Kevin Wong – SLOWLY Startup Diaries, post 6
Translating our Software – Crowdsourcing pros and cons?
Three years on, a review of the pros and cons of using Crowdsourcing for interface translation work.
There was no intention to translate SLOWLY into many languages, especially as resources were tight in the early days. It all started because of a introduction email received from a SLOWLY user, and from then on it was conducted by volunteers in Crowdsourcing model. Today, there are more than twenty languages supported.
A large volunteer team needs a working platform
To have so many translators at the same time (more than 150 at the peak) and yet be manageable, the first prerequisite is to have the right tools. Several translation software and platforms were evaluated, starting with Poedit. It's the most popular and a free one, but it was difficult to manage versions or members.
So I turned to studying some online platforms, and the open source ones set up a few to try it, but there was nothing to recommend (and found that the system resource requirements for translation work were also quite a lot). We then short listed both Poeditor and Crowdin; both have basic functions which are very complete, will not be discussed here. (If you are interested, you can register for a free trial for them.)
Finally, I chose Crowdin and we have been using it until now. The main reason is actually the charging model: only charge according to the number of Source Strings in the project. You can add unlimited members, and the number of target languages is not limited, which is the most suitable setting for us.
The advantages we found
Now that we have found the platform to start, three years have passed in the blink of an eye. Let's start by sharing the benefits of operating in this mode:
Passion and understanding of the product by the translation team – This is important, and it is not difficult to find a translator, but it is difficult to find someone who will take the time to understand how the app works and even the concept behind it. Since the self-nominations are all users, this is a great advantage. I made my decision based on that in the first place.
Low cost – As mentioned in previous articles, we basically have zero income in the first two years, and every penny spent must be carefully considered. Dealing with translation in this way can be said to be almost free. In fact, there are some expenses such as platform fees and time costs for management, in the case of SLOWLY now about 2,000 sets of Source Strings, Crowdin alone charges US$600 per year (almost another markup).
Promotion value – Assuming that self-nominations are based on liking the product and want to contribute, being able to participate in it will bring a little sense of accomplishment. I've seen a volunteer introduce himself on Twitter with the words, “I'm a translator for SLOWLY in a certain language,” ?? so at least one more chance to get his friends to notice a little.
As for the disadvantages
...there are also a few points:
It's hard to control quality – that's my biggest headache. There are several situations that are the most common, one is the handover problem, each person has a different tone of voice, if you find a professional translation team may be avoided. Although some important keywords can be created in a thesaurus, and some problems can still be found with the automatic detection function, it is very difficult to ensure that a certain level of professionalism can be achieved overall.
Different countries and cultural backgrounds – variations of a common language, many of which are official in many countries; and there are differences in the use of words in different natural grammars in different places. The solution can be to add regions and subdivide them into documents in different languages, for example, English can be divided into enGB and enUS, traditional Chinese can be divided into zhHantTW and zhHantHK, and so on... But when you don't even do a good job of the original EN and zh_Hant, it is impossible to break it up again, and the cost of each update will rise again.
Experience level – The third is the level problem, people have volunteered to help you, do you question others? At least that's what I thought. But in exchange for the possibility of facing the level problem, sometimes even receiving a letter of complaint, this has happened several times. In the end, we had to tighten the volunteer admission requirements, and only accept volunteers with more than a certain level of experience.
High attrition rate – Since the participants are motivated, and enthusiasm is completely understandable over time (some of them have helped from the beginning to the present, which is really appreciated 🙇🏻♂️). Since everyone is volunteering, there is no reason to ask you to commit to anything, and this problem is also common in some small and medium-sized projects in the open source world.
Some of the more popular languages are actually fine, but if they are unpopular, it is a bit troublesome. After a few updates, the degree of translation completion will decline, and if it falls to a level, you will have to face the elimination of it or the cost of hiring translators to make up for it, but these unpopular languages themselves are less numerous, and even fewer people need to know English at the same time, so it does not mean that you are willing to pay to find someone to help.
The downside is also cost control – as mentioned in the second point, there are many cases where it may be necessary to find additional manpower, for translation or proofreading. However, supporting these languages was not part of the original plan, which would have avoided recurrent costs. In fact, you can also choose to stop support, and we have reluctantly terminated two languages, but we will be unhappy if we are already using them.
Remediating problems
These situations are difficult to predict, and if you need to find a translator, the more unpopular the language, the more expensive and difficult to find support for it. To save costs, you may have to go to some Freelancer websites such as Fiverr, Upworks, etc. to find helpers, but many times are non-specialists, and the quality is difficult to control, and management can become difficult.
All in all, there are many successful examples of Crowdsourcing, such as some of the most commonly used location databases, such as Foursquare or Google Places, which started out as a large number of users, and have now evolved into a standalone business (and not cheap); Another example is the Captcha that is often seen when filling out forms. As for the use of interface translation, it is necessary to consider the points mentioned above.
Moving to a new paradigm
For all these reasons, we also started to drift away from this model. That is to say, for those languages we have cancelled we will not take the initiative to find someone to fill them. And we will not accept new volunteers to join the translation project, since a few months ago.
Next, we will slowly return to the traditional model, and find a translation team to work with; hoping to optimize quality control and management.
Thanks again to this group of selfless volunteers, we finally completed a stage together! Without them, I think there would only be English and Chinese.
Hope my experience can help you, see you next time.
Cover photo source: Unsplash
🔔 The original text was published in startupslowly.com, on April 13, 2022. Sadly, this URL is now gone, redirecting to the slowly.app generic site. A backup of this site is on Archive.org here.
It also appeared in the original Mandarin language on Archive.org here
Translation Notes and rationale:
This translation was prepared by Yann2, a user, fan and supporter of the SLOWLY project. And it is in no way authorized or condoned by the original author. Any translation error contained herein are my own, and subject to revisions – please send comments if necessary, thank you.
This translator's view is that the inner workings of Kevin's mind are of interest to the passionate fans of the app he has created from scratch. Hence, we bring here an English language version which can be more widely read. Thank you, Kevin!
All 7 original posts, now in English
post 1 : Two startups were not successful, what will happen this time
post 2 : How I found my first users
post 3 : 1345 free cups of coffee from around the world
post 4 : Write a personal introduction letter to the start-up
post 5 : An interview on the Clubhouse app
post 6 : Translating our Software – Crowdsourcing pros and cons?
post 7 : The Hard Truths About Business Development
Famous Last Words
Special thank you to all the folks behind the SLOWLY app.
Letters we share, with pen pals all over the world. A modern day take in the traditional penpal experience. Come and join us, using SLOWLY.
Follow Blog via your Fediverse/Mastodon account :
@friends-near-and-afar-letters-we-share@wordsmith.social
This page created in MarkDown language. Thank you for reading, feel free to comment, via a direct message to my Twitter account – or the Reddit one.
You can also post a public comment in the Reddit thread for it here.
yann2@bsky.social