (Fall 2019) , with a letter prefix used instead of pure numbering to preserve ordering flexibility.

By contrast, chemistry notes, summaries, or reference materials with long-term reuse value should be stored under the domain folder:

11_Science\
└── 03_Chemistry\
    └── 12_Inorganic

This separation ensures that institution-specific artifacts can be archived, while knowledge capital remains accessible.

Maintenance at Life Transitions

This directory structure is designed for periodic compaction, especially during transitions between life stages  (e.g., graduation or job change) .

At each transition, review the outgoing life-stage folder to:

• Delete unnecessary files, especially large or redundant ones.
• Package completed projects with many small files into .zip or .rar archives.
• Move very early life-stage folders down one level  (e.g., archive high school under 01_Organize) .

This keeps total size and file count manageable, improving synchronization speed and reducing long-term maintenance cost.

With this directory naming hierarchy in place, the system directly satisfies Goals 4, 8, 10, and 11 defined earlier.

Personal Media Management Strategy

Why media needs its own strategy

Personal media files—photos and videos—are typically the largest, least structured, and fastest-growing component of personal digital assets. Once the previous sections  (backup rules and directory hierarchy)  are in place, it becomes clear that media dominates:

• long-term storage cost,
• backup complexity,
• and retrieval difficulty.

As a result, media cannot be managed effectively using the same assumptions as documents, code, or datasets. It requires its own strategy.

Cloud services: scope and limitation

Cloud-based photo services such as iCloud or Google Photos offer convenience, fast search, and low initial setup cost. However, they trade long-term autonomy for short-term ease:

• storage cost scales monotonically with time,
• data portability and export are limited,
• long-term access depends on vendor policy and service continuity.

For early-career users planning decades-long retention, these services are best treated as auxiliary access layers, not as the sole or authoritative archive.

The approach described here assumes local ownership with explicit backup, while allowing selective use of cloud platforms when appropriate.

Media as a lifecycle problem

Media management is treated here as a lifecycle problem, not a pure storage problem. The goal is not to keep everything forever, but to:

• reduce volume early,
• guarantee recoverability,
• preserve retrievability over years,
• and retain meaning where it matters.

This lifecycle is organized into four stages.

Cloud-based photo services such as iCloud or Google Photos offer convenience, searchability, and low setup cost. However, they trade long-term autonomy for short-term ease: storage costs scale with time, data portability is limited, and users remain dependent on vendor policies and service continuity. For early-career users planning decades-long retention, these services are best treated as auxiliary access layers, not the sole archive.

Media management is treated as a lifecycle problem rather than a storage problem. The following four steps correspond to reducing volume, ensuring recoverability, maintaining retrievability, and preserving meaning over decades.

Four Major Steps in Personal Media Management

A. Capture — reduce at the source

Volume control is most effective at the moment of capture. Key principles includes:

• Capture what is likely to have future value  (documentation, reference, or meaning) .
• Prefer photos or short videos over long continuous recordings.
• Review and reduce media before uploading to the PC.

Common sources of unnecessary storage include:

• long or redundant videos,
• Live Photos when a still image is sufficient,
• burst photos  (keep only 1–2 best) ,
• failed shots or content with purely temporary importance.

Early reduction prevents exponential growth later.

B. Storage — apply the 3-2-1 rule selectively

Media storage follows the same 3-2-1 backup strategy, with practical constraints. The core practices includes:

• Backup media from phone to PC.
• Backup PC media to an external drive.
• Maintain an off-site copy when feasible.

Additional considerations:

• For very large, low-importance videos  (several GB) , if local storage is limited and loss is acceptable, uploading to platforms such as YouTube  (unlisted or private)  may be used as a non-ideal fallback, not as a primary archive.
• Public platforms  (such as Google Maps reviews or social media)  should never be treated as primary backups. Only content that is explicitly intended to be public should be stored this way, and always with the assumption that it may be removed or lost in the future.
  For example, a video of you climbing a mountain while wearing a hat and sunglasses could reasonably be stored on a public platform: your face is mostly covered, the file is large and costly to keep locally, and the content itself is not critical. If that video is lost one day, the impact is acceptable.

The key distinction is between authoritative storage and incidental redundancy.

C. Review — compress, delete, index

Review is the mechanism that keeps media management bounded over time. The review activities include:

• deleting low-value content,
• compressing older media where quality trade-offs are acceptable,
• deciding which items should be preserved long-term.

To support long-term retrieval  (Goals 7 and 12) , a personal memory index  (PMI)  is recommended. One simple approach is:

• maintain a spreadsheet,
• one sheet per year, with the column like: Date, Note, Link, File cnt, Size  (Mb) , JPG cnt, Video cnt, MISC cen.
• one row per date,
• record what media exists for that date and where it is stored.

This separates personal memory index (PMI) from file system constraints and keeps retrieval time predictable even years later.

D. Print — preserve meaning outside the digital system

Printing is not only a sentimental decoration, but also a format-independent preservation strategy.

• creates offline artifacts immune to file corruption and software obsolescence,
• reduces emotional dependence on digital systems,
• preserves meaning even if digital assets are partially lost.

Meaningful moments can also be turned into physical souvenirs  (e.g., prints, magnets, framed images) , including gifts to people involved, reinforcing social connections while reducing digital burden.

Tools Needed Along the Way (Automation-Oriented)

As personal media collections grow beyond a few thousand files, purely manual management becomes both error-prone and time-consuming. At that scale, even small inefficiencies compound quickly. Lightweight scripting and automation begin to offer substantial leverage—not because the tasks are complex, but because they are repetitive and rule-based.

Relying entirely on existing free or commercial tools often creates new friction. Different tools specialize in different functions, workflows become fragmented across multiple applications, and critical behavior remains opaque and externally controlled. Over time, this leads to tool lock-in, hidden assumptions, and limited adaptability.

For this reason, a surprisingly robust long-term solution is to write small, purpose-built scripts locally. These scripts do not need to be sophisticated or optimized; they simply encode personal rules in a transparent and reusable way. Treated as lightweight utilities rather than “software projects,” they provide full control, eliminate unnecessary dependencies, and double as practical coding practice.

The following sections outline a minimal set of automation capabilities that, once in place, dramatically reduce the long-term cost of managing personal media.

Core automation

• Sort media by date: Read EXIF timestamps from photos and videos, create folders using a yyyy-mm-dd format, and move files accordingly. This step should also handle errors and exceptions, such as missing timestamps, duplicated files, or media exported from messaging apps that lost metadata.
• Measure folder size and file count: Track how many files and how much storage each date folder occupies. Dates with unusually high file counts often correspond to trips, events, or important moments, while unusually large total sizes may indicate long videos that need special storage decisions.
  This also helps answer vague memory questions later, such as “I remember attending a wedding in Fall or Winter 2024—what exact date was it?” Overall, this supports high-level visibility and long-term maintainability.

Highly recommended

• HEIC → JPG conversion: HEIC is efficient but fragile across platforms and tools. Converting to JPG improves long-term compatibility and reduces surprises when moving files between systems.
• Resize photos to a printing-friendly aspect ratio: For example, convert 16:9 images to 4:3 before printing standard 6×4 inch photos. Doing this intentionally avoids unexpected cropping by print services.
• Add timestamps directly onto the image  (optional but powerful) : Ideally place the timestamp in a white margin added during resizing. This makes the photo self-describing even outside digital systems or after metadata loss.
• Batch compression with specified quality targets: Compress older photos or videos to a fixed quality level  (e.g., 80–85%)  to significantly reduce storage and backup cost while preserving acceptable visual quality.

Optional / contextual

• GPS extraction and visualization: Extract GPS EXIF data from photos and open the locations in a browser or map service. This is particularly useful for travel photos or for verifying where a photo was taken. With additional effort, this can be extended into a personal map-based photo review interface, which is powerful but better treated as a separate, larger project.
• Yearly or periodic media statistics: Generate yearly summaries such as total file count, total size, number of photos, number of videos, and notable peaks. These statistics naturally form the backbone of the Personal Memory Index  (PMI)  mentioned earlier, and help transform an unstructured media archive into something observable and manageable.

With recent advances in AI-assisted programming, writing small, task-specific scripts for media organization, metadata extraction, and batch processing is increasingly accessible—even for users without formal programming training.

With these practices in Personal Media Management Strategy, the system directly satisfies Goals 4 and 12, and partially supports Goal 11, as defined earlier.

Closing

Personal digital asset management is difficult for everyone, and I am no exception. Much of this knowledge is learned through trial and error, and formal teaching materials on the topic are surprisingly rare. I have made mistakes along the way, lost data I wish I had preserved, and spent far more time than necessary fixing avoidable problems.

This article is a summary of what I have found to work reasonably well over time. I hope it helps people who are struggling with the same issues, especially those early in their careers who want to build systems that remain usable and meaningful for years down the road.

Back to the beginning of English version.
以下為繁體中文版：

開場

我們第一次擁有自己的個人電腦時，往往缺乏正式的指引來說明該如何在長期內管理數位資產。檔案數量及大小會自然地隨時間累積，備份通常是臨時應付甚至被忽略，而裝置之間的轉移則常常被視為事後才需要處理的問題，直到某次故障真正發生。儘管這個主題重要度極高，個人數位資產管理卻鮮少被明確教授；然而它的內容其實多到足以寫成完整的專書，例如《The Science of Managing Our Digital Stuff》 (Ofer Bergman & Steve Whittaker，MIT Press) 與《Personal Digital Asset Management》 (Priya Prabhakar) 。

本文章並不試圖重現個人資訊管理的完整理論，而是聚焦於一個務實、適合早期職涯的範疇：如何組織檔案、備份系統與個人媒體，使資料遺失、檔案取用與裝置更替在工作量與責任逐漸增加的情況下，仍然能夠被有效掌控。本文的對象包括剛接觸第一台電腦的人 (例如高中生) ，以及希望提升長期生產力與數位資產掌控力的職場人士。在本文中，「個人電腦 (PC) 」指的是桌上型電腦或筆記型電腦。

目標導向的個人數位資產

本文章並非羅列一系列「最佳實務」，而是將個人數位資產管理視為一個目標導向的系統。

每一個設計選擇——無論是備份策略、資料夾結構，或是媒體管理流程——都應該以是否能滿足具體的復原、取用與轉換保證來評估。

核心概念其實很簡單：

如果一個系統在真實的失效情境下無法兌現它的保證，那麼無論它表面看起來多整齊，都不是一個好的系統。

十二項務實指標Twelve Practical Goals

一個良好設計的個人數位資產管理架構，應該能滿足以下十二項目標：

1. 若 PC 發生不可逆損壞或遺失，應能在 6 小時內 復原至最近一次可用的狀態 (包含取回備份與必要的重新設定) 。
2. 若住家遭遇重大災害 (如淹水、火災，或地震) ，且本地端與外接備份皆無法使用，仍應能在一週內復原至最近的可用狀態。
3. 若手機遺失或發生不可逆損壞，最多只應損失 六個月內 拍攝的照片與影片。
4. 需要對外分享或協作的檔案，應能在 5 分鐘內 找到並完成整理與打包。
5. 在外出狀態、未攜帶電腦時，若需要存取電腦中的檔案，應能在 3 分鐘內 完成。
6. 從舊電腦轉移到新電腦的過程，除了實際檔案複製所需時間外，額外設定與整理時間不應超過 1 小時。
7. 若要存取五年前拍攝的照片 (a) 若知道確切日期，應能在 3 分鐘內 找到； (b) 若不知道日期，應能在 10 分鐘內 找到。
8. 多年後重新打開某個專案資料夾時，應能在不需要額外說明的情況下理解其用途、來源與版本狀態。
9. 資料夾名稱與路徑應保持穩定，並能跨平台相容於自動化工具、腳本與版本控制系統。
10. 決定新檔案應該存放的位置，所需時間不應超過 10 秒，且不需猶豫或重新整理架構。
11. 在任何時刻，都能清楚知道哪些檔案有備份、備份頻率為何，以及備份存放在哪裡。
12. 手機拍攝的照片與影片管理，應是一項可定期完成的工作 (例如每季半天) ，而不是一個持續累積、令人逃避的壓力來源。

這些目標刻意混合了技術限制、人類行為，以及情緒負擔。
個人數位資產的失敗，往往不只是因為硬體損壞，而是系統在不知不覺中變得過於沉重、難以維護。

為了達成上述保證，並降低學習、研究與早期職涯階段的生產力損失，本文將依序介紹 3-2-1 備份原則、電腦資料夾命名架構，以及個人媒體  (照片與影片) 管理策略。

章節	主要解決的問題	對應目標
3-2-1備份原則	裝置遺失、居家災害或誤刪資料後的快速復原。	1, 2, 3, 5, 6, 11
電腦資料夾命名架構	多年後仍知道檔案放在哪、為何如此分類。	4, 8, 9, 10, 11
個人媒體管理策略	防止照片與影片成為無法控制的負擔。 自動化來降低 護成本與人為疏失。	3, 7, 11, 12 有自動化：9, 10

3-2-1 備份原則

資料遺失對生產力與人生軌跡的系統性衝擊

資料遺失對生產力及心情的衝擊很大。我在清華化學的大學專題實驗室有個博班學長，他電腦送修後在未被商家告知的情況下重灌，導致資料全部遺失。硬碟救援後只救回 PDF 檔，實驗數據全部找不回來，於是他博士班就休學了，可見資料備份對工作的影響有多大。

更可怕的是，如果大一以前累積的所有數位資產全部消失，硬碟救不回來，家庭也沒有保存，只剩下一些零散地散落在社群媒體上的照片。那不只是「少了一些檔案」，而是會讓人產生一種強烈的不真實感：彷彿自己曾經走過的那段人生，那些成長過程、關係、努力與失敗，被整段抹去了一樣。

3-2-1 備份原則實行細則

3-2-1 原則是一種廣泛採用的備份策略，其核心概念建立在資料冗餘與地理分散之上，具體包含三個要點：

• 三份資料副本：包含原始資料 (也稱為正式使用資料，production data) 以及另外兩份備份。
• 兩種不同的儲存媒介：例如本機硬碟與雲端儲存服務。
• 一份異地備份：儲存在實體位置不同的地方，以防範天災或實體損害造成的全面資料遺失。

實務上可以透過雲端儲存服務搭配外接硬碟來實現。一個適合多數早期職涯使用者的實作方式如下：

1. 主要工作資料夾 (小到中等容量，例如小於 200 GB) ，在每個工作時段結束後 (例如上午、下午或晚上) 同步到雲端服務。這樣可以確保有一份異地備份，同時在本機操作失誤時，也能回復到近期版本。
2. 中到大型的外接備份 (對許多早期職涯使用者而言，通常小於 1 TB) ，以每月或每季的頻率，透過外接硬碟進行更新備份。

個人數位資產所需的總儲存空間，會因職業性質與媒體管理習慣而有很大差異，可能從不到 1 TB 到數 TB 不等，這一點會在後面的媒體管理章節中再討論。不過，對大多數學生與早期職涯工作者來說，這種「雲端備份活躍資料＋定期外接硬碟備份完整資料」的雙層架構，已足以滿足前面所定義的復原需求。

除了資料冗餘與災害復原之外，將主要工作資料夾同步到雲端，還帶來兩個額外的實用好處。

首先是遠端存取能力。當你外出時沒有帶筆電，仍然可以透過手機或其他電腦直接存取主要工作資料夾中的檔案，避免臨時請人幫忙傳檔，或進行各種權宜處理。

再來是搜尋效能的提升。本機檔案搜尋，特別是在資料夾結構很深、檔案數量龐大的情況下，往往需要逐層掃描，速度較慢。相對地，多數雲端儲存服務會事先建立檔名與中繼資料的索引，使得依名稱搜尋檔案時更快、也更可靠。

以 2026 年為例，Google Drive 的 200 GB 方案年費約為新台幣 900 元，而 1 TB 的外接硬碟價格大約落在新台幣 2,000 到 3,000 元之間。對於需要重複進行檔案同步的情境，也可以使用像 Synkron 這類免費工具，自動化本機資料夾與外接硬碟之間的鏡像同步。整體而言，建立一套符合 3-2-1 原則的備份系統，其成本其實不貴。

透過上述方式實作 3-2-1 備份原則後，整個系統可以直接滿足目標 1、2、3、5、6 與 11，並且在一定程度上支援目標 9 與 12。

個人電腦資料夾命名架構

一個良好設計的電腦資料夾命名結構，能大幅提升內容取用的速度與準確度。以下示範一種從高中、大學一路延伸到早期職涯階段都適用的資料夾組織方式。

高階檔案管理：硬碟的資料分區原則

在檔案管理的最高層次，核心概念就是要把「系統基礎設施」跟「個人資料」徹底分開。

• C 槽：放作業系統、程式軟體與相關系統檔案
• D 槽：放所有個人文件、資料、作品(有組織、方便備份、可隨身攜帶)

簡單來說： 程式安裝在 C 槽，個人資料全部放在 D 槽。

這樣的做法有幾個很實際的好處。

• 大幅降低系統出問題時，連帶損壞個人資料的風險。
• 重灌作業系統時，只要重新安裝 C 槽，D 槽的資料幾乎不受影響。
• 備份、還原、遷移資料的流程變得更系統化、更乾淨，

在 macOS 或 Linux 系統上雖然沒有 C、D 槽的概念，但同樣的邏輯依然適用，只是用不同的掛載點來實現，例如：

• /Users/你的使用者名稱/ (macOS 個人資料預設位置)
• /home/你的使用者名稱/ (大多數 Linux 發行版)
• 或是自己額外掛載的 /data/ 等獨立分割區

核心重點在於邏輯上的分離，而不是實際的路徑名稱或作業系統慣例。

在資料磁碟（或等效的掛載點）底下，個人檔案會進一步被整理成少數幾個最上層的功能型資料夾，每一個都對應到不同的使用情境與管理策略。

D:\
├── 01_Primary\
│   ├── 01_ (username) \
│   └── 02_Memory\
├── 02_Storage\
├── 03_Temporary\
├── 05_Dataset\
└── 07_Git_Repo\

在這個層級，重點不是檔案內容的細節分類，而是依照使用模式來做資料分流：

• 01_Primary：經常使用、優先級最高的個人資料
• 02_Storage：體積較大、但存取頻率不高的檔案
• 03_Temporary：暫時性、可隨時刪除的檔案
• 05_Dataset：大型、通常不會再修改的資料集
• 07_Git_Repo：版本控制的程式碼專案，其歷史由 Git 等工具管理

在這些資料夾之中，01_Primary/01_(username) 是整個系統中最重要的一個位置。
它是主要的工作目錄，承載了日常學習、職涯發展，以及長期累積的知識與成果。

接下來的內容，將會專注討論如何在這個主工作資料夾內部進行更細緻、可長期維護的整理方式。

中層檔案管理：主要工作資料夾

主要工作資料夾 (01_Primary/01_〈username〉) 是最重要的地方。這裡的檔案必須在不同人生階段中都能保持容易理解、可重複使用、且容易長期維護。

理想情況下，每一個重要的人生階段——例如高中、大學、研究所，以及早期職涯——都應在這個資料夾中擁有一個專屬的區域。同時，那些可重複使用的知識、職涯相關文件，以及個人行政資料，也必須在各階段之間都能輕鬆取用，而不會被綁死在某個特定年代或身份裡。

以下示範一個中層資料夾結構的範例：

01_ (username) \
├── 00_Personal\
│   ├── 01_ (username) \
│   └── 02_Memory\
├── 01_Organize\
├── 02_Career\
│   ├── 01_CV-SOP\
│   ├── 02_Applications\
│   ├── 03_Internship\
│   └── 04_Interview\
├── 04_Finance \
├── 11_Science\
│   ├── 01_Math\
│   ├── 02_Physics\
│   ├── 03_Chemistry\
│   ├── 04_Biology\
│   └── 05_EAPS\
├── 12_EECS\
│   ├── 01_Python\
│   ├── 02_SQL\
│   ├── 03_Excel VBA\
│   ├── 04_SSH\
│   └── 05_Website\
├── 21_HSNU\
├── 22_NTU\
├── 23_MIT\
├── 31_Music\
└── 32_Maker\

這樣的結構在長期可重複使用與依時間階段封存歸檔之間取得平衡，並會在接下來的子段落中進一步說明。

四個彼此獨立 (orthogonal) 的分類維度

我們在主工作資料夾中，依照四個彼此獨立、各自服務不同目的的分類維度來整理資料夾：

1. 功能 (Function) | (例如：00_Personal、02_Career、04_Finance)
   這一類資料夾放的是跨領域、非專業本體的內容，主要與生活行政、職涯流程，以及個人重要文件相關。
2. 領域 (Domain) | (例如：11_Science、12_EECS)
   領域資料夾用來存放可長期重複使用的專業知識與技能，這些內容應該能跨越不同人生階段、學校或工作單位持續發揮價值。
3. 人生階段 (Life stage) | (例如：21_HSNU、22_NTU、23_MIT)
   人生階段資料夾用來承載特定時間區段內產生的產物，例如課程資料、專題、當下隸屬於某個機構的文件與成果。
4. 興趣 (Hobby) | (例如：31_Music、32_Maker)
   這些資料夾存放與正式學業或工作無直接關係，但會長期演化的興趣與嗜好。

編號策略

這裡的數字用途是穩定排序，而不是時間先後順序。

資料夾會被分配到不同的數字區間，並刻意保留空號，以利未來擴充：

• 00–09：個人與行政相關功能
• 11–19：主要專業領域
• 21–29：人生階段
• 31–39：興趣與個人嗜好

這樣的設計可以避免未來新增領域、轉換職涯或發展新興趣時，必須大量重命名或重排資料夾。

當一個檔案看起來同時屬於「領域」與「人生階段」時，請使用以下判斷原則：

檔案放置的判斷原則

• 人生階段資料夾：放置與特定時期高度綁定、時間性強的內容
• 領域資料夾：放置可長期使用、可跨階段重複利用的知識

舉例來說，在臺大修的一門化學課，課程本身應該放在對應的人生階段資料夾底下：

22_NTU\
└── c-2019-F\
    └── Advanced Inorganic Chemistry II

其中 c-2019-F 代表 2019 年秋季學期，使用字母前綴而非純數字，是為了保留排序上的彈性。

相對地，若是在修課過程中整理出的化學筆記、重點摘要、或未來仍可能教學、複習、寫作時使用的內容，則應該放入對應的專業領域資料夾：

11_Science\
└── 03_Chemistry\
    └── 12_Inorganic

這樣的區分可以讓與特定機構綁定的內容被妥善封存，同時讓真正的知識資本保持可持續存取。

人生階段轉換時的維護策略

這套資料夾結構本來就假設會進行定期壓縮與整理，尤其是在跨越人生階段時（例如畢業、換工作）。

在每次階段轉換時，建議檢視即將封存的人生階段資料夾，並進行以下整理：

• 刪除不再需要的檔案，特別是大型或重複內容。
• 將已完成、但包含大量小檔案的專案打包成 .zip 或 .rar。
• 將非常早期的人生階段資料夾下移一層。(例如把高中資料封存到 01_Organize 底下。)

這樣可以有效控制總容量與檔案數量，提升同步速度，並降低長期維護成本。

在完成這套資料夾命名與結構設計後，整個系統已能直接滿足先前定義的 目標 4、8、10 與 11。

個人媒體管理策略

為什麼媒體檔案需要獨立的一套策略

個人的媒體檔案——照片與影片——通常是所有數位資產中體積最大、結構最鬆散、成長速度最快的一類。

當前面介紹的備份規則與資料夾階層都建立完成後，很快就會發現，真正佔據問題核心的往往不是文件或程式碼，而是媒體檔案本身。媒體檔案主導了幾個關鍵成本：

• 長期儲存成本 (storage cost)
• 備份與同步的複雜度 (backup complexity)
• 多年後的搜尋與回憶成本 (retrieval difficulty)

因此，媒體無法用管理文件、程式碼或資料集的同一套假設來處理，它必須有自己的一套策略。

雲端服務的角色與限制 (Cloud services: scope and limitation)

像 iCloud、Google Photos 這類雲端相簿服務，在便利性上確實非常吸引人：

快速搜尋、幾乎零設定成本、跨裝置即時存取。

但這些便利，實際上是用長期自主權 (autonomy) 換取短期的輕鬆：

• 儲存成本會隨時間單調上升
• 資料可攜性與完整匯出能力有限
• 長期存取權高度依賴廠商政策與服務是否持續營運

對於打算保存數十年媒體的 early-career 使用者而言，這類服務比較適合被視為輔助存取層 (auxiliary access layer)，而不是唯一、也不是主流權威 (authoritative) 的媒體保存位置。

本文採取的前提是：媒體的主控權在本地端 (local ownership)，搭配明確的備份策略；雲端平台則只在合適的情境下被選擇性使用。

把媒體當成一個「生命週期問題」 (Media as a lifecycle problem)

在這裡，媒體管理被視為一個 lifecycle problem，而不是單純的儲存問題。重點不在於「全部留著」，而在於：

• 早期降低數量 (reduce volume early)
• 明確保證可復原性 (guarantee recoverability)
• 在多年後仍可快速取回 (preserve retrievability)
• 在重要的地方保留意義 (retain meaning)

換句話說，媒體管理關心的是「什麼該留下、什麼可以走、什麼值得被記住」，而不只是硬碟空間夠不夠。這個生命週期可以被拆解成四個階段，分別對應於：降低體量、確保安全、維持可找性，以及保存意義。

媒體管理過程中需要的工具 (Automation-oriented)

當個人的照片與影片累積到幾千個檔案以上時，完全靠手動整理很快就會變得又慢又容易出錯。在這個規模下，即使只是一些些零碎動動手指頭的時間，長期累積下來也會消耗大量時間與心力。這時候，輕量化的腳本與自動化工具就能帶來非常大的槓桿效果——不是因為事情變複雜了，而是因為這些工作本質上高度重複、規則明確。

如果完全依賴現成的免費或商業軟體，反而常常會產生新的摩擦成本。不同工具各自只解決一部分問題，流程被切割在多個軟體之間，關鍵行為變成黑盒子，實際控制權也不在自己手上。久而久之，很容易陷入工具綁定、流程僵化、難以調整的狀態。

因此，一個意外地穩健、而且長期來看成本很低的做法，是在本機撰寫一些小而專一的 Python 腳本。這些程式不需要寫得很漂亮、也不需要最佳化，它們只是把「你自己的整理規則」用明確、可重複執行的方式記錄下來。把它們當成工具，而不是專案來看，就能同時達到三個目的：完全掌控流程、避免不必要的依賴，順便累積實用的寫程式經驗，這將會是一個成長的機會。

接下來的段落會列出一組最小但關鍵的自動化能力，只要這些能力到位，就能大幅降低長期管理個人媒體檔案的成本。

A. 核心自動化 (Core automation)

• 依日期排序媒體檔案 (Sort media by date)：讀取照片與影片中的 EXIF 時間戳記，建立 yyyy-mm-dd 格式的資料夾，並將檔案自動移入。這個步驟也需要能處理各種例外狀況，例如：缺少時間資訊的檔案、重複檔案、或是從通訊軟體匯出時遺失 metadata 的媒體。
• 統計資料夾大小與檔案數量 (Measure folder size and file count) ：追蹤每一天資料夾內有多少檔案、總共佔用多少空間。檔案數特別多的日期，通常對應旅行、活動或重要事件；容量特別大的日期，往往意味著長影片，需要額外的儲存策略。這類統計也能幫助未來回答一些模糊的回憶問題，例如：「我記得 2024 年秋冬有參加一場婚禮，到底是哪一天？」從系統層面來看，這能提升整體可視性 (visibility) 與長期可維護性。

B. 高度建議 (Highly recommended)

• HEIC → JPG 轉換：HEIC 格式在空間效率上很好，但在跨平台、跨工具的長期相容性上相對脆弱。轉成 JPG 可以顯著降低未來搬移或換系統時出現意外狀況的機率。
• 調整照片比例以利列印 (Resize for printing) ：例如將 16:9 的照片轉成 4:3，再拿去印標準的 6×4 吋相片。事先自己控制比例，可以避免沖印服務自動裁切造成構圖被破壞。
• 直接在照片上加上時間戳記 (optional but powerful) ：建議在 resize 時額外加一圈白邊，並把拍攝日期印在白邊上。這會讓照片在脫離數位系統、或是 metadata 遺失後，仍然是「自我說明的」。
• 批次壓縮 (Batch compression) ：對較舊的照片或影片設定固定品質 (例如 80–85%) 進行壓縮，可以在視覺品質仍可接受的前提下，大幅降低儲存與備份成本。

C. 進階選用

• GPS 擷取與視覺化 (GPS extraction and visualization) ：從 EXIF 中讀取 GPS 資訊，並在瀏覽器或地圖服務中打開位置。對整理旅行照片、或確認拍攝地點特別有幫助。若再往下延伸，甚至可以做成個人用的地圖式照片回顧介面，但這通常是一個較大的專案。
• 年度或定期媒體統計 (Yearly / periodic statistics) ：例如每年總檔案數、總容量、照片數、影片數，以及特別突出的高峰日期。這些資料實際上會自然形成前面提到的 Personal Memory Index (PMI)，讓原本雜亂的媒體集合，變成可以被觀察與管理的系統。

隨著 AI-assisted programming 的進步，即使沒有正式程式背景，也越來越容易撰寫這類一次性、任務導向的小工具，用來整理媒體、擷取 metadata，或做批次處理。

在完成上述的 Personal Media Management Strategy 後，整個系統已能直接滿足 目標 4 與 12，並在一定程度上支援 目標 11。

結語

個人數位資產管理對每個人來說都不容易，我自己也不例外。這些知識大多是在不斷試錯中累積的，而相關的正式教學資源其實非常稀少。過程中我也犯過不少錯，遺失過希望能留下的資料，也花了很多時間在修補原本可以避免的問題上。

這篇文章整理的是一套「目前看來還算可行」的做法。希望它能幫助正在面對相同困擾的人，特別是還在學習階段或職涯早期、希望建立一套能長期使用、而且在多年後仍然有意義的系統的人。